CANADIAN PARTNERSHIP PARTENARIAT CANADIEN AGAINST CANCER CONTRE LE CANCER Approaches To High-Risk, Resource Intensive Cancer Surgical Care In Canada AUTHORS   LEAD  AUTHORS   CHRISTIAN  FINLEY,  MD  MPH  FRCSC  Thoracic  Surgeon          McMaster  University/St.  Joseph`s  Healthcare  Hamilton   LAURA  SCHNEIDER,  BSc  McMaster  University   SAAD  SHAKEEL,  MPH  McMaster  University EXPERT  BIOSTATISTICIAN   NOORI  AKHTAR-­‐DANESH,  PhD  McMaster  University   EXPERT  SPECIALIST  SURGEONS LAURIE  ELIT,  MD  MSc  FRCSC  Gynecological  Cancer  Surgeon          McMaster  University/Juravinski  Cancer  Centre   ELIJAH  DIXON,  MD  MSc(Epi)  FRCSC  Hepatobiliary  Surgeon      University  of  Calgary/Foothills  Medical  Centre   MCMASTER  HEALTH  FORUM   JOHN  LAVIS,  MD  PhD  McMaster  University   JULIA  ABELSON,  PhD  McMaster  University   FRANCOIS-­‐PIERRE  GAUVIN,  PhD  McMaster  University LEAD  GEOGRAPHERS   NADINE  SCHUURMAN,  PhD  Simon  Fraser  University   BLAKE  BYRON  WALKER,  MSc  Simon  Fraser  University   This  document  has  been  made  possible  through  a  financial  contribution  from  Health  Canada,   through  the  Canadian  Partnership  Against  Cancer.   The  views  or  opinions  presented  in  this  report  are  solely  those  of  the  author  and  do  not   necessarily  represent  those  of  the  Canadian  Partnership  Against  Cancer.   ISBN  978-­‐1-­‐988000-­‐04-­‐6   P a g e    1     TABLE  OF  CONTENTS             ACKNOWELDGEMENTS-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐    4     EXECUTIVE  SUMMARY  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  6   Chapter  1:  Introduction   1.1  Context  Of  Health  Care  Delivery  And  Quality  Improvement  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  133   1.2  The  Burden  Of  Cancer  In  The  Canadian  Health  Care  System  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  155   1.3.1  Esophageal  Cancer  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  166   1.3.2  Pancreatic  Cancer  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  177   1.3.3  Liver  Cancer  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  188   1.3.4  Lung  Cancer  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  199   1.3.5  Ovarian  Cancer  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  200   1.4  Economic  Burden  Of  Cancer  Surgeries  In  Canada  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  211   Chapter  2:  Defining  And  Evaluating  Regionalization   2.1  Defining  Regionalization  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  233   2.2  How  Is  Regionalization  Evaluated?  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  255   2.3  Challenges  Associated  With  Regionalization  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  266   Chapter  3:  Literature  Review:  Existing  Evidence  On  Regionalization   3.1  Literature  Review  And  Methodology   -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  28   3.2  Volume-­‐Outcome  Association  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  29   3.3  Provider  Specialty  And  Outcomes  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  32   3.4  Evidence  For  Regionalization  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  33   3.5  Optimum  Volume  Threshold  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  34   3.6  Limitations  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  35   3.7  Conclusion  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  36   Chapter  4:  Analysis  of  Pan-­‐Canadian  Data   P a g e    2     4.1  Approach  And  Objectives   -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  40   4.2  Analysis  Results  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  41   4.2.1  Esophageal  Cancer  Surgery  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  422   4.2.2  Pancreatic  Cancer  Surgery  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  511   4.2.3  Liver  Cancer  Surgery  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  61   4.2.4  Lung  Cancer  Surgery  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  70   4.2.5  Ovarian  Cancer  Surgery  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  80   Chapter  5:  Regionalization:  Provincial  Case  Studies   5.1  Overview  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  92   5.2  Alberta  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  92   5.3  British  Columbia   -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  93   5.4  Manitoba  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  95   5.5  New  Brunswick  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  97   5.6  Newfoundland  and  Labrador  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  97   5.7  Nova  Scotia  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  99   5.8  Ontario  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  100   5.9  Prince  Edward  Island  and  Territories  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  103   5.10  Saskatchewan  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  104   Chapter  6:  Citizen  Engagement   6.1  Objectives  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  108   6.2  Data  collection  and  analysis   -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  108   6.3  Main  Findings  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  109   Chapter  7:  Summary  and  Key  Recommendations  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  123         References  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  129   Appendix  1:  Summary  of  Literature  Review  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  135   Appendix  2:  ICD-­‐10  diagnosis  and  intervention  codes  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  153   P a g e    3     Appendix  3:  Data  Selection  and  Methodology  for  CIHI  Data  Analysis  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  155   Appendix  4:  In-­‐hospital  Mortality  Analysis  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  157   Appendix  5:  Survival  Analysis  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  162   Appendix  6:  Citizen  Panel  Methodology  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  167   Appendix  7:  Features  of  the  problem  of  delivering  complex  cancer  surgeries  as  described  in  the   citizen  brief  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  169   Appendix  8:  Three  options  for  improving  the  delivery  of  complex  cancer  surgeries  in  Canada  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  170   Appendix  9:  Profile  of  panel  members  -­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐-­‐  171                             P a g e    4     ACKNOWLEDGEMENTS   It  is  only  through  the  contribution  of  many  individuals  and  organizations  that  the   production  of  this  report  was  possible.  We  express  our  gratitude  to  the  following  for  their   contributions  and  assistance  on  this  report:     Strategic  Oversight    provided  by  Canadian  Partnership  Against  Cancer   Mary  Argent-­‐Katwala  PhD   Terrence  Sullivan  PhD   Canadian  Partnership  Against  Cancer   Canadian  Partnership  Against  Cancer   Director,  Diagnosis  and  Clinical  Care   Chair,  Quality  Improvement  and  Systems   Performance  committee     Past  President  and  CEO,  Cancer  Care  Ontario   Geoff  Porter  MD   Professor  of  Surgery   Dalhousie  University,  Halifax,  NS     Canadian  Partnership  Against  Cancer   Senior  Scientific  Leader,  Diagnosis  &  Clinical  Care     Consultations  with  Provincial  Health  Care  Delivery  Experts   We  thank  the  following  individuals  for  consenting  to  discuss  the  structure  of  cancer  surgical   care  and  healthcare  services  as  a  whole  within  their  province.    The  feedback  informed  much  of   the  descriptions  in  Chapter  5  and  provided  an  understanding  of  the  varied  landscape  of   healthcare  provision  throughout  Canada.  Attempts  were  made  to  reach  informants  in   Saskatchewan  and  PEI,  but  were  unable  to  coordinate  formal  discussions     British  Columbia   Christopher  R.  Baliski  MD   Assistant  Clinical  Professor  of  Surgery   University    of  British  Columbia,  Vancouver  BC   Alberta   J.  Gregory  McKinnon  MD     Division  Head,  Surgical  Oncology     University  of  Calgary,  Calgary  AB       Head,  Surgical  Oncology  Network   British  Columbia  Cancer  Agency     Yasmin  Miller     Operational  Leader,  BC  Surgical  Oncology   Network/BC  Cancer  Agency   Manitoba   Helmut  Unruh  MD   Professor  of  Surgery   University  of  Manitoba,  Winnipeg  MB     Co-­‐Lead  ,  VP  &  Chief  Officer  of  Clinical   Services,  Surgical  Oncology   Cancer  Care  Manitoba   Co-­‐Chair,  Alberta  Cancer  Surgery  Council   Ontario   Christian  Finley  MD  MPH   Associate  Professor  of  Surgery   McMaster  University,  Hamilton  ON   P a g e    5     New  Brunswick   S.  Eshwar  Kumar  MD   Co-­‐Chief  Executive  Officer   New  Brunswick  Cancer  Network   New  Brunswick  Department  of  Health     Nova  Scotia   Geoff  Porter  MD   Professor  of  Surgery   Dalhousie  University,  Halifax,  NS     Canadian  Partnership  Against  Cancer   Senior  Scientific  Leader,  Diagnosis  &  Clinical   Care     National   Heather  Logan   Executive  Director,  Canadian  Association  of   Provincial  Cancer  Agencies  (CAPCA)   Director  at  large,  New  Brunswick  Canadian   Cancer  Society  Board  of  Directors   Newfoundland  and  Labrador   Alex  Mathieson  MD   Assistant  Professor  of  Surgery   Memorial  University,  St.  John’s  NL     Eastern  Health  Disease  Site  Member     We  thank  the  following  individuals  for  their  assistance  with  obtaining  data  or  otherwise   assisting  the  authorship  team  with  gaining  access  to  the  vital  information  used  in  this  project.   We  acknowledge  that  the  analyses,  conclusions,  opinions  and  statements  expressed  herein  are   those  of  the  authors  and  not  those  of  CIHI,  ICES,  PopDataBC  or  the  Canadian  Cancer  Society     McMaster  University  Health  Sciences   Canadian  Institute  for  Health   Library   Information  (CIHI)   Laura  Banfield  MLIS  MHSc   John  Ticki   Health  Sciences  Librarian   Senior  Analyst,  Decision  Support  Services   Institute  for  Clinical  Evaluative  Sciences   Population  Data  British  Columbia   (ICES)   (PopDataBC)   Refik  Saskin  MSc   Monique  Gagné   Staff  Scientist,  Data  and  Analytic  Services   Data  Access  Coordinator/Documentation     Nadia  Gunraj  MPH   Lead  Analyst,  Data  &  Analytic  Services   Canadian  Cancer  Society   Data  from  the  Canadian  Cancer  Society  was  accessed  from  the  National  data  reporting,  and   was  utilized  to  assess  population  distributions         P a g e    6     EXECUTIVE  SUMMARY     PROJECT  SCOPE     A   principal   objective   of   health   care   planners   and   administrators   is   to   deliver   care   that   maximizes   patient  outcomes  while  balancing  patient  safety,  satisfaction  and  reasonable  access  to  care.  Cancer   and  associated  care  demands  a  tremendous  burden  on  patients  and  on  limited  health  care  system   resources.  Increasingly,  cancers  are  identified  at  earlier  stages,  positioning  surgery  as  the  optimal   and   often   only   chance   for   cure   and/or   management   of   the   disease.   While   many   patients   fare   well   after   surgery   with   no   complications   and   are   discharged   home   within   expected   timelines,   some   patients  experience  more  complex  post-­‐surgical  courses  with  extended  length  of  stay  and  adverse   events   that   require   additional   care,   re-­‐admission   to   hospital   or   ultimately   death.   Improving   the   outcomes  of  surgical  care  has  the  potential  to  save  lives,  reduce  the  burden  of  disease  on  patients   and  caregivers,  and  reduce  health  care  costs.   There   is   tremendous   variability   in   how   each   province   in   Canada   delivers   cancer   care   services,   resulting   in   disparities   in   patient   outcomes   that   necessitate   attention.   At   least   some   of   these   disparities   can   be   attributed   to   the   low   population   density   in   Canada   and   the   need   to   provide   services   in   geographically   disperse   centers,   where   not   all   of   which   can   provide   highly   specialized   surgical   resources.   There   is   a   special   imperative   in   Canada   as   one   of   the   least   densely   populated,   but   highly   developed   countries   to   articulate   these   issues.   Surgical   cancer   care   is   facing   particular   challenges,   with   rising   incidence   rates   for   most   cancers   and   an   aging   population   increasing   the   demand  for  these  surgeries.  Certain  cancer  surgeries,  such  as  those  for  esophageal,  pancreatic,  liver,   lung   and   ovarian   cancers,   are   particularly   complex   and   are   associated   with   a   high   risk   of   adverse   outcomes,   leading   to   an   increased   burden   on   health   care   resources.   It   is   for   this   reason   that   this   report   focuses   on   deliberate   strategies   that   aim   to   improve   the   organization   of   complex   cancer   surgeries  in  a  way  that  enhances  the  quality  of  health  care  delivery.  One  approach  that  has  gained   considerable   attention   in   the   last   two   decades   is   the   ‘regionalization’   of   surgical   care.   There   is   no   consensus   on   the   precise   definition   of   regionalization   relative   to   the   concept   of   consolidation   of   care   to   high   volume   centres.   Whereas   regionalization   involves   a   deliberate   re-­‐organization   of   services,  consolidation  has  occurred  in  some  locations  where  centres  are  simply  closed  and  patients   receive   care   in   an   alternate   institution.   Based   on   the   literature   and   expert   opinion,   the   following   working  definition  of  regionalization  was  utilized  in  this  report:   “the  deliberate  reorientation  of  cancer  surgical  procedures,  based  on  explicit  and  planned   processes  and  structures,  with  the  intent  of  improving  the  quality  of  care”   This  pan-­‐Canadian  report  provides  an  objective  analysis  and  discussion  about  the  approach  to  high-­‐ risk,   resource   intensive   cancer   surgical   procedures   for   esophageal,   pancreatic,   liver,   lung   and   ovarian   cancers   in   Canada.   We   intend   to   inform   policy   makers,   health   care   planners   and   administrators   on   the   current   state   of   cancer   surgical   care   and   outcomes;   highlight   potential   opportunities  for  improvement;  and  provide  actionable  recommendations  to  improve  the  quality  of   care  for  individuals  and  families  bearing  the  burden  of  cancer.   P a g e    7     METHODOLOGY     This   report   utilized   various   sources   to   inform   a   comprehensive   discussion   on   the   impact   of   regionalization   on   the   quality   of   surgical   cancer   care   delivery   in   Canada.   A   systematic   review   of   the   literature   was   conducted   to   ascertain   the   current   understanding   of   the   relationship   between   regionalization   and   outcomes.   Structured   interviews   with   provincial   clinical   experts   (Acknowledgement   section)   established   the   state   of   care   delivery   processes   implemented   across   Canada.  Patient  outcomes,  access  to  care  and  travel  times  were  evaluated  across  the  country  using   quantitative  analyses  and  validated  Geographic  Information  System  (GIS)  techniques  to  objectively   demonstrate  how  care  consolidation  impacts  the  clinical  outcomes  and  patient  experience.  Finally,   patient  perspectives  and  expectations  of  cancer  care  delivery  were  captured  from  a  series  of  citizen   panels  convened  across  three  different  provinces  with  varying  health  care  delivery  demands.   KEY  FINDINGS     The   literature   provides   evidence   for   the   positive   impact   of   regionalization   on   surgical   cancer  care   • • • The  bulk  of  the  literature  and  pan-­‐Canadian  evidence  supports  the  regionalization  of  cancer   care,  based  upon  objective  measures  such  as  operative  mortality  and  length  of  stay.  This  is   often  expressed  as  a  volume-­‐outcome  relationship,  where  larger  volume  centres  have  better   patient-­‐centred  outcomes.   Institutional   volume   is   regarded   as   the   simplest   and   most   robust   metric   for   describing   volume-­‐outcome   relationships.   Other   institutional   and   surgeon   factors   likely   contribute   to   improved  patient  outcomes,  although  they  are  not  well  discussed  in  the  literature  and  it  is   difficult  to  elucidate  the  effects  of  each  of  the  elements  individually  given  the  multifactorial   nature  of  care.   There  is  no  agreed  upon  volume  threshold  reported  in  the  literature  for  any  of  the  cancer   types   evaluated.   Practically   speaking,   the   volume-­‐outcome   effect   is   demonstrated   even   between   institutions   with   high   case   volumes;   this   suggests   that   it   would   be   imprudent   to   implement  a  single  defined  procedure  volume  threshold.     Higher   volume   centres   predicted   a   significantly   lower   risk   of   in-­‐hospital   mortality   and   shorter  length  of  stay  in  the  Canadian  experience   • • A   reduced   risk   of   in-­‐hospital   mortality   was   significantly   associated   with   an   increase   in   hospital   volume   for   esophageal,   pancreatic,   lung,   and   ovarian   cancer   resections;   whereas   higher   facility   volume   predicted   a   shorter   length   of   stay   for   all   five   types   of   cancer   resections.   The   impact   of   hospital   volume   on   the   risk   of   in-­‐hospital   mortality   was   most   pronounced   for   esophageal   and   pancreatic   cancers.   In   a   statistical   model   controlling   for   the   influence   of   patient-­‐  and  hospital-­‐specific  factors,  every  10  case  increase  in  hospital  volume  predicted  a   risk   reduction   of   in-­‐hospital   mortality   of   21%   and   22%   for   esophageal   and   pancreatic   cancer   surgeries,   respectively.   The   calculated   risk   reduction   was   3%   and   7%   for   lung   and   ovarian  cancer  surgeries,  respectively.   P a g e    8     There   are   significant   disparities   in   patterns   of   practice   and   patient   outcomes   for   surgical   cancer  care  across  Canada   • There   is   a   tremendous   variance   in   in-­‐hospital   mortality,   resection   rate,   and   length   of   stay   outcomes  of  high-­‐risk  cancer  surgical  care  across  Canada.   o In   many   of   the   surgeries   evaluated,   there   was   a   three   to   four   times   difference   in   mortality  rates  between  the  provinces.   This  difference  is  most  broadly  demonstrated  by   the   case   of   esophageal   cancer,   where   patients   in   Newfoundland   and   Labrador   saw   a   2.72%  mortality  rate  while  those  in  Manitoba  have  an  11%  risk  of  death.  These  results   are   both   statistically   and   clinically   significant,   and   are   partly   due   to   hospital   and/or   surgeon   volumes.   This   indicates   a   need   for   provinces   to   thoroughly   investigate   their   current   practices   and   to   consider   implementing   proven   strategies   that   mitigate   or   reduce  the  disparities.   o There   are   substantial   differences   between   provinces   relating   to   resection   rates   for   all   cancers   studied   but   ovarian,   with   the   likelihood   of   receiving   a   potentially   curative   operation  in  a  province  with  high  resection  rates  can  be  double  that  of  provinces  with   lower  rates.  In  the  case  of  liver  cancer,  patients  in  New  Brunswick  have  a  6.9/100,000   resection   rate,   while   patients   residing   in   Alberta   would   expect   a   14/100,000   rate   of   resection.  These  differences  between  rates  can  be  due  to  many  different  factors  that  will   influence   how   many   patients   are   suitable   for   surgery   or   whether   surgical   services   are   available   in   a   given   area.   Provinces   should   examine   the   distribution   of   stage   at   presentation,   patient   co-­‐morbidities,   and   the   travel   distances   to   care   to   determine   if   resection  rates  could  be  optimized  by  screening  programs,  rapid  diagnostic  workup  via   diagnostic   assessment   programs,   referrals   to   high   volume   centres,   or   other   quality   improvement  initiatives  in  order  to  reach  ideal  thresholds  and  improve  patient  survival.   o The   findings   also   report   a   large   variance   in   the   duration   of   hospital   stay   across   the   country,   a   fact   that   speaks   to   differences   in   the   efficiency   of   the   systems   currently   in   place,   or   differing   institutional   policies.   The   length   of   stay   can   vary   by   2-­‐3   days   (~20-­‐ 30%)  on  average  per  patient  depending  on  where  the  procedure  is  performed.  Patients   undergoing   pancreatic   cancer   surgery   demonstrate   the   widest   difference,   with   the   duration   in   hospital   of   Saskatchewan   averaging   14   days   while   patients   in   Ontario   can   expect  to  stay  9  days.  The  reasons  for  this  difference  have  yet  to  be  elucidated,  but  may   indicate   differing   rates   of   complications   or   surgical   complexity   undertaken   depending   on   the   experience   and   training   of   surgical   teams.   It   is   imperative   that   each   province   investigate   causes   of   length   of   stay   variance   in   order   to   improve   the   efficiencies   of   their   respective  systems.       o The   estimated   hypothetical   impact   of   surgical   cancer   care   regionalization   was   considered   by   investigating   the   number   of   lives   that   could   be   potentially   saved   if   all   patients   underwent   surgery   at   the   highest   tertile   volume   centres,   where   they   could   expect  the  outcome  rate  demonstrated  by  the  high  volume  tertile.  An  adjusted  estimate   of   the   potential   number   of   lives   saved   is   presented   in   Figure   1.   When   considering   the   proportion   of   potential   lives   saved   relative   to   the   number   of   deaths,   highest   gains   are   expected   for   pancreatic   cancer   surgeries   (60%),   followed   by   lung   (30%)   and   ovarian   (32%)   cancer   surgeries.   These   potential   gains   provide   the   impetuous   to   consider   P a g e    9     regionalization   approaches   on   a   provincial   level   or   policies   that   discourage   single   surgeons   within   a   centre.   Even   in   the   highest   volume   centres,   quality   improvement   measures  such  as  the  introduction  of  surgical  standards,  staff  training  or  uptake  of  best   practice  operative  techniques  have  the  potential  to  further  reduce  mortality.             Cancer  Type   Esophagus     Pancreas   Liver   Lung   Ovarian   0   100   200   300   400   500   600   700   800   Total  Number  of  Deaths   Potential  Number  of  Lives  Saved   Number  of  Deaths           Figure  1:  Potential  number  of  lives  saved  through  consolidation  in  high-­‐volume  centres     • In   similar   modelling,   if   all   patients   experienced   the   results   of   those   in   hospitals   in   the   highest   volume   tertile,   a   total   of   4775   hospital   days   annually   could   potentially   have   been   saved.   If   all   care   was   provided   in   high   volume   hospitals,   3335   hospital   days   associated   with   lung  cancer  cases,  308  days  for  liver  cancer,  725  days  with  pancreatic  cancer  and  407  days   with   esophageal   cancer   surgeries   would   have   been   saved   annually.   The   number   of   days   could  not  reliably  be  determined  for  ovarian  cancer  surgeries.     There  is  a  lack  of  a  unified  approach  for  the  provision  of  surgical  cancer  care  in  Canada   • • •   Despite  the  significant  role  of  surgery  in  the  cancer  management  pathway,  cancer  surgery   management  is  largely  not  integrated  under  the  umbrella  of  provincial  cancer  agencies.  In   most   provinces,   there   are   no   explicit   roles   to   oversee   the   systematic   provision,   evaluation   and  regulation  of  cancer  surgery  within  a  province.  This  results  in  a  lack  of  accountability  in   most  of  the  country  and  an  inability  to  enact  change  or  to  initiate  a  formal  process  to  track   and  evaluate  outcomes.   A   thorough   screening   of   the   literature   for   relevant   policy   documents   along   with   discussions   with  key  stakeholders  in  each  of  the  provinces  revealed  that  there  is  a  deficit  of  coordinated   efforts  at  either  a  provincial  or  national  level  to  improve  quality  of  complex  cancer  surgeries   across  the  country.   In  most  provinces,  there  is  minimal  regulation  as  to  which  procedures  surgeons  or  hospitals   can  or  should  perform  within  their  specialty  area,  or  how  frequently  they  need  to  perform   these  procedures  to  ensure  their  surgical  skills  remain  up  to  date.   P a g e    10     There  is  a  need  for  national  clinical  guidelines  and  standards  for  surgical  cancer  care   • Ontario   has   the   most   explicit   regionalization   policies   and   guidelines,   with   evidence   of   implementation   for   esophageal,   lung,   pancreatic,   and   liver   cancer   surgeries.   Similar   guidelines   for   ovarian   cancer   have   been   published,   but   implementation   is   still   a   work   in   progress   at   this   time.   Purposeful   steps   have   been   taken,   although   without   published   guidelines,   to   consolidate   cancer   surgery   for   certain   disease   sites   in   British   Columbia   (Thoracic   and   Gynecologic   cancer   surgery)   and   Manitoba.   In   the   other   provinces,   specific   policies   and   guidelines   are   either   absent,   unclear   or   unknown   to   the   authors   despite   discussions  with  local  stakeholders.     Ovarian  cancer:  Is  there  a  case  for  regionalization?   • The   large   number   of   institutions   delivering   surgical   care   for   patients   with   ovarian   cancer   indicates   a   cancer   type   that   may   benefit   from   regionalization   policies   that   would   bring   patients  requiring  ovarian  surgical  cancer  care  to  specialized  higher  volume  centres.  It  has   half   the   number   of   cases   nationally   as   lung   cancer,   but   three   times   the   number   of   institutions   performing   these   surgeries,   with   many   of   the   centres   reporting   a   very   small   total   annual   case   volume.   This   may   improve   patient   outcomes   in   a   manner   similar   to   that   seen  in  other  cancer  types.     Regionalization  has  an  impact  on  travel  time  to  surgical  services   The   median   travel   time   varies   greatly   between   provinces   and   is   largely   influenced   by   geography.   Travel   times   slightly   increased   between   2004   and   2012   almost   irrespective   of   patient  location.   • Across  all  cancer  sites,  the  proportion  of  patients  who  had  to  travel  more  than  one  hour  to   access   care   increased   as   the   number   of   centres   that   provide   surgical   cancer   services   decreased.   This   data   confirms   that   the   regionalization   of   some   services   to   high   volume,   central  locations  occurs  at  the  expense  of  patient  travel  time.     Patients   and   Caregivers   prefer   better   surgical   care   quality   at   the   expense   of   longer   travel   distances   • • • • Three   citizen   panels,   with   38   participants   in   total,   were   conducted   in   Edmonton   AB,   Hamilton  ON  and  Charlottetown  PEI  where  participants  emphasized  a  preference  for  high-­‐ quality  surgical  care  rather  than  convenience.   Participants  emphasized  on  a  strong  need  for  the  provision  of  a  local  support  network  to  be   available   for   patients   and   caregivers   even   in   cases   where   cancer   surgeries   have   been   regionalized.   Participants  felt  that  a  compromise  of  regionalizing  surgical  services  while  facilitating  local   pre-­‐   and   post-­‐operative   clinical   care   or   the   utilization   of   telemedicine   innovations   may   represent   an   optimal   approach   to   regionalizing   these   complex   cancer   services   in   a   way   that   is  acceptable  to  the  Canadian  population.   P a g e    11     KEY  RECOMMENDATIONS     The  significant  variance  in  how  surgical  cancer  care  services  are  delivered  between  the  provinces   greatly  impacts  patient  survival,  efficiency  of  the  healthcare  system  itself  and  how  well  patients  can   access   care.   In   a   prosperous   country   such   as   Canada,   these   disparities   are   unacceptable.   While   there  are  many  theories  for  why  these  disparities  are  present,  this  report  identified  key  issues  that   likely  influence  difference  between  the  provinces.  We  reported  significant  gaps  in  knowledge  that   prevent  informed  decision  making,  a  lack  of  surgical  cancer  care  leadership  within  the  continuum   of  cancer  care,  a  lack  of  a  standardized  quality  monitoring  program,  and  finally  minimal  adoption  of   active   regionalization   policies   that   are   likely   to   enact   change.   These   factors   result   in   an   environment   where   best   practices   cannot   be   disseminated,   and   meaningful   improvements   in   patient  care  are  not  realized  to  their  full  potential.  The  key  recommendations  noted  in  this  report   should   be   considered   for   implementation   at   the   level   of   each   province   and   in   each   cancer   site   to   optimize  patient  care.     Surgical   cancer   care   should   be   integrated   into   the   overall   spectrum   of   provincial   cancer   services   with   the   capacity   for   establishing   systematic   evaluation   and   the   provision   of   sufficient  resources  to  enact  change   There  should  be  an  appointed  Lead  for  Surgical  Cancer  Care  for  each  province  similar  to  the   roles  currently  established  for  medical  and  radiation  oncology.  As  with  the  Heads  of  Medical   and   Radiation   Oncology,   the   Surgical   Lead   should   be   responsible   for   strategic   planning,   budget   allocation,   and   overseeing   both   the   development   of   standards   of   care   and   quality   assurance  activities  within  the  province.     • Given  the  differences  in  cancer  care  organization  between  the  provinces,  it  is  expected  that   this   Surgical   Cancer   Care   Lead   role   will   be   integrated   into   each   province   with   the   consideration  of  current  surgical  care  organization.  Despite  differences  in  implementation,   the  overall  role  and  responsibilities  should  be  similar  between  provinces.       Nationally-­‐implemented  standards  of  care  should  be  developed  for  each  cancer  surgery   • • • • • It   is   recommended   that   a   set   of   national   standards   of   care   be   developed,   similar   to   those   that   currently   exist   in   many   countries.   These   distinct   standards   need   to   be   established   based   on   current   research   and   evidence   in   surgical   cancer   services   to   encourage   excellence   on  both  provincial  and  national  levels.   There   should   be   a   gathering   of   specialty-­‐based   communities   of   practice   focused   on   defining   minimal   standards   for   practitioner   certification   or   training,   the   use   of   consultative   cancer   conferences,  participation  in  national  quality  improvement  registries  and  the  establishment   of  minimal  institutional  and/or  surgeon  case  volumes.   These   specialty-­‐based   communities   must   consider   the   integration   of   specialized   nursing,   critical  care  and  anaesthesia  services,  and  adequate  radiology  and  pathologic  support.   The   evaluation   of   adherence   to   these   surgical   standards   of   care   should   be   integrated   into   existing   national   evaluation   structures   in   order   to   measure   province-­‐level   improvements   and  inter-­‐provincial  variance.   P a g e    12     Purposeful   regionalization   of   cancer   surgical   services   above   and   beyond   simple   consolidation  is  necessary  to  improve  health  care  quality  and  patient  outcomes   • • Volume  itself  is  not  adequate  to  drive  improvements  in  outcomes.  It  is  imperative  that  other   quality  improvement  initiatives  are  incorporated  into  these  centres  of  excellence  based  on   the  requirements  outlined  in  the  cancer  site-­‐specific  standards  of  care,  such  as  staff  training,   technology  uptake  or  infrastructure  requirements.     The  evidence  presented  in  this  report,  accompanied  by  the  reported  citizens’  preference  for   quality   of   care   despite   increased   travel   time,   presents   a   strong   need   and   support   for   regionalization   of   surgeries   for   high-­‐risk,   resource-­‐intensive   cancer.   Non-­‐regionalized   institutions   often   lack   the   organizational   infrastructure   needed   to   host   specialized,   multidisciplinary  care  consistent  with  modern  surgical  cancer  services.     Regionalization  policies  should  be  tailored  to  meet  unique  provincial  needs   It   is   not   feasible   to   establish   one   single   regionalization   policy   for   all   provinces   and   territories   due   to   a   need   to   balance   access   to   care,   geographical   and   health   care   system   governance  and  oversight  practicality  factors  that  are  individual  to  each  jurisdiction.   • The   volume-­‐outcome   effect   is   one   that   should   be   utilized   in   health   care   planning,   but   due   to   the  relative  difference  in  provincial  populations,  it  is  truly  impractical  to  delineate  a  single   national  threshold  for  each  of  the  cancer  types.     Surgical   cancer   care   should   embrace   flexible   implementation   of   regionalization   policies   based  on  access  to  care  concerns  and  patient  preferences   • • • The   issues   relating   to   access   to   care   as   a   consequence   of   regionalizing   to   selected   centres   may   be   ameliorated   through   the   integration   of   innovative   approaches   that   allow   patients   to   remain  at  home  for  part  of  their  cancer  journey.   This   approach   can   take   the   form   of   local   post-­‐surgical   care   centres   and/or   the   ability   to   complete   diagnostic   assessment   pathways,   telemedicine   and   post-­‐surgical   consultations   and  tests  at  local  facilities.     A   structured   benchmarking   process   for   each   specialty   should   be   supported   to   improve   surgical  outcomes  and  inform  policy  decisions   • • • In   the   interest   of   cancer   care   quality   improvement,   all   surgical   care   providers   should   participate   in   national   disease   specific   registries   to   capture   core   data   sets   regarding   patients,  cancers,  procedures,  adverse  events  and  oncologic  outcomes.   Data   collection   must   be   partnered   with   explicit   public   data   reporting,   benchmarking   and   other   quality   improvement   initiatives,   and   ideally   supplemented   and   supported   by   national   society  derived  best  practice  guidelines.   Registries   have   the   potential   to   influence   outcomes   through   a   structured   benchmarking   process,   where   practitioners   are   compared   to   their   peers   via   regular   reporting.   These   reports   can   act   as   decision   making   tools   to   inform   health   care   planners   in   addition   to   encouraging  the  concept  of  positive  deviance  through  the  sharing  of  best  practices  within  a   discipline.   P a g e    13     1.0:  INTRODUCTION   CHAPTER  HIGHLIGHTS   ! ! ! ! Cancer   represents   a   significant   burden   to   individuals,   the   health   system   and   society.   With   population  growth  and  aging,  cancer  incidence  has  increased  and  is  expected  to  continue  on  this   trend.  Surgery  offers  the  best  chance  for  cure  for  the  high-­‐risk  cancers  under  study   There  is  no  one  cancer  surgery  delivery  approach  implemented  in  Canada,  but  there  is  increasing   interest  in  the  regionalization  of  cancer  surgery  services,  an  approach  that  has  the  potential  to   improve  patient  outcomes  as  the  cornerstone  of  a  robust  quality  improvement  initiative   Esophageal,  pancreatic,  liver,  lung  and  ovarian  cancers  require  complex  surgery  to  ideally  reduce   the  high  morbidity  and  mortality  rates,  but  outcomes  are  highly  variable  on  the  stage  of  disease,   extent  of  surgery  and  location  of  tumour   High-­‐risk,   complex   cancer   surgery   carries   a   significant   economic   burden   to   patients   and   the   Canadian   health   care   system.   While   improving   patient   outcomes   is   ultimately   the   most   important   objective,   any   initiative   that   can   reduce   the   burden   of   disease   to   the   patient   while   also  efficiently  using  limited  health  care  resources  is  valuable  to  all  stakeholders   1.1  CONTEXT  OF  HEALTH  CARE  DELIVERY  AND  QUALITY  IMPROVEMENT   The  absence  of  health  impacts  an  individual’s  quality  of  life  on  a  day  to  day  basis,  and  also   affects   the   ability   of   a   society   to   function   effectively   on   a   broader   level.     It   is   for   this   reason   that   governments   and   organizations   devote   considerable   resources   and   energy   to   improving   the   health   status   of   populations.   The   World   Health   Organization   (WHO)   states   that   obtaining   the   highest   possible   standards   of   health   is   a   fundamental   right   of   every   human   being,   including   access   to   timely,   acceptable,   and   affordable   health   care   of   appropriate  quality.1  Deficiencies  in  care  such  as  gaps  in  access  or  poor  delivery  of  services   affect   individuals’   health   and   quality   of   life.   The   challenge   lies   in   devising   a   health   care   system  that  delivers  the  best  possible  care  while  also  balancing  the  other  needs  of  a  society.     Globally,   all   health   care   systems   are   imperfect   models   of   care,   with   no   one   system   representing  the  ultimate,  ideal  system.  A  good  health  care  system  combines  cost  conscious   care   with   optimal   outcomes,   but   the   challenge   lies   in   developing   health   care   delivery   strategies   that   engender   the   greatest   improvement   in   patient   outcomes.   Health   care   organization   and   planning   are   continuous   processes   that   must   adapt   to   the   needs   of   the   population   and   the   availability   of   resources   to   achieve   improvements   in   the   quality   of   health  care.  In  Canada,  the  number  of  new  cancer  cases  is  expected  to  increase  partly  due   to   population   growth   and   aging,   so   there   is   a   definite   need   to   re-­‐evaluate   the   delivery   of   P a g e    14     cancer  care  services  to  ensure  continuous  quality  improvement,  particularly  with  regard  to   resource-­‐intensive  cancer  surgery.       As  the  Canadian  health  care  system  considers  novel  approaches  to  improving  the  delivery   of   cancer   surgical   care,   the   Donabedian   conceptual   framework   for   quality   care   improvement   helps   to   describe   the   relationships   between   the   processes,   structure   and   outcomes   involved   in   health   care   system   improvement.2   Improvements   in   outcomes   are   partly  contingent  on  having  quality  processes,  which  would  entail  the  application  of  best-­‐ practice,   evidence-­‐based   medical   care.   Examples   include   the   initiation   of   clinical   trials   to   develop   new   surgical   techniques   or   test   new   chemotherapy   drugs,   the   increasing   use   of   minimally-­‐invasive   surgical   techniques,   the   establishment   and   measurement   of   quality   benchmarks,   the   creation   of   recommendations   requesting   the   specialization   of   surgeons   and  the  use  of  multidisciplinary  teams  to  coordinate  the  care  of  patients.3  Likewise,  patient   outcomes   can   also   be   improved   through   initiatives   that   focus   on   the   structure   of   care,   such   as   adaptable   hospital   infrastructure   and   strong   regulatory   bodies   and   guidelines.   Interventions   that   target   the   structure   of   care   require   a   large   amount   of   coordination   between   health   care   system   managers,   clinicians,   hospital   administrators,   and   other   key   stakeholders,   but   can   also   have   a   high   impact   on   patient   outcomes.   Structure   and/or   process   improvements   are   inter-­‐related   with   improved   patient   outcomes,   and   the   strongest  quality  improvement  initiatives  are  those  that  integrate  both  factors.     Over   the   past   three   decades,   there   has   been   increasing   interest   in   the   regionalization   of   cancer  surgical  care  into  specialized  centres  of  excellence.  This  approach  to  improving  the   quality   of   cancer   care   offers   an   opportunity   to   integrate   both   process-­‐based   and   structure-­‐ based  improvement.    There  is  considerable  variance  in  how  these  regionalization  strategies   have   been   implemented   across   Canada,   since   these   decisions   are   left   to   the   discretion   of   each   province.   Each   province   has   the   ability   to   establish   quality   improvement   programs   and   approaches   to   health   care   delivery   for   high-­‐risk,   resource   intensive   cancer   services.   This  has  led  to  great  variability  across  the  country  on  surgical  outcomes  and  the  delivery  of   care.  Heterogeneous  quality  of  surgical  care  leads  to  higher  mortality  rates,  longer  length  of   stay,   more   post-­‐operative   complications,   and   greater   chances   of   hospital   readmission,   thus   impacting  patient  health  outcomes  but  also  increasing  the  overall  cost  of  health  care.     Throughout   this   discussion   paper,   we   will   present   an   objective   analysis   and   discussion  about  the  regionalized  quality  improvement  policy  approaches  that  have   been   undertaken   for   the   provision   of   high-­‐risk,   resource   intensive   cancer   surgical   procedures  across  Canada.  We  present  the  potential  benefits  and  limitations  of  this   approach  in  the  Canadian  context  to  assist  with  evidence  based  decision  making  for   health  care  policy  planners  and  decision  makers.   P a g e    15     1.2  THE  BURDEN  OF  CANCER  IN  THE  CANADIAN  HEALTH  CARE  SYSTEM   While   other   chronic   health   conditions   play   a   role   in   the   increasing   burden   placed   on   the   health   care   system,   the   management   of   cancer   care   is   particularly   resource   intensive.   Cancer   poses   a   significant   burden   to   individuals   and   incurs   a   large   financial   cost   to   the   health   care   system.   According   to   the   2014   Canadian   Cancer   Society   report,   43%   of   Canadians  are  expected  to  develop  cancer,  and  25%  are  expected  to  die  from  cancer  over  a   lifetime.4  It  is  the  leading  cause  of  death  in  Canada,  representing  nearly  30%  of  all  deaths  in   2011.4   The   burden   of   disease   associated   with   cancer   is   expected   to   increase   with   the   growing  aging  population,  with  about  89%  cases  developing  in  patients  over  the  age  of  50   which  will  lead  to  an  increase  in  case  complexity  and  costs.4   The  increasing  burden  of  cancer  care  is  coinciding  with  an  economic  climate  where  health   care   expenditures   have   increased   consistently   from   7%   of   the   Canadian   Gross   Domestic   Product   in   1975   upwards   to   10.9%   in   2014,   with   $214.9   billion   spent   in   total.5   This   increase  in  health  care  costs  has  occurred  concurrently  with  rising  inflation,  an  increased   burden   of   chronic   diseases   and   cancer,   more   costly   modern   treatment   approaches   and   a   growing  aging  population,  a  single  factor  that  is  greatly  associated  with  higher  health  care   costs  due  to  increased  co-­‐morbidities.5,6   Cancer   screening   remains   the   most   cost-­‐effective   modality   for   reducing   the   burden   of   disease.   Conversely,   cancer   treatment   requires   a   significant   use   of   health   care   services,7   with   cancer   reported   to   be   the   4th   leading   cause   of   hospital   admissions   in   2008.8   An   estimated  $17.4  billion  dollars  was  spent  on  the  direct  costs  of  health  care  (physician  and   hospital   expenses),   and   the   indirect   costs   associated   with   a   loss   of   productivity   due   to   cancer   therapies.9   Unfortunately,   screening   is   presently   a   small   part   of   the   puzzle,   since   many   cancers   are   detected   at   a   later   stage   and   thus   require   considerably   more   extensive   surgeries  and  oncologic  services  than  if  a  tumour  was  detected  earlier.  This  is  particularly   true  for  lung,  esophageal,  liver,  pancreatic  and  ovarian  cancers.  In  all  these  cases,  surgery   usually   offers   the   greatest   potential   for   cure   relative   to   chemotherapy   and   radiation   therapy.7   Surgery  can  be  used  to  remove  tumours  or  cancerous  cells,  to  shrink  tumours  in   preparation  for  radiation  or  chemotherapy,  or  to  provide  palliative  relief  from  pain.   Whereas   the   procedures   and   risks   for   cataract   removal   or   a   knee   arthroplasty   are   relatively   homogeneous   between   patients,   cancer   surgeries   vary   in   associated   risk   and   resource   implications.   The   five   listed   cancer   sites   are   considered   to   be   particularly   high-­‐ risk,  resource  intensive  to  treat  surgically.  These  complex  procedures  carry  a  higher  risk  of   adverse   outcomes,   although   there   is   considerable   variance   in   adverse   event   rates   depending   on   non-­‐modifiable   patient   related   factors   (i.e.   age,   co-­‐morbidities,   tumour   stage),  or  potentially  modifiable  factors  such  as  access  to  best-­‐practice  care,  the  selection  of   appropriate   resection   approaches,   surgeon   skill   level,   availability   of   hospital   resources,   appropriate   use   of   minimally   invasive   techniques,   access   to   new   technologies,   hospital   infrastructure  and  staff  training.     P a g e    16     Regardless   of   cancer   site,   the   absolute   number   of   cancer   surgeries   increased   from   137,100   in   2004-­‐05   to   146,000   in   2007-­‐08   despite   stable   age-­‐standardized   rates,10     representing   an   evolving  burden  on  health  care  services  in  the  coming  years.  Esophageal,  pancreatic,  liver,   lung   and   ovarian   cancer   surgeries   share   this   upward   trend   and   will   be   introduced   in   the   following  sections.       1.3.1  ESOPHAGEAL  CANCER   The   esophagus   (or   oesophagus)   is   a   multi-­‐layered   muscular   tube   and   forms   a   portion   of   the   digestive   system.   Cancer   can   develop   at   any   site   along   the   length   of   esophagus.   The   two   major   types   of   esophageal   cancers   are   adenocarcinoma   and   squamous   cell   carcinoma.   In   addition,   cancerous   tumours   can   spread   from   other   parts   of   the   body   (e.g.   lungs,   liver,   or   kidneys)  to  cause  metastatic  disease.     The   lifetime   probability   of   developing   esophageal   cancer   is   higher   in   males   (1   in   117)   than   females  (1  in  319)  in  Canada,  with  an  expected  5-­‐year  survival  of  14%  (13%  in  males  and   5%   in   females).4   The   expected   survival   rate   for   surgically   resected   early   stage   cancer   is   much   higher.   However,   the   annual   incidence   rates   are   significantly   increasing   for   males,   showing   an   annual   1.4%   increase   from   2001   to   2010.4   Surgery   remains   the   major   curative   option  for  esophageal  cancer  when  used  in  combination  with  chemotherapy  and  radiation   treatments.11   It   involves   the   surgical   removal   of   part   or   all   of   the   esophagus.   The   degree   of   the  resection  depends  on  the  location,  stage,  histology,  and  grade  of  esophageal  cancer,  as   well  as  the  patient’s  health  status.     450   Number  of  Surgeries   400   354   350   300   314   301   283   304   294   2006   2008   323   334   303   250   200   150   100   50   0   2002   2004   2010   2012   2014   Fiscal  Year   Figure  1.1  Number  of  Esophageal  Cancer  Surgeries  in  Canada  (2004-­‐12)     P a g e    17     According   to   hospital   admissions   data   sourced   from   the   Canadian   Institute   of   Health   Information,   a   total   of   2810   esophageal   cancer   resections   were   performed   in   Canada   (excluding  Quebec)  between  fiscal  years  2004  to  2012.  The  annual  number  of  esophageal   resections   stayed   relatively   stable,   with   a   modest   6%   increase   over   the   9-­‐year   period   (Figure  1.1).   1.3.2  PANCREATIC  CANCER     The   pancreas   is   a   gland   that   produces   digestive   enzymes   that   are   transported  to  the  small  intestine  through  the  pancreatic  duct.  The   pancreas   also   produces   insulin   and   other   hormones.   Cancer   can   develop   in   either   the   pancreas   (typically   endocrine   tumours)   or   the   pancreatic   duct   (typically   exocrine   tumours).   Pancreatic   cancer   is   the  4th  leading  cause  of  cancer-­‐related  death  for  males  and  females.4   The  lifetime  probability  of  developing  pancreatic  cancer  is  1.4%  (1   in  71  for  males;  1  in  69  for  females)  in  Canada  (2009).4  Although  no  significant  changes  in   incidence  and  mortality  rates  are  observed  for  pancreatic  cancer  over  time,  the  expected  5-­‐ year  survival  rate  remains  at  7%  for  both  genders.4     Surgery   could   involve   the   partial   or   total   removal   of   the   pancreas.   The   type   of   surgical   procedure   depends   mainly   on   the   stage   of   cancer,   tumour   location,   and   overall   patient   health.   While   in-­‐hospital   mortality   rates   are   variable   depending   on   resection   type,12   the   overall  mortality  rate  after  pancreatic  cancer  resections  has  been  declining  over  the  last  20   years.12-­‐15  More  complex  procedures  are  associated  with  higher  mortality  rates,  with  most   post-­‐operative  deaths  due  to  surgical  complications.16     In   the   period   of   2004-­‐2012,   a   total   of   3990   relevant   pancreatic   cancer   resections   were   performed   in   Canada   (excluding   Quebec).   The   annual   number   of   pancreatic   cancer   surgeries   increased   77%,   from   337   procedures   in   2004   to   599   in   2012   (Figure   1.2),   representing  a  tremendous  increase  in  the  burden  of  care  to  the  health  care  system.     P a g e    18     700   Number  of  Surgeries   600   599   533   500   416   396   400   401   400   300   456   452   337   200   100   0   2002   2004   2006   2008   2010   Fiscal  Year   2012   2014     Figure  1.2:  Number  of  Pancreatic  Cancer  Surgeries  in  Canada  (2004-­‐12)   1.3.3  LIVER  CANCER   Liver  cancer  is  one  of  the  cancer  types  with  the  fastest  increase  in   incidence   rates   in   Canada.4,7   The   annual   incidence   and   mortality   rates   have   significantly   increased   for   males   (2.3%   and   3.2%,   respectively)  and  females  (2.4%,  and  2%),  with  an  expected  5-­‐year   survival  rate  of  18%.4     Surgery   remains   a   crucial   curative   treatment   option   to   control   such   tumours,   except   for   patients   with   advanced   disease.17,18   Hepatic   resection   is   a   high   risk   procedure   and   there   has   been   an   increase   in   the   number   of   such   procedures   performed   over   the   last   20   years.19,20   Two-­‐thirds   of   liver   resections   are   associated   with   secondary,   metastasized   colorectal   cancers,   which   significantly   increases   the   degree   of   surgical   complexity.   The   impact   of   this   complexity   on   surgical   outcomes   varies   depending   on   the   extent   of   surgical   resection,   cancer   type   and   patient   co-­‐ morbidities.21     There   has   been   an   increase   in   the   number   of   liver   resections   performed   over   the   last   20   years.   A   pan-­‐Canadian   study   reported   an   80%   increase   in   the   rate   of   liver   resections   performed  from  1995  to  2004,  an  increase  from  3.2  cases  per  100,000  adults  in  1995  to  5.9   cases   per   100,000   adults   in   2004.19   More   recent   data   obtained   from   the   Canadian   Institute   of   Health   Information   Discharge   Abstract   Database   (DAD)   reported   that   the   number   of   liver  resection  cases  has  increased  from  693  in  2004  to  1265  in  2012,  representing  an  82%   increase  over  this  9-­‐year  period  (Figure  1.3).   P a g e    19     Number  of  Surgeries   1600   1400   1265   1202   1200   1000   800   600   692   745   796   784   883   981   985   400   200   0   2002   2004   2006   2008   2010   2012   2014   Fiscal  Year     Figure  1.3:  Number  of  Liver  Cancer  Surgeries  in  Canada  (2004-­‐12)   1.3.4  LUNG  CANCER   Lung   cancer   is   the   leading   cause   of   cancer-­‐related   deaths   in   Canada,   accounting   for   27%   of   deaths   in   both   males   and   females   and   an   expected   5-­‐year   survival   rate   of   17%.4     The   lifetime   probability   of   developing   lung   cancer   is   1   in   12   for   males   and   1   in   14   for   females.4   The   annual   incidence   and   mortality   rates   for   males   have   significantly   decreased   from   2000  to  2009,   while  no  significant  changes  have  been  observed   over   the   same   time   for   females.4   Surgical   resection   rates   for   lung   cancer   have   increased   over   this   same   period,   with   in-­‐patient   hospital   records   reporting   a   24%   increase   in   lung   resections,   from   2942   cases   in   2004   to   3795   in   2012   (Figure  1.4).     Surgery   is   the   optimal   curative   option   for   lung   cancer   in   early   stage   disease,   with   the   anticipated   five-­‐year   survival   for   surgically   resected   stage   I   and   II   disease   reported   to   be   near   70%.22   Lung   cancer   surgery   is   performed   to   remove   the   diseased   lung   tissue.   There   are  various  procedure  options  depending  on  the  stage,  location,  and  type  of  cancer.  Some  of   the   common   surgical   procedures   include:   Pneumonectomy   (removal   of   a   whole   lung);   Lobectomy   (removal   of   one   or   more   lobes   of   a   lung);   and   a   Segmentectomy   or   Wedge   resection   (removal   of   part   of   a   lobe   within   a   lung).   The   surgery-­‐related   mortality   rates   depend  on  the  type  of  surgical  technique  applied,  the  extent  of  resection,  and  the  associated   complexities  of  the  procedure.     P a g e    20     4500   Number  of  Surgeries   4000   3500   3000   3795   3542   3516   3383  3325   3343   3199  3239   2942   2500   2000   1500   1000   500   0   2002   2004   2006   2008   2010   2012   2014   Fiscal  Year     Figure  1.4:  Number  of  Lung  Cancer  Surgeries  in  Canada  (2004-­‐2012)   1.3.5  OVARIAN  CANCER   Ovarian   cancer   is   the   most   lethal   gynecological   malignancy.23   This   cancer  is  relatively  uncommon  and,  given  the  non-­‐specific  nature  of   symptoms,   is   often   diagnosed   at   a   later   stage   of   cancer.23,24   Early   diagnosis   often   takes   place   incidentally   during   surgery   for   other   indications.24  The  lifetime  probability  of  developing  ovarian  cancer   in  Canada  is  1.4%  (1  in  72),  and  the  expected  5-­‐year  survival  rate  is   42%.4   Although   there   has   been   a   significant   1.1%   reduction   in   the   annual   incidence   rate   from   2001-­‐2010,   the   annual   mortality   rates   have   significantly   increased   by   2.2%   over   the   same   time   period.4   From   2004   to   2012,   a   total   of   16,949   gynecological   procedures   for   primary   ovarian   and   fallopian   tube   cancers   were   performed   in   Canada,   representing   a   modest   12%   increase   over   the   8   year   period   (Figure  1.5).       Surgery  plays  a  key  role  in  the  management  of  ovarian  cancer,  by  determining  the  extent  of   cancer  spread  (stage)  and  influencing  decisions  on  the  use  of  chemotherapy  in  addition  to   the  removal  of  the  tumour  itself  (debulking).25   Due  to  the  close  interconnectedness  of  the   female   reproductive   system,   ovarian   cancer   surgeries   are   associated   with   an   incredible   heterogeneity   of   risk.   For   instance,   surgery   may   involve   the   removal   of   one   ovary   (unilateral   oophorectomy)   or   both   ovaries   (bilateral   oophorectomy),   the   fallopian   tubes   (salpingectomy),  the  uterus  (hysterectomy),  or  pelvic  and  para  aortic  node  dissection  along   with   the   removal   of   as   much   clinically   evident   tumour   as   possible   (omentectomy).   Depending   on   the   spread   of   tumour   to   nearby   organ   systems,   omentectomy   may   involve   P a g e    21     the   resection   of   bowel,   diaphragm,   or   peritoneum.   Naturally,   the   degree   of   tumour   resection  will  influence  the  morbidity  and  mortality  rates  after  surgery.   Number  of  Surgeries   2500   2000   2030   1949  1951   1877   1848   1 847   1 825   1814   1808   1500   1000   500   0   2002   2004   2006   2008   2010   2012   2014   Fiscal  Year     Figure  1.5:  Number  of  Gynecological  Resections  for  Ovarian  Cancer  Surgeries  in  Canada  (2004-­‐12)     1.4  ECONOMIC  BURDEN  OF  CANCER  SURGERIES  IN  CANADA   The  incidence  of  cancer  is  rising  for  the  most  part  in  Canada4.   Surgery   is   typically   regarded   as   the   preferred,   and   often   definitive,   treatment   option   for   high   risk   cancers.   The   costs   associated  with  cancer  surgical  procedures  are  significant  and  are  associated  with  higher   resource  consumption  due  to  the  high  complexity.  An  Ontario  study  looking  at  cancer  data   from   1997-­‐2007   reported   the   mean   post-­‐diagnosis   costs   of   treatment   for   high-­‐risk,   high   resource   intensive   treatment   for   patients   who   survived   beyond   one   year.26   The   highest   costs  were  reported  for  esophageal  cancer  ($50,620)  from  among  the  21  cancers  studied,   which  can  be  attributed  to  the  need  for  an  extended  length  of  hospital  stay  and  a  greater   need   for   post-­‐treatment   follow-­‐up,   resulting   in   higher   physician   service   costs.   The   post-­‐ diagnosis   costs   for   patients   surviving   beyond   the   first   year   for   other   high-­‐risk   resource   intensive   cancers   were:   pancreas   ($41,846);   liver   ($32,717);   lung   ($29,878);   and   ovarian   cancer   ($29,640).   The   costs   included   expenditures   that   are   associated   with   inpatient   hospital   admissions,   physician   services,   chemotherapy,   radiotherapy,   outpatient   medications,  same  day  surgery,  diagnostic  tests,  long-­‐term  care,  continuing  care  and  home   care.   These   are   imperfect   costing   numbers,   as   they   were   produced   for   wait   times   allocation   discussions   that   included   the   consideration   of   supplementary   health   care   services   for   several  of  these  cancers.  Although  these  costs  do  not  represent  the  actual  costs  of  surgery   in  Canada,  this  is  the  best  available  estimate  on  the  cost  associated  with  the  treatment  of   P a g e    22     high-­‐risk,  resource  intensive  surgical  procedures.  These  costs  are  borne  by  the  health  care   system,   but   also   in   part   by   the   patient   themselves,   with   the   time   required   to   travel   to   appointments,   transportation   costs   and   lost   opportunity   costs   in   addition   to   the   fees   not   covered  by  provincial  health  plans.  Box  1  describes  a  simplified  demonstration  of  the  travel   and  time  costs  associated  with  all  hospital  visits  for  the  treatment  of  esophageal  cancer.     BOX  1:  COSTING  CASE  STUDY  CONSIDERING  APPOINTMENTS  AND  COST  OF  PATIENT  TRAVEL   Patient  Stories   John lives in the centre of Niagara Falls, Ontario. He is currently receiving treatment for esophageal cancer at the Juravinski Cancer Centre (JCC) in Hamilton, ON for medical oncology/radiation oncology). He received surgical care at St. Joseph’s Healthcare Hamilton. The distance from his home to JCC is 71.3km (45 minutes), and 73.6km (50 min) to SJHH. The parking fee at JCC was $90 for 30 uses and $144.50 at SJHH ($3.50/hour with each appointment taking roughly 3 hours + one 2-week pass costing $50) For an average patient requiring only Perioperative chemotherapy: Ø Total of 52 visits, not including in-home weekly Community Care Access Centre (CCAC) coordinated home care, 31 trips to JCC, 21 trips to SJHH over 9 months. Total of 52 return trips with parking each day. Total cost of parking is $415. Total cost of car travel, based on an estimate of 40cents/km: To JCC: 4420.6 km or $1768.24 total. To SJHH: 3091.20km or $1236.48 total. Total cost is $3420 for parking and travel to appointments and nearly 100 hours in travel time (98hrs) For an average patient requiring Pre-operative chemotherapy/radiation therapy: Ø Total of 66 visits, not including 6 months of weekly in-home CCAC visits, 45 to JCC and 21 to SJHH). Total of 66 return trips with parking each day. Cost of parking at JCC is $450 ($90*5 for 50 visits, cheapest) and SJHH is $144.50 for a total parking cost of $595. Total cost of car travel, based on an estimate of 40cents/km: To JCC: 6417km or $2566.80. To SJHH: 3091.20km or $1236.48 total. Total cost is $4398.28 for parking and travel to appointments and 102.5 hours in travel time High-­‐risk,   complex   cancer   surgery   carries   a   significant   economic   burden   to   the   Canadian   health   care   system.   While   improving   patient   outcomes   is   ultimately   the   most   important   objective,   any   initiative   that   can   reduce   the   burden   of   disease   to   the   patient   while   also   efficiently  using  limited  health  care  resources  is  valuable  to  all  stakeholders.     In  the  next  section,  we  will  discuss  the  definition  of  regionalization  of  care,  and  present   associated  challenges  and  limitations  of  this  approach.   P a g e    23     2.0:  DEFINING  AND  EVALUATING   REGIONALIZATION   CHAPTER  HIGHLIGHTS     ! ! ! ! The  definition  of  regionalization  varies  based  on  particular  discipline  of  study.  Furthermore,  our   best   efforts   could   not   find   a   consensus   definition   of   regionalization   of   cancer   surgical   care   in   the   literature   Regionalization  is  more  than  a  mere  shift  of  patients  from  multiple  small  centres  into  one  central   facility,  but  is  instead  a  planned  process  specifically  targeted  at  improving  quality  and  outcomes   of  care.  Based  on  expert  opinion,  we  define  regionalization  as:  “the  deliberate  reorientation  of   cancer   surgical   procedures,   based   on   explicit   and   planned   processes   and   structures,   with   the   intent  of  improving  the  quality  of  care”   A  lack  of  consensus  definition  limits  the  ability  to  evaluate  outcomes  of  ‘regionalization’  across   various   jurisdictions.   If   this   is   a   strategy   to   improve   quality   of   care,   we   need   to   know   exact   steps   taken  to  implement  it  elsewhere.  More  often  than  not,  volume  is  used  as  a  surrogate  measure  to   evaluate  the  impacts  of  ‘regionalization’   Volume   alone   is   not   a   quality   improvement   measure,   but   it   is   an   important   component   of   a   multifaceted   organizational   quality   improvement   approach   that   emphasizes   specialized   staff   training,   application   of   new   knowledge,   and   benchmarking   outcomes   over   time   among   other   initiatives.     2.1  DEFINING  REGIONALIZATION   Regionalization   is   a   broad,   loosely   defined,   and   context-­‐specific   term   that   has   been   used   in   the   disciplines   of   geography,   globalization,   sports,   politics,   health   care   policy   and   administration,  business,  and  emergency  planning,  to  mention  a  few.  The  definitions  vary   based   on   the   specific   discipline   and   intent   of   regionalization.   For   example,   in   emergency   planning  it  is  defined  as,  “the  matching  of  medical  resources  to  patient  needs  to  maximize   health  benefits  and  outcomes  while  minimizing  cost  and  use  of  resources  over  a  specified   geographic   area”.27   On   a   broader   health   care   system   level,   it   is   often   used   to   define   the   transference   of   responsibilities   for   health   care   planning,   organization,   and   delivery   of   health  care  services  to  local  regions.28  The  recent  development  of  Local  Health  Integration   Networks   and   Regional   Health   Authorities   are   examples   of   forms   of   regionalization.   The   main   goals   of   regionalization   of   health   care   services   to   local   regions   include   cost   P a g e    24     containment,   improving   accountability   and   responsiveness   for   local   health   needs,   and   increasing  public  participation  in  health  care  decision-­‐making.28   A   review   of   literature   reporting   the   impacts   of   cancer   surgery   allocation   to   high   volume   centres   showed   that   a   consensus   has   yet   to   be   reached   for   a   consistent   definition   of   regionalization.   This   term   is   often   used   interchangeably   with   ‘centralization’,   ‘consolidation’   or   ‘designation’   of   cancer   care.   According   to   Raval,   Bilimoria,   &   Talamonti   (2010),  “Regionalization  is  a  term  used  to  describe  shifting  a  set  of  procedures  to  certain   hospitals   (e.g.   high-­‐volume)   to   improve   surgical   outcomes”.29   This   means   that   typically   lower   volume   centres   (often   located   in   rural   areas)   are   closed,   directing   the   patients   towards   higher   volume   centres   (typically   in   urban   areas).   Regionalization   of   surgical   procedures   can   also   contain   cost,   increase   accountability   and   responsiveness   to   local   needs,   but   it   can   also   be   directed   at   enhancing   quality   of   care   initiatives   to   improve   outcomes  of  surgery.     The  lack  of  an  available  definition  for  regionalization  in  the  literature  leads  to  an  inability   to   compare   outcomes   of   regionalization   across   published   studies.   As   with   clinical   data   collection,   it   is   impossible   to   consider   differences   between   approaches   when   there   is   no   standardized  terminology.  Processes  that  do  not  necessarily  constitute  regionalization  are   often   classified   as   such,   and   these   confound   the   study   of   regionalization   outcomes.   ‘Passive   Centralization’   is   such   an   example.   Passive   centralization   is   the   ‘unintentional’   consolidation   of   surgical   procedures   to   specific   sites   based   on   natural   geography   (e.g.   a   central  service  site  for  scattered  rural  remote  areas),  for  cost  containment,  or  availability  of   surgeons   and   staff   in   urban   centres.   This   is   in   contrast   to   ‘Active   Centralization’,   which   is   a   process  that  is  actively  undertaken  with  the  intent  to  improve  outcomes  of  surgery  while   ensuring   safe,   effective,   and   efficient   patient   care.   Regionalization   of   high   risk,   high   resource   intensive   cancer   surgical   procedures   in   Ontario   is   an   example   of   ‘active   centralization’.   Given   the   challenges   described   above,   a   more   suitable   definition   of   regionalization  was  required  in  order  to  classify  cancer  treatment  services  as  ‘regionalized’   to   facilitate   making   comparisons   across   the   provinces   in   Canada.   Based   on   the   field   knowledge  of  our  team  members,  review  of  the  literature,  and  discussion  with  experts,  we   defined  regionalization  as:     “the deliberate reorientation of cancer surgical procedures, based on explicit and planned processes and structures, with the intent of improving the quality of care"   P a g e    25     Therefore,  a  mere  increase  in  facility  volume  (centralization)  does  not  necessarily  equate   to   regionalization.   Centralization   is   not   regionalization.   This   definition   closely   matches   with   policies   adopted   by   Cancer   Care   Ontario   to   regionalize   high   risk   cancer   surgical   procedures.     2.2  HOW  IS  REGIONALIZATION  EVALUATED?     How   we   choose   to   evaluate   the   outcomes   of   regionalization   of   cancer   surgery   has   implications   on   effective   resource   utilization   and   on   patient   quality   of   care.   Evaluation   usually   takes   place   using   either   existing   literature   or   retrospectively   reviewed   administrative   databases.   Both   these   approaches   have   certain   trade-­‐offs   based   on   the   quality   of   the   data   and   how   it   applies   to   each   particular   population.   As   noted   above,   existing  literature  may  not  be  compatible  due  to  varying  definitions  of  regionalization.  With   no  clear  definition  of  the  system  under  discussion,  it  may  be  more  akin  to  comparing  apples   with   oranges.   Furthermore,   the   precision   of   the   data   may   be   unknown   due   to   inconsistencies  in  data  collection  or  other  biases.  Administrative  datasets  carry  their  own   threats   to   validity   and   depend   very   much   on   the   standardization   of   data   definitions   and   collection  procedures  between  centres.     Regardless   of   how   well   the   data   is   collected,   the   impact   of   changes   in   a   complex,   multi-­‐ dimensional   system   is   difficult   to   predict.   The   statistical   and   economic   models   require   extensive   data   to   confidently   predict   the   impact   of   changing   certain   factors   on   outcomes   of   interest.  High  quality  data  for  such  models  on  a  national  level,  for  all  possible  variables  that   may  impact  the  performance  of  a  system,  is  not  always  available  and  indeed,  a  large  scale   evaluation   of   regionalization   has   yet   to   be   completed   before   this   report.   Therefore,   considering   all   the   optimization   efforts   presently   being   applied   within   a   complex   system,   we   are   able   to   predict   many   but   not   all   eventual   repercussions   of   these   manipulations.   Sometimes   the   impact   of   changes   to   the   delivery   of   services   implemented   at   a   system   level   is  only  evident  later  in  time.     The   effects   of   early   adoption   of   regionalization   has   been   evaluated   in   the   literature   in  an   attempt   to   demonstrate   the   benefits   of   regionalization   approaches.   In   addition   to   evaluating   the   outcomes   of   regionalization   in   monetary   terms,   evaluations   can   also   focus   on   the   improvement   of   patient   outcomes,   the   successful   implementation   of   high   quality   surgical   procedures   or   adherence   to   evidence-­‐based   guidelines   and   standards,   among   others.   The   majority   of   studies   use   administrative   data   to   compare   outcomes   of   high   vs.   low  volume  surgical  centres.     Volume  as  a  Surrogate  Measure   Institutional  and  surgeon  volume  have  been  used  to  evaluate  the  impacts  of  regionalization   on   outcomes.   While   the   literature   tends   to   demonstrate   a   positive   association   between   a   higher  volume  of  cases  and  better  patient  outcomes,  there  are  other  quality  of  care  factors   P a g e    26     in   high   volume   or   tertiary   care   centres   that   also   affect   outcomes.   Our   literature   review   alludes  to  the  fact  that  volume  alone  does  not  predict  the  reduction  in  mortality  rates  and   improvements   in   overall   survival   that   have   been   seen   over   time.   Other   potential   factors   include   provider   specialty,   the   use   of   multidisciplinary   teams,   specialized   hospital   infrastructure,   improved   surgical   instrumentation,   nursing   staff   experience   and/or   specialization,  rescue  of  ill  patients  and  better  surgical  techniques.     One   of   the   potential   concerns   regarding   the   use   of   volume   as   a   predictor   of   outcomes   is   that   it   may   unfairly   discriminate   against   low-­‐volume   providers   with   favourable   outcomes.170   In   the   interest   of   improving   the   quality   of   care,   it   may   seem   reasonable   to   apply   benchmarking   techniques   to   highlight   centres   that   are   more   successful   at   selected   outcomes.   There   are   concerns   regarding   the   low   statistical   power   associated   with   small   sample  size  in  individual  low  volume  providers  that  may  hinder  a  reliable  comparison  with   either  their  high  volume  counterparts  or  against  a  national  benchmark,  but  this  approach   may  prove  to  be  useful.     Even   with   these   considerations,   the   benefits   of   using   volume   outweigh   its   limitations.   Volume  is  an  easily  measurable  and  frequently  reported  variable,  and  is  clearly  associated   with   improved   surgical   outcomes.29,30   It   is   easily   acquired   from   administrative   data   for   hospitals   and   for   surgeons.   Although   the   debate   remains   unsettled   on   a   direct   causal   relationship   between   volume   and   outcomes,   better   outcomes   are   still   achieved   for   patients   if   surgeries   are   performed   at   high   volume   sites.10   Therefore,   volume   remains   the   best   available  indictor  to  evaluate  surgical  quality  in  the  absence  of  other  viable  alternatives.29   While  understanding  its  limits,  volume  will  be  used  as  a  surrogate  measure  for  evaluating   outcomes  of  cancer  surgical  procedures  in  this  report.   2.3  CHALLENGES  ASSOCIATED  WITH  REGIONALIZATION   While  regionalization  of  cancer  surgical  care  has  a  potential  for  quality  improvement  that   may   serve   the   interests   of   patients,   providers,   and   the   payers,   it   is   not   without   criticism.   Any   shifts   in   the   structure   of   care   directly   or   indirectly   impacts   the   users,   providers,   and   funders   of   healthcare   services.   Table   2.3.1   summaries   some   of   the   speculated   risks   and   benefits  of  regionalization  policies.                 P a g e    27     TABLE  2.3.1:  BENEFITS  AND  UNINTENDED  CONSEQUENCES  OF  REGIONALIZATION   PROS   Improved  patient-­‐based  outcomes   • Lower  mortality  rates;  shorter  length   of  stay       Allows  for  dissemination  of  guidelines  and   standards     Enables  capturing  of  reliable  data  for   evaluating  quality  of  surgical  care     Provides  opportunities  for  partnering  with   multidisciplinary  staff   Encourages  collaboration  for  research  and   innovation  to  improve  processes  of  care   Potential  24/7  access  to  specialty  care  at   designated  centres     Promotes  benchmarking  and  other  surgical   quality  improvement  measures     CONS   Out-­‐of-­‐pocket  patient  cost  and  burden  of  travel32   (also  depicted  in  Box  1)   Loss  of  local  pre-­‐and  post-­‐cancer  surgical   services   Loss  of  surgical  expertise  at  local  hospitals31     Longer  surgical  wait  times  in  high  volume   centres   Potential  lack  of  access  to  informal  care/lack  of   support  for  families  and  caregivers     Providers’  resistance  among  providers  towards   shift  in  established  care  processes   Significant  challenges  to  implementing   regionalization  if  hospital’s  revenue  is  directly   tied  to  the  number  of  treated  patients119     In  short,  the  regionalization  of  surgical  care  often  presents  patients  and  caregivers  with  a   dilemma,   requiring   them   to   choose   between   quality   care   and   accessible   care.   The   concerns   of   providers   revolve   around   resistance   towards   any   shifts   away   from   existing   practices,   financial  implications,  and  disagreements  regarding  the  use  of  volume  as  a  quality  metric.   The   system-­‐level   concerns   focus   on   the   practical   challenges   faced   by   policy   makers   and   healthcare   planners   to   devise   comprehensive   strategies   and   guidelines   that   adequately   address  stakeholder  concerns,  geographic  specificities,  and  healthcare  demands.     The   successful   implementation   of   regionalization   policies   and   standards   may   require   provincial   cancer   agencies   to   provide   strategic   direction   and   leadership   for   surgical   care.   The  Ontario  experience  of  the  regionalization  of  thoracic  surgeries  provides  an  example  of   strategies   that   have   been   adopted   to   assist   with   the   implementation   of   regionalization   policies.   Although   an   evaluation   of   risk-­‐adjusted   outcomes   has   yet   to   be   conducted,   the   success   of   this   policy   has   been   reported   by   the   resultant   increase   in   surgical   volume   at   designated   centres.119   In   Ontario,   other   strategies   that   have   been   utilized   to   promote   the   implementation   of   regionalization   include:   1)   Extensive   discussions   between   concerned   healthcare   planners   (cancer   agency   senior   leadership),   administrators   (Chief   Executives   of   hospitals),  and  thoracic  surgeons;  and  2)  Use  of  additional  funds  to  encourage  compliance   of   hospitals   with   surgical   standards   and   to   lower   wait   times   by   increasing   number   of   procedures.119     Regionalization   policies   that   adequately   address   the   stakeholders’   concerns   can   improve   the   success   of   quality   improvement   initiatives   by   ensuring   coordination,   collaboration,   and   client-­‐satisfaction.     P a g e    28     3.0:  LITERATURE  REVIEW:  EXISTING   EVIDENCE  ON  REGIONALIZATION   CHAPTER  HIGHLIGHTS   ! ! ! Overall,   compelling   evidence   from   numerous   studies   contrasting   provider   (hospital,   surgeon)   volume,  or  provider  specialty  and  improved  outcomes  (in-­‐hospital  mortality,  morbidity,  length  of   stay,  or  overall  survival)  has  been  published  in  last  two  decades.   The   bulk   of   the   literature   and   pan-­‐Canadian   evidence   would   support   the   regionalization   of   lung,   esophageal,  liver,  pancreatic  and  ovarian  cancers  on  hard  objective  measures  such  as  mortality   and  length  of  stay.     The   relative   significance   of   facility   volume,   surgeon   volume,   surgeon   specialty,   or   institution   specialization  on  such  outcomes  is  still  undetermined.     The   extent   of   conflicting   literature   can   be   confusing   for   patients   and   families   seeking   information  to  aid  decision-­‐making  about  where  and  with  whom  to  seek  cancer  surgical   care.  This  can  impact  their  willingness  to  travel  farther  distances  for  care.   Fundamentally,   it   is   not   reasonable   or   practical   to   identify   a   single   number   threshold   that   defines  a  high  volume  centre.     There  is  a  lack  of  studies  that  consider  more  difficult  to  measure  outcomes  such  as  adherence  to   best   practice   standards,   complication   rates,   survival   and   long   term   results,   largely   because   these   o ! ! outcomes  are  not  reliably  captured  in  administrative  datasets,  if  at  all,  and  there  has  only  been   recent  interest  in  the  evaluation  of  such  endpoints  within  the  academic  community   o Survival   and   complications   affect   all   patients   undergoing   the   surgery   and   thus   would   offer   a   significantly   better   power   to   evaluate   the   centralization   of   surgical   cancer   services.   3.1  LITERATURE  REVIEW  AND  METHODOLOGY   In   this   section,   a   brief   summary   of   existing   literature   on   the   modifiable,   provider-­‐related   (facility,  surgeon)  factors  that  can  impact  outcomes  of  care  is  presented.  Any  evidence  on   the  impact  of  regionalization  at  the  patient  or  system-­‐level  will  be  discussed  in  this  section.   In   total,   110   relevant   studies   were   abstracted   from   the   literature   (see   Appendix   1   for   detailed  search  criteria  and  summary  tables).  The  studies  varied  based  on  population  size,   country   of   origin,   and   reported   outcomes.   Although   a   formal   methodological   quality   assessment   was   not   conducted,   only   population-­‐based   studies   were   included   in   order   to   avoid   biases   in   single   institution   studies,   such   as   those   related   to   socioeconomic   status,   P a g e    29     race,   or   co-­‐morbidities.   The   selection   of   population-­‐based   studies   allows   for   a   theoretically   greater   generalizability   of   conclusions   to   an   overall   population   leading   to   the   identification   of  factors  that  can  lead  to  population-­‐wide  improvements  in  care.  Dimick  et  al.  supported   the   claim   that   data   from   single-­‐centre   studies   is   not   a   good   representation   of   national   trends  in  rates  of  procedures  performed  over  time.20  The  findings  of  such  studies  may  not   be   generalizable   to   a   broader   range   of   institutional   settings   given   the   differences   in   standards   and   procedures   used   to   provide   care.   Figure   3.1   demonstrates   the   flow   of   the   completed  literature  search  and  review  process.   3.2  VOLUME-­‐OUTCOME  ASSOCIATION   The   positive   impact   of   facility/surgeon   volume   on   post-­‐operative   outcomes   for   various   cancer   types   has   been   demonstrated   extensively   in   the   literature.   The   results   of   these   studies   have   prompted   various   researchers   and   policy   makers   to   urge   for   the   regionalization   of   complex   surgical   procedures.21,34-­‐39   It  is  important  to  note  that  there   was   a  disparity  in  the  number  of  studies  reporting  on  volume-­‐outcome  association  depending   on  the  cancer  type,  with  liver  and  ovarian  cancer  represented  by  fewer  analyses  than  the   other   cancers.   In-­‐hospital   mortality   was   the   most   reported   outcome   while   reporting   of   other  outcomes  such  as  length  of  stay,  morbidity,  and  overall  survival  was  inconsistent  or   incomplete.  Only  significant  associations  after  adjusting  for  patient  case-­‐mix  are  reported   over   the   course   of   this   review   document.   It   is   also   important   to   note   that   the   type   of   procedure   being   evaluated   in   a   study   was   identified   and   reported.   The   procedures   differ   in   complexity   and   associated   mortality   and   morbidity   rates,   hence,   the   reports   that   fail   to   account  for  the  types  of  resection  could  potentially  bias  the  results.   In   spite   of   inherent   differences   in   surgical   approaches   and   the   complexity   of   procedures,   numerous  studies  reported  improved  clinical  outcomes,  even  if  not  statistically  significant,   at  high  volume  centres  for  esophagus,  lung  and  pancreatic  cancers  (see  Appendix  1).  These   associations  held  true  after  adjusting  for  patient  case-­‐mix  for  the  majority  of  studies  (Table   3.1).  The  strongest  evidence  for  a  reduction  in  mortality  rates  with  increasing  volume  was   reported   for   esophageal   cancer   surgery.   A   large   body   of   evidence,   including   systematic   reviews  and  meta-­‐analyses,  support  referral  to  regional,  specialized,  high-­‐volume  hospitals   for   esophageal   surgical   procedures.18,40   The   operative   mortality   reported   in   these   studies   varied  widely  possibly  due  to  the  heterogeneity  of  the  study  populations  or  differences  in   study  periods.18,19   Relatively  fewer  studies  reported  on  length  of  stay  and  overall  survival,   however,   the   length   of   period   over   which   these   survival   rates   were   calculated   was   not   clearly  stated.41-­‐47         P a g e    30             Identification     Screening             24  additional  records  identified   through  hand  searching  references       1751  records  after  duplicates  removed         5923  records  identified  through   database  searching     1661  records  excluded   after  title  screening     269  titles  included     13  records  excluded  after   abstract  screening           Eligibility     256  full-­‐text  articles   assessed  for  eligibility     30  full-­‐text  articles   excluded     Single  centre  study   Data  collected  before  1990     Published  before  2 000     No  Data  Available           Included     39  Esophagus   110  studies  included  in   qualitative  synthesis     21  Pancreas     12  Liver       12  Ovarian       26  Lung     Figure  3.1:  Flow  diagram  depicting  literature  screening  process   The   volume–outcome   relationship   for   hepatic   cancer   surgical   procedures   has   rarely   been   explored.48   The   most   commonly   reported   outcomes   include   mortality   and   length   of   stay.18-­‐ 21,37,38   While   an   overall   downward   trend   in   mortality   and   morbidity   rates   post-­‐ hepatobiliary  surgical  procedures  was  observed  in  general,  a  significantly  larger  decrease   in   mortality   and   length   of   stay   at   high   volume   centres   was   reported.20,38,49-­‐54   In   a   pan-­‐ Canadian   study,   the   increase   in   number   of   procedures   being   performed   in   high   volume   centres   was   associated   with   a   reduction   in   the   rates   of   perioperative   mortality   (from   5.8%   P a g e    31     in   1995–1996   to   4.2%   in   2003–2004).19   The   mean   length   of   stay   decreased   from   12.1   to   13.2  days  over  the  course  of  1995  to  2004.19     The   most   inconsistent   and   conflicting   evidence   in   favour   of   centralizing   surgical   cancer   services   was   found   for   ovarian   cancer   surgery.   The   absolute   reduction   in   mortality   for   ovarian   cancer   after   regionalization   was   less   than   that   shown   for   lung,   esophagus,   and   pancreatic  cancer  procedures.55   In-­‐hospital  mortality  was  seldom  discussed  as  an  outcome   in  these  studies.  This  may  reflect  the  proportionally  lower  mortality  in  these  cases  and  as   such,   discrepancies   are   more   difficult   to   detect.   While   treatment   in   tertiary,   high   volume   centres   was   shown   to   be   an   independent   prognostic   factor   predicting   better   prognosis   and   survival   of   ovarian   cancer,     staging   of   disease   and   complete   cytoreduction   also   predicted   outcomes  of  surgery.23,56,57  In  ovarian  cancer  surgery,  hospital  volume  was  reported  to  be   associated  with  other  metrics  of  quality  care  including  lower  re-­‐operation  rate,  provision  of   guideline-­‐based   care,   better   staging   for   early-­‐stage   disease,   more   extended   cytoreduction   for   advanced-­‐stage   cancers,   performance   of   lymphadenectomy,   receipt   of   post-­‐operative   chemotherapy,   fewer   prolonged   hospitalizations,   and   lower   ostomy   (complication)   rates.55,58,59   The   adherence   to   guidelines   is   particularly   pertinent   as   it   is   associated   with   better  outcomes  and  the  literature  reports  that  low  volume  hospitals  and  surgeons  are  less   likely  to  adhere  to  the  guidelines.60   Across   all   cancer   types   studied,   post-­‐operative   complications   are   a   major   cause   of   in-­‐ hospital   mortality.16   An   improvement   in   rates   of   complications   at   high   volume   centres   were   reported   in   a   few   studies.12,13,16,58,59,61-­‐64   Regrettably,   there   were   no   single   standard   criteria  for  classification  of  post-­‐operative  complications  for  any  of  the  cancer  types,  so  it   was   difficult   to   assess   reported   complication   severity.   Wright   et   al.   found   that   the   high   volume   centres   for   ovarian   cancer   had   higher   adjusted   complication   rate,   yet,   showed   better   survival.59   Further   analysis   concluded   that   patients   experiencing   a   complication   at   low-­‐volume   hospitals   were   nearly   50%   more   likely   to   die   as   compared   to   those   in   high   volume   centres.   Ghaferi   et   al.   also   found   similar   complication   rates   for   gastrectomy,   pancreatectomy,   and   esophagectomy   at   different   hospital   volumes,   however,   a   higher   failure   to   rescue   patients   from   adverse   events   was   associated   with   low   volume   hospitals.66   Therefore,  the  ability  of  an  institution  to  manage  complications,  even  if  they  do  not  reduce   the  overall  rate  of  complications  themselves,  seems  to  predict  mortality.   Contrarians  of  volume-­‐outcome  associations  argue  that  it  is  incorrect  to  attribute  positive   outcomes   to   high-­‐volumes   alone,   a   position   that   is   supported   by   some   literature.   The   volume-­‐outcome   relationship   could   be   a   result   of   various   complex   interactions   between   factors   that   may   also   lead   to   improved   outcomes.   For   example,   the   benefits   of   restricting   complex   high   risk   procedures   to   high-­‐volume   centres   could   be   a   result   of   superior   surgical   outcomes   at   a   particular   site   leading   to   more   referrals,   which   could   increase   procedural   volume   at   that   site.11   The   improvement   in   outcomes   could   also   be   attributed   to   better   P a g e    32     patient   selection,   advanced   operative   techniques,   better   perioperative   and   post-­‐operative   management  of  cases,  and  specialized  surgeon  training.50,68     Surgeon  Volume  and  Outcomes   The   association   between   surgeon   volume   and   mortality,   morbidity,   or   length   of   stay   was   less   frequently   reported   in   the   literature.   Only   twelve   studies   included   in   this   review   reported  the  impacts  of  high  surgeon  volume  on  patient  specific  outcomes.16,23,25,48,55,58,69-­‐73   Although  reduction  in  mortality  rate  for  pancreatic,16,72,73  esophageal,41,69,74,75  lung,69-­‐71and   ovarian55   cancer   was   reported   with   increased   surgeon   volume,   arguments   remain   unsettled   for   the   benefits   of   consolidating   surgeries   to   high   volume   surgeons   vs.   high   volume   centres.   While   studying   the   impacts   of   volume   on   short   term   morbidity,   30-­‐day   mortality,   and   long   term   prognosis,   Sundelof   et   al.   reported   a   high   level   of   association   between  surgeon  and  hospital  volume,  hence  the  reported  outcomes  were  similar  for  both   the   variables.76   Uncertain   results   were   found   in   studies   that   tried   to   find   the   relative   significance  of  one  variable  or  the  other.  For  instance,   Derogar  et  al.  compared  the  relative   significance   of   hospital   and   surgeon   volumes   on   short   term   and   long   term   risk   of   mortality.45   After  adjusting  for  the  effect  of  hospital  volume  and  surgeon  clustering  effect,   short   term   mortality   risk   was   not   significantly   associated   with   surgeon   volume.   The   authors  asserted  the  conclusion  that  factors  related  to  hospitals  (i.e.  intensive  care  units)   but   not   volume,   impacted   short   term   mortality   risk.   However,   the   risk   of   long-­‐term   mortality   was   significantly   associated   with   surgeon   volume,   even   after   adjusting   for   hospital  volume  and  surgeon  clustering  effect.45   3.3  PROVIDER  SPECIALTY  AND  OUTCOMES     Surgeon   speciality   was   not   studied   in   the   majority   of   the   literature   associated   with   pancreatic,  hepatobiliary,  lung,  or  esophageal  cancers,  with  the  exception  of  ovarian  cancer.     In   lung   cancer   regionalization   literature,   the   few   studies   that   investigated   provider   speciality   reported   non-­‐significant   differences   between   rates   of   mortality,   length   of   stay,   and  overall  survival  when  comparing  patients  outcomes  of  surgeries  performed  by  thoracic   vs.   general   surgeons.78-­‐80   The   outcomes   were   consistent   across   pneumonectomy,   lobectomy,   and   general   lung   resections.   However,   the   number   of   lung   cancer-­‐specific   studies  is  too  small  for  any  conclusive  conclusions  to  be  made.     In  gynecological  surgery,  the  extensive  literature  has  reported  that  initial  surgical  care  by  a   gynecological   oncologist   has   been   found   to   be   associated   with   a   higher   likelihood   of   comprehensive   surgical   staging,   optimal   primary   cytoreductive   surgery,   adherence   to   recommended   guidelines,   and   superior   long-­‐term   outcomes.81-­‐86   The   advantages   associated   with   specialized   surgeons   also   appears   to   extend   to   advantages   to   using   specialized   hospitals   to   perform   ovarian   cancer   surgery.   Vernooij   et   al.   reported   that   specialized   hospitals   are   shown   to   have   significantly   lower   mortality   and   better   overall   P a g e    33     survival   compared   to   general   hospitals.23   In   addition,   the   level   of   specialization   of   the   hospital  was  positively  associated  with  adequately  staged,  optimally  debulked  patients.23     The  interplay  between  surgeon-­‐specific  factors  and  hospital-­‐related  factors  and  outcomes   was   demonstrated   in   some   studies   as   well.   For   example,   low   volume   surgeons   at   high   volume   centres   were   reported   to   have   comparable   outcomes   for   hepatic   procedures   as   those  of  high  volume  surgeons  at  the  same  centre.31,87   Similarly,  gynecological  oncologists   working   in   cancer   centres   were   reported   to   be   twice   as   likely   to   attain   optimal   cytoreduction,  when  compared  to  general  gynaecologists.88   TABLE  3.1.  SUMMARY  OF  STUDIES  REPORTING  POSITIVE  OUTCOMES  FOR  HOSPITAL  AND   SURGEON  VOLUME       Positive  Outcomes  for  Hospital   Volume   Positive  Outcomes   for  Surgeon   Volume   Esophagus         Pancreas   Mortality   Complications   Length  of  Stay   Survival   Mortality   [41,45,69,74,75]   [63]   -­‐-­‐   [45,76]   [16,56,72,73]         Liver         Lung           Ovarian           Complications   Length  of  Stay   Survival   Mortality   Complications   Length  of  Stay   Survival   Mortality   Complications   Length  of  Stay   Survival   Mortality   Complications   Length  of  Stay   Survival   [34,36,46,62,113,134-­‐139,141,145,151]       [62,138]   [138,139,145]   [76,118,141,148]   [12,13*,15,16,64*,94,101,103,104,106,   107]   [13*,15,16*]   [16*,64*,101,103,104,106]   [13*,16,105]   [18,19*,20,21,  37,38]     -­‐-­‐-­‐   [19*,37,38]   [48*]   [34,70,88,112,116,119,121,122,125,128*]   -­‐-­‐-­‐   [88,121]   [118,122]   [55*,25*,58*,59,158,57,160*]     -­‐-­‐-­‐   [55,58,59]   [25*,57,58,158,160*]   [16*]   [16*]   -­‐-­‐-­‐   -­‐-­‐-­‐   -­‐-­‐-­‐   -­‐-­‐-­‐   [48]   [69,70,71]   -­‐-­‐-­‐   -­‐-­‐-­‐   -­‐-­‐-­‐   [55,24*]     -­‐-­‐-­‐   [25*,55*,58]   [23,58*]   *Association  insignificant  after  adjustment  for  case-­‐mix   3.4  EVIDENCE  FOR  REGIONALIZATION     Early   adoption   of   cancer   surgery   regionalization   has   led   to   the   initiation   of   studies   reporting   on   outcomes   of   regionalization   as   a   cornerstone   of   quality   improvement   programs.  The  impacts  of  regionalization  on  surgical  procedures  have  been  studied  across   various  settings  and  in  different  jurisdictions.     P a g e    34     The   literature   appears   to   report   a   positive   association   between   regionalization   and   morbidity   and/or   mortality   across   all   five   cancer   types,   even   if   it   is   a   weak   association.   Across   many   jurisdictions,   the   literature   consistently   reports   improvements   in   survival   (2-­‐ 5   year),   and   both   post-­‐operative   and   30-­‐day   mortality   and   morbidity   rates   following   the   implementation   of   regionalization   policies   for   esophagectomy.11,40,89-­‐91   This   observation   appears   to   hold   true   with   pancreatic   surgery,   which   has   seen   a   gradual   increase   in   the   number  of  resections  performed  in  high  volume  centres.12,13   One  study  in  the  Netherlands   showed   a   significant   reduction   in   rates   of   mortality   (24%   to   2.6%),   and   morbidity   (82%   to   38%)   after   pancreatic   cancer   surgery   while   comparing   pre-­‐and-­‐post   regionalization   outcomes.13   Similarly,   the   proportion   of   ovarian   cancer   patients   undergoing   ovarian   resections   in   high   volume   centres,   and   the   proportion   of   procedures   being   done   by   high   volume   surgeons   is   reported   to   have   increased   over   the   last   14   years.55   Regardless   of   cancer   site,   significant   reductions   in   mortality   rates   were   reported   in   studies   comparing   results   of   providing   care   in   designated   centres   pre-­‐   and   post-­‐centralization.56,92   A   significant   regionalization   of   hepatic   cancer   surgical   procedures   over   the   last   decade   to   high-­‐volume  centres  has  also  been  reported  in  the  US  and  Canada,  with  reported  significant   improvements   in   mortality   rates.19,39   This   consistent   reporting   of   outcomes   improvement   across   all   studied   disease   sites   leads   one   to   believe   that   the   regionalization   of   surgical   cancer  services  offers  positive  benefits  to  patients.   3.5  OPTIMUM  VOLUME  THRESHOLD     One   of   the   main   objectives   of   this   systematic   review   was   to   provide   an   evidence-­‐based   volume   threshold   for   each   cancer   type   at   which   optimum   outcomes   could   be   achieved.   Early  results  of  the  regionalization  of  pancreatic,  esophageal,  and  liver  cancer  procedures   using  different  volume  thresholds  and  standards  has  been  reported  in  the  literature.     The   extent   of   the   literature   studying   volume-­‐outcome   associations   for   esophagectomy   allowed   pooling   of   data   and   controlling   for   potential   confounding   variables   (e.g.   age,   sex,   co-­‐morbidities)   for   meta-­‐analyses.   While   positive   results   have   been   reported   from   the   meta-­‐analysis   of   data,   it   has   been   reported   that   it   is   necessary   to   perform   at   least   20   esophagectomies  per  year  to  reduce  mortality  to  less  than  5%.91  This  observation  appears   to   be   supported   by   other   literature   reporting   higher   and   widely   varying   mortality   rates   for   less  than  20  esophagectomy  cases  per  year.93  For  pancreatic  cancer,  a  volume  threshold  of   10  pancreatic  procedures  per  year  was   proposed.94   Lung  cancer  presents  a  curious  case  in   that   although   the   annual   case   volume   for   lung   cancer   is   larger   than   that   of   pancreatic   or   esophageal  cancer,  there  is  relatively  limited  evidence  to  support  volume  standards  in  lung   cancer   surgery95   and   there   were   no   common   criteria   across   studies   to   indicate   potential   volume   cut-­‐offs.   Lastly,   according   to   the   ovarian   cancer   guidelines   disseminated   by   the   National   Comprehensive   Cancer   Network   (NCCN),   the   hospitals   are   classified   as   high-­‐ volume   if   they   perform   >20   cases   of   ovarian   cancer   per   year,   and   the   surgeons   are   P a g e    35     classified   as   high-­‐volume   if   they   perform   >10   cases   of   ovarian   cancer   per   year.60   The   volume   cut-­‐off   for   liver   cancer   surgeries   was   not   reported   in   the   included   literature.   Relatively  small  number  of  studies  limits  our  ability  to  provide  such  comparisons  as  well.     The   review   of   literature   demonstrated   disagreements   in   the   use   of   volume   thresholds   to   classify   institutions   or   surgeons   as   high   or   low   volume.   Studies   comparing   the   annual   volume   of   procedures   often   had   overlapping   volume   thresholds   defining   high   vs.   low   volume  categories.96  This  further  complicates  the  process  of  comparing  studies  to  identify   volume   thresholds   in   order   to   achieve   best   possible   patient   outcomes.   In   addition,   it   has   been   suggested   that   the   open-­‐ended   volume   thresholds   in   the   highest   volume   categories   may  underestimate  the  actual  mean  number  of  surgeries  performed  per  year.97  Therefore,   the   outcomes   demonstrated   in   high   volume   categories   may   be   a   result   of   actual   (higher)   mean  volume  rather  than  the  threshold  used  for  the  high-­‐volume  category.  This  threshold   is   expected   to   vary   for   each   of   the   cancer   types   given   the   varying   levels   of   complexities   associated  with  each  of  the  surgical  procedures.     Fundamentally,   it   is   not   reasonable   to   identify   a   single   number   threshold   that   defines   a   high  volume  centre.  Practically  speaking,  the  volume-­‐outcome  effect  is  demonstrated  even   with   institutions   with   the   highest   case   volume,   although   it   eventually   does   not   make   a   clinically  significant  difference  and  would  be  impractical  to  implement  a  defined  value.  As   such,  the  volume-­‐outcome  effect  is  one  that  should  be  utilized  in  health  care  planning,  but   not  with  such  granularity  as  to  demand  a  threshold.   3.6  LIMITATIONS     The   association   between   volume   consolidation   and   positive   outcomes   is   not   without   criticism.   In   general,   the   literature   is   lacking   consensus   on   the   relative   importance   of   facility   vs.   surgeon   volume   at   affecting   the   outcomes   of   surgery.   The   heterogeneity   in   patient   co-­‐morbidities,   surgeon   specialty,   facility   size,   and   volume   thresholds   limits   the   ability   to   draw   valid   comparisons   between   different   studies   while   varying   study   designs,   non-­‐consensus-­‐based   volume   thresholds,   conflicting   findings,   small   sample   sizes,   low   quality   of   data   sources,   heterogeneity   of   comparison   groups,   and   definitions   of   outcome   measures   limits  the  ability  to  provide  conclusive  results.29  The  small  number  of  available   studies   and   heterogeneity   of   data   limits   the   ability   to   conduct   a   meta-­‐analysis   for   most   cancer   surgeries.95   Finally,     long   term   survival   assessment   is   a   highly   patient-­‐important   metric,   but   it   is   also   dependent   on   tumour   stage,   which   is   not   always   included   in   the   analysis.89   Even  though  the  majority  of  studies  reported  improved  outcomes  in  high  volume  centres   or  for  high  volume  surgeons  relative  to  low  volume  hospitals  and  surgeons,  the  difference   was   not   always   statistically   and/or   clinically   significant.   This   limits   the   ability   to   draw   statistically  sound  conclusions  on  whether  the  high  volume  centres  and  surgeons  actually   P a g e    36     show   improved   outcomes.   In   addition,   the   quality   of   data   used   to   study   volume-­‐outcome   association  has  inherent  limitations.  Most  of  the  studies,  if  not  all,  used  administrative  data   to   study   the   volume-­‐outcome   association.   Such   data   is   collected   for   administrative   and   billing  purposes  and  thus  direct  cause-­‐and-­‐effect  relationships  cannot  be  determined  with   confidence   from   the   analysis   of   such   data   given   the   possible   differences   in   data   coding   practices.     Lastly,  the  review  found  that  comprehensive  reporting  of  cancer  surgery  systems  in  place   for   the   delivery   of   high-­‐risk,   resource   intensive   surgical   procedures   in   each   Canadian   province  or  territory  is  largely  non-­‐existent.  While  steps  to  improve  quality  of  surgical  care   have  been  initiated  in  other  provinces,  Ontario  is  the  only  province  with  formal  steps  and   published   guidelines   for   regionalization   of   certain   high-­‐risk   procedures   (pancreatic,   hepatic,  esophagus,  lung,  gynecological  cancers),17,60,98  although  a  few  other  provinces  have   taken  steps  to  actively  regionalize  some  cancer  site  procedures.  In  addition,  there  is  little   research  into  whether  these  efforts  have  helped  or  hindered  the  quality  of  cancer  care  from   a  system-­‐level  perspective.  Further  research  is  required  to  compare  system-­‐level  outcomes   such  as  administration  costs  prior  to  regionalization  activities  with  the  post-­‐regionalization   period  in  order  to  adjudicate  the  overall  benefit.   3.7  CONCLUSION       The   success   to   which   consolidation   of   care   to   high   volume   centres   or   surgeons   improves   patient   outcomes   has   yet   to   be   fully   studied,   although   it   appears   that   the   potential   to   greatly  impact  patient  survival,  length  of  stay  and  morbidity  is  possible,  if  implemented  in   ideal  situations.  The  conflicting  evidence  of  the  relative  significance  of  provider  (surgeon,   hospital)   volume,   surgeon   specialty   or   hospital   designation   alludes   to   the   fact   that   improvements   in   outcomes   are   more   likely   to   be   caused   by   the   cumulative   effect   of   practitioner   training   and   multidisciplinary   collaboration,   an   adequate   number   of   practitioners   in   one   centre,   balanced   case   volume   and   complexity,   readily   available   consultation  services,  specialized  nursing  and  anesthesia  services,  and  adequate  radiology   and  pathologic  support.       Volume   is   not   a   stand-­‐alone   factor   explaining   all   improvements   in   outcomes.   The   presence   of   multidisciplinary   teams,   well-­‐trained   staff,   and   the   resources   available   to   efficiently   recognize   and   treat   post-­‐operative   complications   are   all   factors   associated   with   high   volume   centres   that   cumulatively   improve   outcomes   of   care.   The   best   possible   surgical   outcome  may  be  achieved  through  the  utilization  of  expert  surgeons  in  high  volume  centres   equipped   with   cutting   edge   technology,   perioperative   institutional   standards   and   supported   by   specialized   multidisciplinary   teams.   The   experience   of   surgical   team,   their   ability   to   rescue   complicated   cases,   and   availability   of   multiple   surgeons   for   cooperation   and  planning  of  complex  cases  may  result  in  better  outcomes  of  care.     P a g e    37     A   majority   of   the   literature   points   towards   improved,   even   if   not   statistically   significant,   outcomes   associated   with   high   volume   hospitals.   Evidence   regarding   the   impact   of   high   surgeon   volume   is   comparatively   less   clear.   Skepticism   remains   in   the   use   of   findings   from   the  literature  to  provide  an  evidence-­‐based  discussion  on  the  impact  of  regionalization  on   surgical   outcomes.   The   limited   strength   of   evidence   and   heterogeneity   of   results   should   be   accounted  for  in  identifying  volume  thresholds.     Fundamentally,  the  volume  of  cases  completed  at  a  given  centre  is  not  an  all-­‐encompassing   metric.   It   is   not   just   the   simple   sum   of   all   the   components   in   care,   as   those   components   are   inextricably   linked   and   overlapping.     This   is   what   makes   dissecting   down   further   impossible   and   explains   the   heterogeneity   of   results.   It   is,   however,   the   most   robust   statistic   available   to   decision   makers   and   as   such,   is   useful   in   measuring   and   evaluating   institutions.  It  is  potentially  best  utilized  as  a  starting  point  for  the  development  of  a  cancer   surgery  quality  improvement  strategy  and  must  be  a  key  contributor  to  that  conversation.                                       P a g e    38     4.0:  ANALYSIS  OF  PAN-­‐CANADIAN   DATA   CHAPTER  HIGHLIGHTS   ! ! ! ! Hospital   volume   was   a   significant   predictor   of   mortality   for   four   of   the   five   cancer   surgeries,   even   in   the   presence   of   surgeon   volume   and   specialty.   Surgeon   volume   significantly   predicted   both   outcomes   only   for   liver   cancer,   while   controlling   for   the   influence   of   hospital   related   factors.  However,  it  is  necessary  to  note  that  these  three  variables  are  not  mutually  exclusive.  A   list   of   factors   significantly   associated   with   in-­‐hospital   mortality   and   length   of   stay   in   adjusted   analysis  is  presented  in  Table  4.0a.   Ontario,  being  the  most  populous  province,  showed  the  biggest  reduction  in  number  of  hospitals   performing   these   surgeries   for   lung,   liver,   pancreatic   and   esophagus   cancers.   Differences   exist   in   provinces   in   crude   in-­‐hospital   mortality   rates   and   length   of   stay.   The   trends   vary   based   on   specific   cancer   surgery   of   interest,   with   no   one   province   demonstrating   the   best   outcomes   across  all  cancer  types     Travel   times   slightly   increased   between   the   beginning   and   the   end   of   the   study   period   almost   irrespective   of   patient   location.   As   surgical   centres   closed,   whether   as   a   result   of   passive   or   active   regionalization   activities,   the   catchment   areas   of   the   remaining   centres   widened,   revealing   a   longer   travel   time   for   all   but   the   most   urban   centres.   Across   all   cancer   sites,   the   proportion   of   patients   who   had   to   travel   more   than   one   hour   to   access   care   increased   as   the   number  of  centres  that  provide  surgical  cancer  services  decreased.  This  data  confirms  that  the   regionalization   of   some   services   to   high   volume,   central   locations   occurs   at   the   expense   of   patient  travel  time.   There   were   differences   between   the   provinces   in   age-­‐adjusted   resection   rates   per   capita.   However,  a  clear  association  between  in-­‐hospital  mortality  rates  and  associated  resection  rates   could  not  reliably  be  established.     Limitations:     ! ! As  can  be  expected  based  on  the  population  distribution  of  Canada,  there  is  a  large  variation  in   the  number  of  surgical  cases  completed  per  province,  even  temporally  during  the  study  period.   This   limits   the   ability   to   perform   year-­‐to-­‐year   comparisons   on   mortality   rates,   so   a   9-­‐year   (2004-­‐ 2012)   average   mortality   rate   is   reported   throughout   this   section   to   facilitate   comparisons   and   discussion.     The   small   annual   surgical   volume   in   less   populous   provinces   limited   our   ability   to   reliably   determine  inter-­‐provincial  differences  in  outcomes  of  interest  in  regression  and  survival  analysis.       P a g e    39     ! ! It  should  be  noted  that  the  surgeon  specialty  could  not  be  independently  verified,  which  limits   our   ability   to   draw   accurate   comparisons   between   specialist   and   general   surgeons   in   the   outcomes  of  interest.   The   inter-­‐provincial   differences   could   not   reliably   be   determined   given   the   small   surgical   case   volume  in  certain  jurisdictions.   TABLE  4.0A:  SIGNIFICANT  ASSOCIATIONS  BETWEEN  SELECTED  COVARIATES  AND  MORTALITY   Covariates     Esophagus   IHM   LOS   Pancreas   IHM   LOS   Liver   Lung   IHM   LOS   IHM   Ovary   LOS   IHM   LOS   Hospital  Volume           ⨯   ⨯           Surgeon  Volume   ⨯     ⨯         ⨯     ⨯     Surgeon  Specialty   ⨯   ⨯   -­‐-­‐   -­‐-­‐   -­‐-­‐   -­‐-­‐   ⨯   ⨯       Age                         Co-­‐morbidity  Score                       Sex   ⨯     ⨯   ⨯     ⨯       -­‐-­‐   -­‐-­‐   Resection  Type   ⨯                 ⨯     Primary  vs.     Secondary  Cancer   -­‐-­‐   -­‐-­‐   -­‐-­‐   -­‐-­‐       -­‐-­‐   -­‐-­‐   -­‐-­‐   -­‐-­‐   Year   ⨯   ⨯   ⨯             ⨯     Ontario  vs.     Rest  of  Canada   ⨯   ⨯   ⨯     ⨯     ⨯     ⨯   ⨯   IHM=  In-­‐hospital  mortality;  LOS=  30-­‐day  Length  of  stay   =  Significant  association;               ⨯ =  Insignificant  association   P a g e    40     4.1  APPROACH  AND  OBJECTIVES   This   section   provides   a   detailed   discussion   on   the   current   state   of   cancer   surgeries   for   esophageal,   pancreatic,   liver,   lung,   and   ovarian   cancer   surgeries   across   Canada.   The   primary  objective  of  our  analyses  were:  1)  To  provide  quantitative  evidence  on  the  impacts   of  regionalization  activities  on  patient  outcomes  established  through  risk-­‐adjusted  volume-­‐ outcome   associations   for   the   five   surgical   procedures   of   interest;   2)   To   demonstrate   the   disparities  across  the  provinces  in  regards  to  patient-­‐based  outcomes  of  interest;  and  3)  To   present  the  temporal  and  geographic  trends  in  the  distribution  of  surgical  care  across  the   provinces,   and   associated   burden   of   travel.   The   main   outcomes   of   interest   included   in-­‐ hospital   mortality,   30-­‐day   length   of   stay,   per   capita   resection   rates,   and   travel   distances.   In   addition,   the   potential   years   of   lives   and   the   number   of   hospitalization   days   saved   through   consolidation   to   high   volume   centres   are   also   presented   in   this   section.   A   detailed   description   of   data   selection   and   statistical   analysis   methodologies   are   described   in   Appendix  2  and  3.     The   national-­‐level   data,   excluding   Quebec,   was   extracted   from   the   Canadian   Institute   of   Health   Information   Discharge   Abstract   Database   (DAD),   which   contains   hospital   records   for   in-­‐patient   admissions.   Except   for   liver   cancer,   only   cases   with   primary   cancers   and   associated   surgical   procedure(s)   over   the   9-­‐year   period   (2004-­‐2012)   were   included   in   this   analysis   (See   Appendix   2   for   detailed   inclusion   criteria).   Surgeon   volume,   surgeon   specialty,   and   hospital   volume   were   used   as   proxy   measures   to   predict   the   impacts   of   regionalization   of   these   complex   high-­‐risk   cancer   surgeries   to   higher   volume   centres.   Mortality   and   length   of   hospital   stay   were   selected   as   outcomes   of   interest   due   to   the   suitability   of   the   datasets   as   well   as   the   fact   that   these   two   outcomes   are   considered   highly   important  to  both  patients  and  health  care  planners.  Multivariate  regression  and  survival   analyses   were   performed   to   assess   the   adjusted   risk   of   mortality   and   length   of   stay   associated  with  patient  and  provider  specific  variables  of  interest.  Geographic  Information   Systems   (GIS)   spatial   mapping   techniques   were   employed   to   investigate   how   regionalization  of  surgical  cancer  services  has  evolved  over  the  study  period  and  how  this   impacts  patient  travel  times  to  access  surgical  care  (Appendix  3).     Lastly,   age-­‐standardized   resection   rates   were   used   to   determine   the   differences   between   provinces  in  proportion  of  resected  tumours.  The  decision  to  undertake  surgical  approach   depends   on   cancer-­‐specific   factors   (i.e.   stage,   extent   of   tumour   spread),   patient-­‐related   factors  (age,  comorbidities,  or  choice  of  treatment);  however,  studies  have  shown  resection   rate   to   be   associated   with   provider’s   decision   to   adopt   surgical   treatment   even   in   high-­‐risk   cases.162,163   It   has   been   speculated   that   any   increase   in   resection   rates   could   lead   to   improved   survival.166   Since   improvements   in   resection   rates   represent   an   opportunity   to   improve   survival,   we   present   the   differences   in   resection   rates   across   provinces   for   the   cancer  surgeries  of  interest.  If  higher  resection  rates  do  affect  survival,  any  inter-­‐provincial   differences   in   the   proportion   of   resected   cancer   tumours   could   potentially   indicate   P a g e    41     disparities   in   the   equitable   access   to   high   quality   care.   The   age-­‐standardized   resection   rates   were   calculated   using   the   1991   Canadian   population   structure   as   a   reference   population.   Provincial   population   distributions   in   2005   and   2011   were   standardized   and   multiplied   by   3   to   obtain   the   denominator   as   person-­‐years,   as   opposed   to   the   annual   rate(s).   The   numerator   for   resection   rate   calculation   was   the   sum   of   surgeries   taking   place   over  in  the  first  and  last  3  years  of  the  study  cohort.  The  age-­‐standardized  incidence  rates   for   esophageal,   pancreatic,   and   lung   cancers   in   2010   are   reported   for   comparison   purposes.  The  incidence  rates  for  ovarian  and  liver  cancers  could  not  be  reported  as  all  the   diagnosis   codes   for   ovarian   cancer(s)   used   in   this   study   and   the   incidence   of   secondary   liver  tumours  were  not  available  from  Statistics  Canada.             4.2  ANALYSIS  RESULTS   Overall,   differences   were   observed   in   the   total   number   of   surgeries   performed   for   each   cancer  (Table  4.1).  Moreover,  inter-­‐provincial  differences  in  the  proportion  of  surgeries  for   a   particular   cancer   type   were   also   apparent,   related   mainly   to   the   variance   in   population   density   across   Canada.   A   majority   (49-­‐52%)   of   procedures   for   all   five   cancers   were   performed  in  Ontario,  with  British  Columbia  and  Alberta  constituting  the  second  and  third   highest   proportions,   respectively.   This   section   will   summarize   the   state   of   surgical   care   for   each  cancer  type  in  Canada,  and  the  results  of  volume-­‐outcome  association.  See  Appendix  4   and  5  for  detailed  statistical  analyses.   TABLE  4.1:  TOTAL  NUMBER  OF  RESECTIONS  IN  CANADA  (2004-­‐2012)   Province   Alberta  (AB)   British  Columbia   (BC)   Manitoba  (MB)   New  Brunswick  (NB)   Newfoundland  (NL)   Nova  Scotia  (NS)   Ontario  (ON)   Prince  Edward  Island   (PEI)   Esophagus   (%)   311                (11.1)   508                (18.0)   114                  (4.0)   84   (3.0)   29   (1.1)   171                  (6.1)   1,472               (52.3)   *                   Lung   (%)   2,946   (9.7)   4,691     (15.5)   1,986                  (6.5)   1,518                  (5)   455   (1.5)   1,734                   (5.7)   15,732               (52)   *     Liver   Ovarian   Pancreas   Total   (%)   (%)   (%)   1,241               2,182   648                 7328   (14.9)   (12.9)   (16.2)   1,516               3,051   674   10440   (18.2)   (18)   (17)   345               895   164   3504      (4.1)   (5.2)   (  4.1)   59                 526   85   2272    (0.71)   (3.1)   (2.1)   155   362   80   1080   (1.8)   (2.1)   (2.0)   505                 731   181   3322    (6.0)   (4.3)   (4.5  )   4,315               8,340   2,053         31912   (51.7)   (49.2)    (51.5)   *                   62   -­‐-­‐   69     (0.4)   -­‐-­‐   P a g e    42     Saskatchewan  (SK)   119                  (4.3)   2,810   Total   1,218                 (4.02)   30,284   194                  (2.3)   8,333   800   (4.7)   16,949   105   (2.6)   3,990   2436   62366   *numbers  suppressed  (<5  surgeries)   4.2.1  ESOPHAGEAL  CANCER  SURGERY   Esophageal   resections   had   the   smallest   annual   case   volume   among   the   five   cancers   of   interest.   A   total   of   2810   relevant   esophageal   cancer   resections   were   performed   in   71   hospitals   across   Canada   (2004-­‐2012).   The   number   of   institutions   performing   esophageal   resections   declined   over   time,   mainly   led   by   a   50%   reduction   across   Ontario   (Figure   4.2.1a).   However,   the   annual   number   of   esophageal   resections   stayed   relatively   stable   during  this  time  period,  increasing  only  6.4%  (Table  4.2.1a).  Figure  4.2.1b  depicts  annual   provincial  volumes  and  changes  in  the  number  of  procedures  over  time,  with  case  volumes   only   minimally   increasing   over   the   study   period   and   differing   only   slightly   between   the   provinces.   Due   to   the   small   annual   case   load,   the   median   hospital   volume   was   12   esophagectomies   per   year   (Interquartile   range   [IQR]=13),   and   median   annual   surgeon   volume   was   5   esophagectomies   [IQR=6].   Although   the   proportion   of   specialist   surgeons   varied  between  provinces  and  over  the  years,  thoracic  surgeons  performed  nearly  2/3rd  of   all   esophagectomies   across   Canada   between   2004   and   2012.   The   mean   patient   age   was   64.1  (Standard  Deviation  [SD]=10.2)  years,  with  a  majority  (79%)  being  male.         TABLE  4.2.1A:  TOTAL  NUMBER  OF  RESECTIONS  FOR  ESOPHAGEAL  CANCER  IN  CANADA  (2004-­‐ 2012)     AB   BC   MB   NB   NL   NS   ON   PEI   SK   Canada   2004   33   51   14   *   *   21   173   *         11   314   2005   27   48   13   9   *   20   169   *         13   301   2006   22   44   14   8   *   18   160   *         12   283   2007   27   59   13   7   *   25   157   *         13   304   2008   37   51   15   10   *   24   145      *         12   294   *  Number  suppressed  if  number  of  surgeries  <5       2009   34   63   9   14   *   15   172   *         13   323   2010   37   64   9   11   *   12   150   *         16   303   2011   45   62   11   7   *   22   185   *         17   354   2012   49   66   16   13   *   14   161   *         12   334   Total   311   508   114   84   29   171   1472   *         119   2810   P a g e    43     200   Number  of  Surgeries   180   160   140   120   100   80   60   40   20   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Alberta   Atlantic   British  Columbia   Ontario     Prairies     Figure  4.2.1a:  Number  of  esophageal  cancer  surgeries,  by  province  (2004-­‐12)       Number  of  Hospitals   35   30   25   20   15   10   5   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Alberta   Atlantic   British  Columbia   Ontario   Prairies     Figure  4.2.1b:  Number  of  institutions  performing  esophageal  cancer  surgeries,  by  province  (2004-­‐12)     Travel  Distance     The   median   number   of   kilometers   travelled   by   the   patients   to   the   surgical   centres   increased   over   time,   with   a   reduction   in   the   number   of   facilities   providing   esophageal   resections.   The   difference   in   median   travel   distances   from   2004-­‐06   to   2010-­‐12   is   presented   in   Table   4.2.1b.   A   geographic   representation   of   changes   in   distribution   and   number  of  institutions  between  these  two  time  periods  is  presented  in  Figures  4.2.1c  and   4.2.1d.     P a g e    44     The   magnitude   of   change   in   additional   travel   distances   varied   by   province;   with   Newfoundland   and   Saskatchewan   showing   the   highest   increase.   However,   the   smaller   sample  size  in  Newfoundland  may  jeopardize  the  validity  of  these  results.  On  the  contrary,   a  reduction  in  median  kilometers  travelled  by  the  patients  was  observed  in  Manitoba  and   British   Columbia   over   the   two   time   periods   (24   and   5   kilometers,   respectively).   As   expected,   the   provinces   with   the   highest   increase   in   median   number   of   kilometers   also   depicted   a   reduction   in   proportion   of   patients   within   one   hour   travel   distance   (Table   4.2.1b).  In  Ontario  an  additional  6  (median)  kilometers  were  added  to  the  patient  journey   as  a  result  of  a  50%  reduction  in  number  of  facilities.  Between  fiscal  years  2010-­‐12,  61%  of   patients   were   still   within   a   one   hour   travel   distance   from   the   surgical   centre,   down   3%   from  2004-­‐06.             It   should   be   noted   that   the   provinces   differ   in   population   densities,   hence,   the   relative   proportions  of  populations  within  1  hour  of  travel  distance  is  expected  to  vary.  The  relative   differences  across  different  disease  sites  within  the  same  province  might  serve  as  a  better   indicator   of   disparities   in   structures   of   care   for   difference   surgical   procedures.   Intra-­‐ provincial   differences   in   patient   travel   times   for   different   surgeries   are   presented   in   Chapter  5.       TABLE  4.2.1B:  CHANGE  IN  MEDIAN  TRAVEL  DISTANCES  FROM  2004-­‐06  TO  2010-­‐12   Province     2004-­‐06   2010-­‐12   Median  Travel   Median  Travel   %  in  1  hour   Distance  in   Distance  in   travel  time   kilometers  [IQR]   kilometers  [IQR]   AB   BC   MB   NB   NL   NS   ON   SK   15  [81]   33  [95]   38  [68]   19  [54]   8  [111]   63  [160]   25  [81]   74  [221]   66   50   54   68   55   46   64   44   47  [160]   28  [88]   14  [104]   20  [41]   146  [442]   93  [282]   31  [79]   162  [229]   %  in  1  hour   travel  time   49   57   53   65   33   44   61   29   Change  from   2010-­‐12  to   2004-­‐06   Difference  in   Median  Travel   Distances   (kilometers)   +32   -­‐5   -­‐24   +1   +137   +29   +6   +87   P a g e    45     Figure  4.2.1c:  Patient  travel  times  and  pan-­‐Canadian  distribution  of  hospitals  performing  esophageal  cancer  surgeries,  2004-­‐06     P a g e    46     Figure  4.2.1d:  Patient  travel  times  and  pan-­‐Canadian  distribution  of  hospitals  performing  esophageal  cancer  surgeries,  2010-­‐12     P a g e    47     Resection  Rate     Figure   4.2.1e   depicts   resection   rates   over   time   in   comparison   with   the   age-­‐adjusted   incidence   rates.   Overall,   the   minimal   variations   in   age-­‐standardized   resection   rates   were   observed   between   provinces   in   2010-­‐12   fiscal   years.   The   rates   varied   from   3.9   (per   100,000)   in   Saskatchewan   to   1.7   (per   100,000)   in   Newfoundland   in   the   2010-­‐12   fiscal   years.  Apart  from  Manitoba  (2.7  per  100,000),  the  resection  rates  for  other  provinces  were   higher   than   3   resections   per   capita.   However,   the   rates   varied   over   the   year   within   the   provinces.   For   instance,   an   increase   in   resection   rates   occurred   in   Alberta,   British   Columbia,  New  Brunswick,  and  Saskatchewan  between  the  first  three-­‐year  interval  (2004-­‐ 2006)  and  the  later  interval  in  2010-­‐2012.  The  rates  of  esophageal  resection  stayed  stable   in   Ontario   over   the   two   time   periods.   Newfoundland   had   the   lowest   rates   of   2   resections   per   capita   among   all   the   provinces.   The   provincial   resection   rates   observed   over   time   do   not  depict  a  clear  association  with  relative  9-­‐year  mortality  rates  reported  in  Figure  4.2.1g.   However,   the   resections   rates   were   lower   than   the   incidence   rates   for   Manitoba,   New   Brunswick,   Nova   Scotia,   and   Ontario.   The   implications   of   such   differences   on   survival   rates   and  quality  of  care  need  further  investigation.    Rate  (/100,000)   6   5   4   3   2   1   0   AB   BC   MB   Resection  Rate  (2010-­‐12)   NB   NL   NS   Resection  Rate  (2004-­‐06)   ON   SK   Incidence  Rate  (2010)   Figure  4.2.1e:  Age-­‐standardized  resection  rates  per  100,000  populations  for  esophageal  cancer   surgeries   In-­‐Hospital  Mortality     The   overall   in-­‐hospital   mortality   for   esophagectomy   cases   over   the   9-­‐year   period   was   5.6%,  declining  from  6.3%  in  2004  to  3.3%  in  2012  (Figure  4.2.1f).  As  expected,  there  were   differences  between  provinces  in  the  9-­‐year  average  mortality  rate,  even  after  adjusting  for   age   differences   among   the   provinces   (Figure   4.2.1g).   Manitoba,   New   Brunswick,   Ontario,   and   Saskatchewan   had   9-­‐year   mortality   rates   above   the   national   average,   with   other     P a g e    48     provinces   such   as   Newfoundland   markedly   lower.   It   must   be   noted   that   these   age-­‐adjusted   mortality   rates   do   not   consider   the   number   of   patients   who   opt-­‐out   of   surgery,   are   more   highly   staged   and   thus   ineligible   for   resection,   or   experience   a   greater   number   of   co-­‐ morbidities.  More  significantly,  the  rates  cannot  consider  the  effect  of  even  a  small  number   of  deaths  in  a  small  sample,  as  would  be  the  case  with  Newfoundland.  These  factors  may   influence   the   reported   mortality   rates   in   esophageal   cancer,   but   also   in   the   other   cancer   sites  investigated  in  this  report.         9   8   Mortality  Rate  (%)   7   6   5   4   3   2   1   0   2003   2004   2005   2006   2007   2008   2009   2010   2011   2012   2013   Fiscal  Year     Figure  4.2.1f:  Annual  age-­‐adjusted  mortality  rates  for  esophageal  cancer  surgeries  (2004-­‐12)   14   11%   Mortality  Rate  (%)   12   10   6.61%   8   6.27%   6.41%   6   4   3.95%   4.10%   2.72%   2.75%   2   0   AB   BC   MB   NB   NL   Province     NS   ON   SK     Figure  4.2.1g:  Age-­‐adjusted  mortality  rates  for  esophageal  cancer  surgeries  (2004-­‐12)       P a g e    49     60 40 0 20 Mortality Rate (%) 80 100 In   univariate   analysis,   hospital   volume,   surgeon   volume   and   specialty   were   all   associated   with   differences  in  the  in-­‐hospital  mortality  rate.  Figure  4.2.1h,  demonstrates  a  linear  reduction  in   observed  mortality  rates  with  an  increase  in  annual  hospital  volume.  In  multivariate  analysis,   hospital   volume   was   significantly   associated   with   reduction   is   risk   of   mortality,   after   controlling   for   the   effect   of   patient-­‐related   covariates,   surgeon   volume,   specialty,   and   significant  interactions  between  provider  volume  and  specialty  (see  Appendix  4).  A  significant   effect   of   hospital   volume   on   risk   of   mortality   persisted   after   controlling   for   provincial   variations  in  the  model.  Among  the  non-­‐modifiable  factors,  patient  age  and  co-­‐morbidity  score   significantly   predicted   the   risk   of   in-­‐hospital   mortality.   Procedural   type   (open   vs.   minimally   invasive   resection),   and   patient   gender   did   not   significantly   predict   the   risk   of   in-­‐hospital   mortality.       Surgeon   volume   or   specialty   was   insignificantly   associated   with   mortality   risk   after   controlling   for   hospital   volume   and   random   differences   between   hospitals.   A   strong   interaction   between   provider   volume   and   surgeon   specialty   was   also   observed,   further   highlighting   the   complexity   of   volume-­‐outcome   association.   It   also   supports   the   premise   that   improved   outcomes   at   high   volume   hospitals   could   be   the   result   of   a   combination   of   factors   including   hospital   volume,   surgeon   volume   and   surgeon   specialty.   Factors   including   staff   training   and   access   to   advanced   treatment   measures   could   not   be   accounted   for   in   present   data.  There  were  no  differences  in  risk  of  mortality  between  Ontario  and  rest  of  Canada  after   controlling  for  other  factors  in  the  model.     0 10 20 Annual Hospital Volume 30 40   Figure  4.2.1h:  Annual  hospital  volume  and  mortality  rate  for  esophageal  cancer  surgeries  (2004-­‐12)       P a g e    50     ! The  impact  of  hospital  volume  on  the  risk  of  in-­‐hospital  mortality  was  most   pronounced  for  esophageal  cancer.  In  a  simplistic  model,  every  10  case  increase   in  hospital  volume  predicted  a  risk  reduction  of  in-­‐hospital  mortality  of  21%.     ! Assuming  that  the  quality  of  care  and  outcomes  are  the  same  across  all  the   hospitals  in  the  highest  volume  tercile,  theoretically  a  total  of  30  additional  lives   could  have  been  saved  after  esophageal  cancer  surgeries  in  Canada  (2004-­‐12).     Length  of  Stay     TABLE  4.2.1C:  MEDIAN  LENGTH  OF   The   overall   median   length   of   stay   for   esophageal   STAY       surgeries   over   the   9-­‐year   period   was   13   days   Province   Median  length  of   [IQR=12].  In  spite  of  slight  variations  over  the  years,   stay  [IQR]   the   median   length   of   stay   has   been   reduced   by   one   AB   15  [10]   day   across   Canada   from   2004   to   2012   (Table   4.2.1b).   BC   12  [11]   Similarly,   provincial   differences   were   apparent   as   MB   14  [11]   well,  with  New  Brunswick  and  Saskatchewan  having   NB   16  [20]   NL   14  [3]   the   highest   median   length   of   stay   at   16   days   and   NS   13   [10]   British   Columbia   with   the   lowest   at   12   days   (Table   ON   13  [12]   4.2.1c).  Without  accounting  for  any  other  differences   SK   16  [13]   between   provinces   affecting   length   of   stay,   this   difference  implies  a  costly  additional  476  hospital  days  for  Saskatchewan,  and  336  for  New   Brunswick   relative   to   British   Columbia   over   this   time   period.   While   differences   in   discharge   rate   at   different   time   period   are   evident   for   open   procedures,   the   relative   differences   between   low   and   high   volume   hospitals   at   each   period   appear   constant   (Figure   4.2.1i).   TABLE  4.2.1B:  MEDIAN  LENGTH  OF  STAY  IN  CANADA,  BY  YEAR     Median  length  of  stay   [IQR]   2004   13   [14]   Fiscal  Year   2005   2006   2007   2008   2009   2010   2011   2012   15   14   14   13   13   13   13   12   [13]   [12]   [13]   [12]   [12]   [12]   [10]   [10]     In  survival  analysis,  patient  age,  co-­‐morbidities,  open  procedures,  and  male  patients  were   significantly  associated  with  a  risk  of  longer  length  of  stay.  The  survival  model  is  presented   in   Appendix   5.   Surgeon   specialty   was   not   significantly   associated   with   length   of   stay.   Higher  surgeon  volume  was  significantly  associated  with  a  shorter  length  of  stay  for  those   discharged   alive,   with   this   association   marginally   significant   for   hospital   volume.   There   were  also  no  differences  between  Ontario  and  rest  of  Canada  on  length  of  hospital  stay.       P a g e    51     ! Assuming   that   the   quality   of   care   and   outcomes   are   the   same   across   all   the   hospitals   in   the   highest   volume   tercile,   a   median   of   407   hospital   days   could   potentially   have   been   saved   annually   across   Canada   for   esophageal   cancer   surgeries  (2004-­‐12).   1 .9 Discharge Rate (%) .8 .7 .6 .5 .4 .3 .2 .1 0 0 5 10 10 Days 15 20 Annual Hospital Volume 13 Days 25 30 35 21 Days   Figure  4.2.1i:  Predicted  discharge  rate  for  open  esophagectomy  in  Canada  (2004-­‐12)   4.2.2  PANCREATIC  CANCER  SURGERY   In   the   period   spanning   2004   to   2012,   3990   relevant   pancreatic   resections   were   performed   in   92   hospitals   across   Canada.   The   mean   age   of   patients   was   63.7   (SD=11.6)   years,   with   51%   of   patients   being   male.   The   median   hospital   volume   was   23   resections   per   year   [IQR=30],   and   median   surgeon   volume   was   8   resections   annually   [IQR=8].   The   annual   number  of  pancreatic  cancer  surgeries  increased  77%,  from  337  procedures  in  2004  to  599   in   2012   (Figure   4.2.2a).   An   increase   in   number   of   procedures   was   observed   in   Ontario,   Alberta,   Manitoba,   and   British   Columbia   (Table   4.2.2a).   At   the   same   time,   the   number   of   institutions   performing   surgeries   decreased   from   56   in   2004   to   39   in   2012   (Figure   4.2.2b).   It   is   interesting   to   note   that   a   50%   reduction   in   the   number   of   institutions   performing   pancreatic   resections   was   observed   in   Ontario   and   New   Brunswick   (26   to   13;   and   6   to   3   respectively],   potentially   demonstrating   active   regionalization   efforts   in   Ontario   and   the   apparent  consolidation  of  surgical  cancer  services  in  New  Brunswick.               P a g e    52     TABLE  4.2.2A:  NUMBER  OF  PANCREATIC  CANCER  SURGERIES  IN  CANADA,  BY  YEAR  AND   PROVINCE  (2004-­‐12)     AB   BC   MB   NB   NL   NS   ON   SK   Canada   2004   47   60   13   16   14   21   155   11   337   2005   60   76   7   14   6   13   209   11   396   2006   56   73   16   7   7   28   200   13   400   2007   45   87   16   11   *   14   213   10   401   2008   79   76   13   *   6   24   199   17   416   2009   73   72   19   9   14   25   238   6   456   2010   85   55   33   6   7   22   235   9   452   2011   92   83   16   11   8   18   288   17   533   2012   111   92   31   9   13   16   316   11   599   *Number  suppressed  if  number  of  surgeries  <5       Number  of  Surgeries   350   300   250   200   150   100   50   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Alberta   Atlantic     British  Columbia   Ontario   Prairies     Figure  4.2.2a:  Number  of  pancreatic  cancer  surgeries,  by  province  (2004-­‐12)       Total   648   674   164   85   80   181   2053   105   3990   P a g e    53     Number  of  Hospitals   30   25   20   15   10   5   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Alberta   Atlantic   British  Columbia   Ontario   Prairies     Figure  4.2.2b:  Number  of  institutions  performing  pancreatic  cancer  surgeries,  by  province  (2004-­‐12)     The  outcomes  of  surgery  also  depend  on  complexity  of  surgical  approach.  Among  the  three   surgical   approaches,   Whipple   procedures   (pancreaticoduodenectomy)   were   performed   in   the  majority  of  cases  (72%),  with  open  distal  pancreatectomy  accounting  for  a  remaining   25%   of   procedures.   A   minimally   invasive   approach   was   adopted   for   only   2.8%   of   the   procedures.     Travel  Distance     Overall,   a   reduction   in   the   proportion   of   population   within   one   hour   travel   distance   of   pancreatic  surgical  cancer  care  was  noticed  for  all  provinces,  albeit  of  different  magnitudes.   Median  travel  distances  increased  in  all  provinces  when  comparisons  were  made  between   first   and   last   three   fiscal   years   (Table   4.2.2b).   The   greatest   increase   was   found   in   Newfoundland,   with   median   of   193   extra   kilometers   travelled   by   the   patients   to   receive   pancreatic   cancer   surgery.   In   Ontario,   the   50%   reduction   in   number   of   institutions   performing  pancreatic  surgeries  resulted  in  a  median  of  5  additional  kilometers,  associated   with  3%  lower  proportion  of  population  within  1  hour  travel  distance.  The  reduction  in  the   number   of   facilities   in   New   Brunswick   affected   the   median   travel   distance,   but   not   the   proportion  of  population  within  one  hour  of  travel  time.  The  remaining  provinces  showed   changes   in   patients   traveling   more   than   an   hour   without   any   major   changes   in   structure   of   care   i.e.   facility   distribution.   Figures   4.2.2c   and   4.2.2d   can   be   used   to   visualize   the   geographic   distribution   of   facilities   providing   pancreatic   cancer   surgical   care   in   2004-­‐06   and  2010-­‐12.         P a g e    54     TABLE  4.2.2B:  CHANGE  IN  MEDIAN  TRAVEL  DISTANCES  FROM  2004-­‐06  TO  2010-­‐12   Province     AB   BC   MB   NB   NL   NS   ON   SK   2004-­‐06   Median  Travel   Distance  in   kilometers  [IQR]   17  [128]   17  [59]   10  [104]   60  [105]   79  [400]   43  [164]   27  [73]   118  [220]   2010-­‐12   %  in  1  hour   travel  time   64   60   64   35   44   47   64   43   Median  Travel   Distance  in   kilometers  [IQR]   22  [165]   28  [59]   16  [94]   102  [146]   272  [405]   115  [292]   32  [76]   155  [204]   %  in  1  hour   travel  time   57   57   59   35   25   36   61   35   Change  from   2010-­‐12  to   2004-­‐06   Difference  in   Median  Travel   Distances  (km)   +5   +11   +6   +42   +194   +72   +5   +37     Resection  Rate   Overall,   a   majority   of   the   provinces   reported   an   increase   in   adjusted   resection   rates   for   pancreatic   cancer   over   time,   except   from   New   Brunswick   and   Newfoundland.   Figure   4.2.2e   shows   that   Alberta   had   the   highest   resection   rate   for   pancreatic   cancer   surgeries,   increasing   from   4.5   to   7.8   resections   per   capita   over   the   two   time   periods.   Manitoba   and   Ontario   also   saw   an   increase   in   resection   rates   over   time,   2.9   and   1.9   (per,   100,000)   respectively.   There   was   substantial   variation   between   the   highest   and   lowest   resection   rates   between   provinces   in   2010-­‐12.   In   quantitative   terms,   the   rate   varied   from   3.4   (per   100,000)   in   Newfoundland   to   7.8   (per   100,000)   in   Alberta.   Although   the   relative   differences  varied,  the  incidence  rates  for  pancreatic  cancer  were  higher  than  the  observed   resection  rates  for  all  the  provinces.     12   Rate  (/100,000)   10   8   6   4   2   0   AB   BC   Resection  Rate  (2010-­‐12)   MB   NB   NL   NS   Resection  Rate  (2004-­‐06)   ON   SK   Incidence  Rate  (2010)   Figure  4.2.2e:  Age-­‐standardized  resection  rates  per  100,000  population  for  pancreatic  cancer   surgery     P a g e    55     Figure  4.2.2c:  Patient  travel  times  and  pan-­‐Canadian  distribution  of  hospitals  performing  pancreatic  cancer  surgeries,  2004-­‐06     P a g e    56     Figure  4.2.2d:  Patient  travel  times  and  pan-­‐Canadian  distribution  of  hospitals  performing  pancreatic  cancer  surgeries,  2010-­‐12     P a g e    57     In-­‐Hospital  Mortality     The   overall   in-­‐hospital   mortality   rate   after   pancreatic   resection   for   malignant   indications   over   a   9-­‐year   period   was   3.7%.   Marked   variations   in   annual   mortality   rates   were   observed   over  time,  with  the  lowest  annual  mortality  rate  in  Canada  for  pancreatic  cancer  surgeries   found   to   be   2%   in   2011   and   6%   in   2005   (Figure   4.2.2f).   In   addition   to   variations   in   mortality  rate  observed  over  time,  provincial  differences  in  9-­‐year  average  mortality  rates   were   also   apparent.   Alberta,   Manitoba,   Nova   Scotia,   and   Saskatchewan   had   rates   higher   than  the  national  9-­‐year  average  (Figure  4.2.2g).         7     6   Mortality  Rate  (%)     5   4   3   2     1   0   2003   2004   2005   2006   2007   2008   2009   2010   2011   2012   2013   Fiscal  Year     Figure  4.2.2f  Annual  age-­‐adjusted  mortality  rates  for  pancreatic  cancer  surgeries  (2004-­‐12)   8   Mortality  Rate  (%)   6   5   5.75%   6.36%   7   4.63%   4.28%   3.46%   3.22%   4   2.13%   3   2   1.29%   1   0   AB   BC   MB   NB   NL   Province   NS   ON   SK     Figure  4.2.2g:  Age-­‐adjusted  mortality  rates  for  pancreatic  cancer  surgeries  (2004-­‐12)     P a g e    58     60 40 0 20 Mortality Rate (%) 80 100 According   to   the   multivariate   regression   model,   the   risk   of   in-­‐hospital   mortality   for   pancreatic   cancer   resections   increased   with   increasing   age   and   co-­‐morbidity   score   (Appendix  4).  Gender  and  year  did  not  significantly  predict  the  risk  of  in-­‐hospital  mortality.   A   Whipple   procedure   was   associated   with   a   significantly   higher   risk   when   compared   to   distal  pancreatectomy.       In   univariate   analysis,   hospital   volume   was   significantly   associated   with   the   rate   of   mortality  (p=0.003),  but  not  surgeon  volume.  Figure  4.2.2h  represents  a  simplistic  view  of   the   association   between   increasing   annual   hospital   volume   and   mortality   rate   for   pancreatic   cancer   resections   in   Canada.   A   steeper   reduction   in   mortality   rate   is   apparent   with   an   initial   increase   in   volume,   with   the   trend   stabilizing   around   40   procedures   per   year.  Hospital  volume  was  a  significant  predictor  of  mortality  in  the  multivariate  regression   model,   after   controlling   for   random   hospital   effect   and   provincial   variations.   Surgeon   volume   was   only   significantly   associated   with   mortality   risk   without   hospital   related   factors   in   the   model.   There   were   no   significant   differences   in   risk   of   mortality   between   Ontario  and  rest  of  Canada.   0 20 40 60 Annual Hospital Volume 80 100   Figure  4.2.2h:  Annual  hospital  volume  and  mortality  rate  (%)  for  pancreatic  cancer  surgeries     (2004-­‐12)   ! According   to   the   model,   an   annual   increase   in   10   cases   in   a   given   hospital   predicted   a   22%   decrease   in   the   risk   of   in-­‐hospital   mortality   after   controlling   for   the   influence   of   covariates   and   random   clustering   effects   of   individual   hospitals.     ! Assuming   that   the   quality   of   care   and   outcomes   are   the   same   across   all   the   hospitals  in  the  highest  volume  tercile,  theoretically  a  total  of  89  additional  lives   could  have  been  saved  after  pancreatic  cancer  surgeries  in  Canada  (2004-­‐12).     P a g e    59     Length  of  Stay     TABLE  4.2.2C:  MEDIAN  LENGTH  OF   The   median   length   of   stay   after   a   pancreatic   STAY       resection   in   Canada   during   2004   to   2012   was   11   Province   Median  Length  of   days   [IQR=9].   Over   the   course   of   nine   years,   the   Stay   median  length  of  stay  decreased  from  13  to  12  days    [IQR]   in   Canada   (Table   4.2.2b).   The   provincial   median   AB   12  [10]   hospital   days   ranged   from   10   in   Ontario   to   15   in   BC   13  [10]   Saskatchewan   with   a   majority   of   provinces   having   MB   14  [11]   NB   14  [11]   values   between   13   and   14   days   (Table   4.2.2c).   An   NL   13  [16]   increase   in   discharge   rate   with   increasing   volume   for   NS   13  [14]   both   Whipple   procedure   and   distal   pancreatectomy   ON   10  [8]   resection  types  at  median  length  of  stay  is  depicted  in   SK   15  [16]   Figure   4.4.2i.   Figure   4.2.2j   a)   and   b)   demonstrate   differences  in  predicted  discharge  rates  between  procedures  at  different  points  in  time.     Survival   analysis   confirmed   a   statistically   significant   association   between   patient   age   and   co-­‐morbidity   score   with   a   longer   length   of   stay   (Appendix   5).   There   were   no   statistically   significant  differences  in  length  of  stay  between  males  and  females.  As  expected,  Whipple   procedures,  being  more  complex,  had  longer  length  of  hospital  stay  as  compared  to  distal   pancreatectomy.   Furthermore,   the   requirement   of   additional   organ   resections   (i.e.   gall   bladder,  bowel  resections)  also  predicted  a  higher  number  hospitalized  days.  The  year-­‐to-­‐ year   reductions   in   length   of   stay   (Table   4.2.2b)   were   also   confirmed   after   controlling   for   the   effect   of   other   covariates.   Both   hospital   and   surgeon   volumes   were   significantly   associated  with  shorter  length  of  stay.  Moreover,  patients  in  Ontario  also  had  higher  chance   of  early  discharge  as  compared  to  rest  of  Canada.     TABLE  4.2.2B:  MEDIAN  LENGTH  OF  STAY  IN  CANADA,  BY  YEAR     Median  Length  of  Stay   [IQR]     Fiscal  Year   2004   2005   2006   2007   2008   2009   2010   2011   2012   13   13   12   11   12   11   11   10   10   [10]   [11]   [11]   [11]   [8]   [10]   [8]   [8]   [9]   P a g e    60     1 .9 .8 Discharge Rate (%) .7 .6 .5 .4 .3 .2 .1 0 0 25 50 Annual Hospital Volume Distal Pancreatectomy 75 100 Whipple Procedure 1 1 .9 .9 .8 .8 Discharge Rate (%) Discharge Rate (%)   Figure  4.2.2i  Predicted  discharge  rate  11  days  after  pancreatic  cancer  procedures  (2004-­‐12)   .7 .6 .5 .4 .3 .7 .6 .5 .4 .3 .2 .2 .1 .1 0 0 0 25 50 Annual Hospital Volume 7 Days 14 Days 75 11 Days 21 Days 100 0 25 50 Annual Hospital Volume 7 Days 14 Days 75 100 11 Days 21 Days   Figure  4.2.2j.  Predicted  discharge  rate  after  a.  Whipple  Procedure  b.  Open  Distal  Pancreatectomy   (2004-­‐12)   ! Assuming   that   the   quality   of   care   and   outcomes   are   the   same   across   all   the   hospitals   in   the   highest   volume   tercile,   a   median   of   725   hospital   days   could   potentially   have   been   saved   annually   across   Canada   for   pancreatic   cancer   surgeries.         P a g e    61     4.2.3  LIVER  CANCER  SURGERY   Liver   cancer   had   the   third   largest   surgical   volume   among   the   five   high-­‐intensity   cancer   types   studied.   Overall,   a   total   of   8083   liver   resections   were   performed   on   patients   with   mean  age  of  62  (SD=11.7)  years.  Primary  liver  cancers  constituted  24%  of  all  procedures,   while   the   remaining   procedures   were   performed   for   secondary   liver   tumours.   A   total   of   118   institutions   performed   liver   surgeries   over   the   9-­‐year   period   across   Canada.   As   the   annual   number   of   surgeries   increased   82%,   the   number   of   hospitals   performing   these   resections  decreased  from  61  in  2004  to  42  in  2012.  An  increase  in  number  of  procedures   was   observed   in   all   provinces   except   New  Brunswick   and   Saskatchewan   (Table   4.2.3a,   and   Figure  4.2.3a).  Overall,  the  shift  in  the  number  of  facilities  over  time  was  less  pronounced   than   that   observed   for   esophageal   and   pancreatic   cancer   surgeries   (Figure   4.2.3b).   A   systematic  shift  in  the  number  of  facilities  performing  liver  resections  from  2004  to  2012   was   evident   in   Manitoba   (4   facilities   to   1)   and   Ontario   (29   centres   to   20).   Both   of   these   provinces   report   good   mortality   rates   and   length   of   hospital   stay   relative   to   the   pan-­‐ Canadian   mean,   which   appears   to   demonstrate   a   positive   benefit   to   these   regionalization   efforts.   In   the   remaining   provinces,   the   annual   fluctuation   in   the   number   of   facilities   appeared   to   be   a   result   of   institutions   undertaking   a   small   liver   surgical   caseload   (≤2/year).   Among   the   118   institutions   providing   liver   resection   over   9-­‐year   period,   the   median   hospital   volume   was   61   resections   per   year   [IQR=88].   The   number   of   physicians   performing  liver  resections  stayed  relatively  constant  over  the  years  (119  in  2004  to  117  in   2012)  with  a  median  surgeon  volume  of  20  resections  per  year  [IQR=18].     There  are  two  types  of  resections  for  liver  cancers:  open  and  minimally  invasive.  A  majority   (93%)   of   resections   were   open   procedures.   Over   the   years,   a   shift   towards   minimally   invasive  procedures  has  been  observed.  In  2004,  only  1%  of  minimally  invasive  procedures   were   performed,   while   this   number   has   increased   to   about   11%   in   2012   (Figure   4.2.3c).   Minimally   invasive   procedures   had   a   lower   mortality   rate   (1.32%)   as   compared   to   open   procedures  (3.1%)  over  the  study  period.     TABLE  4.2.3A:  NUMBER  OF  LIVER  CANCER  SURGERIES  IN  CANADA,  BY  YEAR  AND  PROVINCE       AB   BC   MB   NB   NL   NS   ON   SK   Canada   2004   116   121   19   12   15   42   347   20   692   2005   109   147   17   7   15   49   387   14   745   2006   94   155   33   5   13   52   418   25   796   *Number  suppressed  if  number  of  surgeries  <5       2007   102   147   28   1   9   41   434   22   784   2008   130   141   44   6   23   48   468   23   883   2009   137   181   39   8   17   70   511   18   981   2010   153   175   43   7   13   68   505   21   985   2011   176   228   67   7   21   64   607   31   1202   2012   224   221   55   6   29   71   638   20   1265   Total   1241   1516   345   59   155   505   4315   194   8333   P a g e    62     700   Number  of  Surgeries   600   500   400   300   200   100   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Alberta   Atlantic   British  Columbia   Ontario   Prairies   Figure  4.2.3a:  Number  of  liver  cancer  surgeries,  by  province  (2004-­‐12)       Number  of  Hospitals     35   30   25   20   15   10   5   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Alberta       Atlantic   British  Columbia   Ontario   Prairies     Figure  4.2.3b:  Number  of  institutions  performing  liver  cancer  surgeries,  by  province  (2004-­‐12)   P a g e    63     1200   Number  of  Surgeries   1000   800   600   400   11%   200   1%   3%   4%   5.6%   7%   8%   11%   8%   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Minimally  Invasive   Open   Figure  4.2.3c:  Shift  in  liver  cancer  procedures  in  Canada  (2004-­‐12)       Travel  Distances     The  change  in  the  number  of  institutions  performing  liver  resections  was  least  pronounced   for  liver  cancer  over  time  among  the  five  cancer  types  of  interest  (Figure  4.2.3b).  It  is  worth   noticing  that  there  was  at  least  one  institution  in  each  province  carrying  out  the  majority  of   procedures   each   year.   The   fluctuation   in   number   of   institutions   from   year   over   year   was   due   to   institutions   with   <5   cases   annually.   The   shift   in   the   number   and   location   of   institutions,  and  its  impact  on  patient  travel  times  can  be  visualized  through  a  comparison   of  Figures  4.2.3d  and  4.2.3e.     Except   for   Nova   Scotia   and   Saskatchewan,   no   substantial   variations   in   travel   distances   and   proportion   of   population   within   one   hour   travel   distance   were   observed   across   the   provinces.   In   Nova   Scotia,   one   major   centre   performed   liver   resections   over   the   9   year   study   period   with   small   number   of   resections   performed   in   other   5   centres.   Therefore,   the   median  increase  in  travel  distance  in  Nova  Scotia  could  be  a  result  of  random  fluctuations   in   the   location   of   patients   within   the   province.   Similar   reasoning   appears   to   apply   in   the   context   of   Saskatchewan,   with   a   substantial   reduction   in   the   median   travel   distance   over   the   two   time   periods   (114   kilometers).   In   spite   of   reductions   in   the   number   of   facilities   from   5   (average   2004-­‐06)   to   2   in   2010-­‐12,   the   proportion   of   patients   within   one   hour   travel   distance   increased   from   31%   to   50%   in   Saskatchewan.   The   potential   for   bias   in   provinces  with  a  small  patient  population  should  be  kept  in  mind  while  interpreting  these   results.   Ontario   saw   the   biggest   shift   in   number   of   institutions   over   time-­‐the   systematic   shift  could  be  seen  in  data  with  a  sudden  drop  in  volume  in  one  institution  in  a  given  year   followed  by  a  sudden  jump  in  volume  in  another.  These  findings,  paired  with  the  fact  that   median  travel  distance  increased  (3  kilometers)  and  the  proportion  of  patients  within  one     P a g e    64     hour   travel   time   decreased   (5%)   may   hint   towards   possible   impacts   of   regionalization   of   liver  cancer  surgeries.         TABLE  4.2.3B:  CHANGE  IN  MEDIAN  TRAVEL  DISTANCES  FROM  2004-­‐06  TO  2010-­‐12     Province   AB   BC   MB   NB   NL   NS   ON   SK   2004-­‐06   Median  Travel   Distance  in   kilometers  [IQR]   19  [129]   17  [44]   13  [83]   29  [110]   113  [547]   101  [229]   28  [56]   186[247]   Change  from   2010-­‐12  to   2004-­‐06   Difference  in   Median  Travel   Distances  (km)   -­‐1   +2   0   +5   -­‐18   +92   +3   -­‐114   2010-­‐12   %  in  1   hour  travel   time   59   64   62   54   35   41   62   31   Median  Travel   Distance  in   kilometers  [IQR]   18  [144]   20  [60]   13  [85]   33  [124]   95  [286]   193  [280]   31  [72]   72  [216]   %  in  1  hour   travel  time   62   62   64   55   40   31   57   50     Resection  Rate   Similar   to   the   trends   seen   in   pancreatic   cancer   resections,   an   overall   increase   in   age-­‐ adjusted  resection  rates  took  place  across  all  provinces  from  2004-­‐06  to  2010-­‐12  (Figure   4.2.3f).   Alberta   had   the   highest   (14   per  100,000)   while   New   Brunswick   had   the   lowest   (6.9   per   100,000)   resection   rates   in   last   3   years   (2010-­‐2012).   The   highest   improvement   in   resection   rates   for   pancreatic   cancers   was   observed   in   Manitoba,   increasing   from   5.3   to   12   resections   per   100,000   populations.   A   comparison   with   Figure   4.2.3g   points   towards   a   general   trend   that   the   provinces   with   highest   resection   rates   had   among   the   lowest   in-­‐ hospital  mortality  rates.   16   Rate  (/100,000)   14   12   10   8   6   4   2   0   AB   BC   MB   NB   Resection  Rate  (2010-­‐12)   NL   NS   ON   SK   Resection  Rate  (2004-­‐06)   Figure  4.2.3f:  Age-­‐standardized  resection  rates  per  100,000  populations  for  liver  cancer  surgeries     P a g e    65     Figure  4.2.3d:  Patient  travel  times  and  pan-­‐Canadian  distribution  of  hospitals  performing  liver  cancer  surgeries,  2004-­‐06         P a g e    66     Figure  4.2.3e:  Patient  travel  times  and  pan-­‐Canadian  distribution  of  hospitals  performing  liver  cancer  surgeries,  2010-­‐12     P a g e    67     In-­‐Hospital  Mortality     The   overall   in-­‐hospital   mortality   rate   for   liver   resections   over   9-­‐year   period   was   3%,   declining  from  3.9%  in  2004  to  2.6%  in  2012  (Figure  4.2.3g).  The  adjusted  mortality  rates   were   higher   than   the   national   average   for   New   Brunswick,   Newfoundland   and   Saskatchewan,   however,   the   small   number   of   cases   could   explain   this   large   variation,   as   demonstrated  by  the  error  bars  in  Figure  4.2.3h.     5   Mortality  Rate  (%)   4   3   2   1   0   2003   2004   2005   2006   2007   2008   2009   2010   2011   2012   2013   Fiscal  Year     Figure  4.2.3g:  Annual  age-­‐adjusted  mortality  rates  for  liver  cancer  surgeries  (2004-­‐12)   8   4.87%   Mortality  Rate  (%)   7   6   3.50%   4   3   4.35%   4.23%   5   2.63%   2.42%   2.27%   3.14%   2   1   0   AB   BC   MB   NB   NL   Province     NS   ON   SK     Figure  4.2.3h:  Age-­‐adjusted  mortality  rates  for  liver  cancer  surgeries  (2004-­‐12)       P a g e    68     60 40 0 20 Mortality Rate (%) 80 100 In   multivariate   analysis,   age,   co-­‐morbidities,   male   patients,   and   open   procedures   were   significantly  associated  with  a  greater  risk  of  in-­‐hospital  mortality,  whereas  resections  for   secondary   liver   tumours   carried   a   lower   mortality   risk.   Non-­‐significant   differences   in   mortality   rates   between   Ontario   and   rest   of   Canada   were   observed.   The   associated   mortality  risk  for  liver  resections  also  significantly  reduced  over  time.  Figure  4.2.3i  shows  a   trend   towards   lower   mortality   rates   with   increasing   annual   hospital   volume.   These   differences   in   mortality   rates   with   increasing   hospital   volume   were   non-­‐significant   after   adjusting   for   random   hospital   or   provincial   differences.   Surgeon   volume,   however,   did   predict  a  lower  mortality  risk  in  adjusted  models  (Appendix  4).     0 50 100 Annual Hospital Volume 150 200 Figure  4.2.3i:  Annual  hospital  volume  and  mortality  rate  (%)  for  liver  cancer  surgeries  (2004-­‐12)   ! Assuming   that   the   quality   of   care   and   outcomes   are   the   same   across   all   the   hospitals   in   the   highest   volume   tercile,   theoretically   a   total   of   15   additional   lives   could  have  been  saved  after  liver  cancer  surgeries  in  Canada  (2004-­‐12).     Length  of  Stay     The  overall  median  length  of  stay  for  liver  resections   was   7   days   [IQR=5],   with   a   median   reduction   of   one   day   from   2004   to   2012   (Table   4.2.3b).   Saskatchewan   had   the   highest   median   length   of   stay   for   liver   resections   (Table   4.2.3c).   As   observed   for   pancreatic   cancer   resections,   minimally   invasive   procedures   require   shorter   hospitalization   as   compared   to   open   procedures  (Figure  4.2.3j  (a)).  However,  the  predicted   discharge   rate   increased   with   increasing   institutional     TABLE  4.2.3C:  MEDIAN  LENGTH  OF   STAY       Province   AB   BC   MB   NB   NL   NS   ON   SK   Length  of  Stay   (Median  [IQR])   8  [5]   7  [5]   7  [4]   8  [4.5]   7  [4]   7  [4]   7  [4]   9  [7]   P a g e    69     volume  for  both  the  procedures.   In  survival  analysis,  a  significantly  longer  length  of  stay  was  associated  with  patient  age,  co-­‐ morbidities,  and  open  procedures.  The  number  of  expected  hospital  days  was  predicted  to   have  declined  over  the  years.  Ontario  had  a  lower  risk  of  prolonged  length  of  stay,  as  did   resections  for  secondary  tumours.  As  opposed  to  the  trend  seen  in  other  cancer  surgeries,   surgeon   volume   predicted   shorter   hospital   stay,   while   hospital   volume   was   not   a   significant   predictor   of   length   of   hospital   stay.   See   Appendix   5   for   complete   survival   analysis.   TABLE  4.2.3B:  MEDIAN  LENGTH  OF  STAY  IN  CANADA,  BY  YEAR     Median  Length  of  Stay   [IQR]   Fiscal  Year   2004   2005   2006   2007   2008   2009   2010   2011   2012   8   8   7   8   7   7   7   7   7   [6]   [5]   [5]   [5]   [4]   [4]   [4]   [5]   [6]   1 1 .9 .9 .8 .8 Discharge Rate (%) Discharge Rate   .7 .6 .5 .4 .3 .7 .6 .5 .4 .3 .2 .2 .1 .1 0 0 0 0 25 50 75 100 125 Annual Hospital Volume Minimally Invasive Open 150 175 25 50 75 100 125 Annual Hospital Volume 5 Days 10 Days 150 175 7 Days 14 Days   Figure  4.2.3j:  Predicted  discharge  rate  after  liver  cancer  surgeries  a.  At  Median  7  Days,  b.  Open  liver   resections  (2004-­‐12)   ! Assuming   that   the   quality   of   care   and   outcomes   are   the   same   across   all   the   hospitals  in  the  highest  volume  tercile,  a  total  of  308  hospital  days  could  potentially   have  been  saved  annually  across  Canada  after  liver  cancer  surgeries.           P a g e    70     4.2.4  LUNG  CANCER  SURGERY   Lung   cancer   surgical   procedures   had   the   highest   volume   of   cases   over   the   9-­‐year   study   period  in  Canada.  In  total,  30,284  relevant  procedures  were  performed  in  76  institutions.   The   mean   age   of   patients   was   66.7   (SD=10)   years,   with   52%   of   patients   being   male.   The   median   hospital   volume   was   129   surgeries   per   year   [IQR=108],   whereas   the   median   surgeon   volume   was   40   [IQR=27].   The   annual   number   of   lung   surgeries   increased   29%,   from   2942   in   2004   to   3795   in   2012.   An   increase   in   number   of   procedures   was   observed   in   all  provinces,  except  Nova  Scotia  (Table  4.2.4a,  Figure  4.2.4a).  Over  the  same  time  period,   the   number   of   institutions   performing   lung   surgeries   decreased   from   65   to   43   (Figure   4.2.4b).   The   three   known   provinces   that   have   taken   steps   to   actively   regionalize   lung   cancer  surgical  care  are  Ontario,  British  Columbia,  and  Manitoba.  Subsequent  reductions  in   number   of   institutions   performing   these   resections   were   observed   in   Ontario   (37   to   24),   and   British   Columbia   (8   to   4).   Although   a   shift   from   3   to   2   institutions   was   observed   in   Manitoba,  there  was  one  centre  carrying  out  94%  of  lung  cancer  surgeries  in  2012.       Lung   resections   were   classified   into   five   categories   based   on   surgical   complexities   associated  with  each  approach.  In  order  of  increasing  complexity,  these  include:  1)  Video   Assisted   Thoracoscopic   Surgery   minimally   invasive   (VATS)   sub-­‐lobar   resection,   2)   Open   thoracotomy   sub-­‐lobar   resection,   3)   VATS   lobectomy,   4)   Open   lobectomy,   and   5)   Pneumonectomy.    Open  lobectomy  was  the  most  performed  procedure  (35%),  consisting  of   the   removal   of   a   full   lobe   of   the   lung   followed   by   open   sub-­‐lobar   resection   (28%),   which  is   the  removal  of  only  a  portion  of  a  lobe.  Lung  cancer  surgeries  are  normally  performed  by   Thoracic  surgeons,  although  25%  of  lung  surgeries  in  Canada  were  performed  by  general   surgeons.   A   higher   proportion   of   surgeries   completed   by   general   surgeons   is   reported   in   New   Brunswick   52%,   Ontario   (32%)   and   Saskatchewan   (47%).   It   should   be   noted   that   coding  practices  for  surgeon  specialty  are  subject  to  debate,  as  described  in  Appendix  3.   TABLE  4.2.4A:  TOTAL  NUMBER  OF  LUNG  CANCER  SURGERIES  IN  CANADA  (2004-­‐12)     AB   BC   MB   NB   NL   NS   ON   PEI   SK   Canada   2004   241   464   205   122   44   183   1550   *   130   2942   2005   285   482   224   147   49   181   1680   *   150   3199   2006   291   484   188   154   39   207   1737   *   139   3239   *Number  suppressed  if  number  of  surgeries  <5         2007   330   506   212   139   35   184   1837   *   140   3383   2008   340   523   214   189   49   197   1689   *   124   3325   2009   348   583   243   166   61   205   1781   *   129   3516   2010   326   504   228   186   65   206   1694   *   134   3343   2011   382   531   246   202   64   195   1800   *   122   3542   2012   403   614   226   213   49   176   1964   *   150   3795   Total   2946   4691   1986   1518   455   1734   15732   *   1218   30284   P a g e    71       Number  of  Surgeries   2500   2000   1500   1000   500   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Alberta   Atlantic   British  Columbia   Ontario   Prairies     Figure  4.2.4a:  Number  of  lung  cancer  surgeries,  by  province  (2004-­‐12)   40   Number  of  Hospitals   35   30   25   20   15   10   5   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Alberta   Atlantic   British  Columbia   Ontario   Prairies     Figure  4.2.4b:  Number  of  institutions  performing  lung  cancer  surgeries,  by  province  (2004-­‐12)       Travel  Distance     Table   4.2.4b   presents   a   comparison   of   changes   in   median   travel   distances   over   time   across   the  provinces  for  lung  cancer  surgery.  An  interesting  finding  is  that  the  implementation  of   standards  for  regionalization  of  thoracic  cancer  surgeries  in  Ontario  and  accounts  of  such   measures  in  British  Columbia  and  Manitoba  had  mixed  impacts  on  travel  times.  An  increase   in   median   travel   time   and   the   respective   proportion   of   patients   within   1   hour   driving     P a g e    72     distance   to   the   surgical   centre   was   observed   in   Ontario   with   a   move   towards   regionalization.  However,  the  median  travel  time  for  lung  resections  was  reduced  in  British   Columbia   and   Manitoba   between   the   earliest   and   latest   3   year   intervals.   These   findings   are   astounding  given  the  clear  evidence  of  a  shift  to  fewer  facilities.  In  British  Columbia,  there   were  8  facilities  performing  lung  surgeries  in  2004,  but  there  were  only  4  left  in  2012.  In   spite   of   occasional   surgeries   in   other   centres,   the   data   shows   consistent   consolidation   to   these  4  high  volume  centres.  Still,  the  proportion  of  patients  within  1  hour  driving  distance   increased   by   6%,   thus   shortening   the   median   travel   distance   over   time.   Similarly,   a   shift   to   2   facilities   (from   3)   was   evident   in   Manitoba   around   2009-­‐10   without   any   changes   in   proportion   of   population   within   1   hour   commute.   In   concordance   with   the   findings   for   British   Columbia,   the   median   travel   time   decreased   by   5   kilometers   over   time.   While   it   could  not  be  confirmed,  Saskatchewan  did  not  have  a  formal  process  for  regionalization  of   cancer  surgical  care,  yet  a  gradual  reduction  from  3  to  2  facilities  is  obvious  in  the  data.  The   median   travel   time   also   decreased   with   an   increase   in   the   proportion   of   patients   having   driven   less   than   an   hour   for   surgery.   A   similar   move   from   4   facilities   to   3   in   Alberta   resulted   in   minuscule   changes   travel   distance   and   time.   The   geographic   maps   presented   in   Figures   4.2.4c   and   4.2.4d   can   further   assist   in   visualization   of   distribution   of   facilities   across  the  provinces  and  any  changes  in  these  distributions  over  time.     TABLE  4.2.4B:  CHANGE  IN  MEDIAN  TRAVEL  DISTANCES  FROM  2004-­‐06  TO  2010-­‐12   Province     AB   BC   MB   NB   NL   NS   ON   SK     2004-­‐06   2010-­‐12   Median  Travel   %  in  1  hour   Median  Travel   Distance  in   travel  time   Distance  in   kilometers  [IQR]   kilometers  [IQR]   17  [128]   32  [102]   18  [120]   37  [92]   95  [324]   81  [150]   18  [52]   118  [197]   62   52   63   50   40   44   71   38   18  [118]   25  [93]   13  [94]   48  [115]   292  [539]   71  [148]   26  [69]   73  [186]   %  in  1  hour   travel  time   62   58   63   45   30   45   63   46   Change  from   2010-­‐12  to   2004-­‐06   Difference  in   Median  Travel   Distances   (kilometers)   +1   -­‐7   -­‐5   +11   +198   -­‐10   +8   -­‐45   P a g e    73     Figure  4.2.4c:  Patient  travel  times  and  pan-­‐Canadian  distribution  of  hospitals  performing  lung  cancer  surgeries,  2004-­‐06     P a g e    74     Figure  4.2.4d:  Patient  travel  times  and  pan-­‐Canadian  distribution  of  hospitals  performing  lung  cancer  surgeries,  2010-­‐12     P a g e    75     Resection  Rate   Figure   4.2.4e   shows   that   New   Brunswick   had   the   highest   resection   rate   for   lung   cancer   surgeries,   increasing   from   43   to   58   resections   per   capita   over   the   two   time   periods.   Manitoba   and   Nova   Scotia   did   not   show   such   a   sharp   rise   in   resection   rates   over   the   intervals   under   study,   with   rates   that   were   already   markedly   higher   than   the   other   provinces   (51   and   48   resections     per   capita,   respectively)   in   2010-­‐2012.   The   rates   increased   across   all   provinces   from   2004-­‐06   to   2010-­‐12,   except   Saskatchewan.   Between   Newfoundland,  with  the  lowest  resection  rate  for  lung  surgeries,  and  New  Brunswick  with   the   highest   rate,   there   was   a   difference   of   30   resections   per   capita.   These   findings   combined  with  the  relative  differences  between  the  incidence  and  resection  rates  possibly   indicate  disparities  in  care  delivery  based  on  geographic  location.       70    Rate  (/100,000)   60   50   40   30   20   10   0   AB   BC   Resection  Rate  (2010-­‐12)   MB   NB   NL   NS   Resection  Rate  (2004-­‐06)   ON   SK   Incidence  Rate  (2010)     Figure  4.2.4e:  Age-­‐standardized  resection  rates  per  100,000  populations  for  lung  cancer   surgeries     In-­‐Hospital  Mortality     The  overall  lung  cancer  surgery  in-­‐hospital  mortality  rate  for  the  9-­‐year  period  was  2.3%,   with  a  large  (50%)  reduction  in  mortality  occurring  between  2004  (3.2%)  and  2012  (1.6%)   (Figure   4.2.4f).   Differences   between   provinces   were   also   apparent.   New   Brunswick   had   a   mortality   rate   1.4%   higher   than   the   9-­‐year   national   average   and   Ontario,   Saskatchewan,   and   Nova   Scotia   had   marginally   higher   age-­‐adjusted   mortality   rates.   A   reduction   in   mortality   rates   over   time   within   each   province   was   seen   as   well   (Figure   4.2.4g).   For   instance,  Ontario  had  a  crude  mortality  rate  of  3.1%  in  the  first  3  years  (2004-­‐06)  which   reduced  to  1.8%  in  last  3  years  (2010-­‐12).  Similarly,  a  reduction  from  2.38%  to  1.09%  in   British  Columbia,  and  2.08%  to  1.6%  in  Alberta  was  recorded  over  the  same  period.  Such   differences   were   difficult   to   present   for   provinces   with   smaller   caseloads.   The   relatively     P a g e    76     recent   large   scale   adoption   of   minimally   invasive   techniques,   a   paradigm   shift   that   is   unique  to  lung  surgery,  may  have  also  influenced  the  rate  of  mortality.       4   Mortality  Rate  (%)   3.5   3   2.5   2   1.5   1   0.5   0   2003   2004   2005   2006   2007   2008   2009   2010   2011   2012   2013   Fiscal  Year   Figure  4.2.4f:  Annual  age-­‐adjusted  mortality  rate  for  lung  cancer  surgeries  (2004-­‐12)     5   4.5   3.78%   Mortality  Rate  (%)   4   3.5   3   2.54%   2.52%   2.5   2   1.65%   1.66%   1.60%   2.38%   1.52%   1.5   1   0.5   0   AB   BC   MB   NB   NL   NS   ON   SK   Province   Figure  4.2.4g:  Age-­‐adjusted  mortality  rates  for  lung  cancer  surgeries  (2004-­‐12)       In   the   multivariate   regression   model,   higher   age,   co-­‐morbidity   score,   and   males   had   a   higher   risk   of   in-­‐hospital   mortality   after   resection   for   lung   cancer.   Open   procedures   had   twice   the   risk   of   mortality,   whereas   pneumonectomy   had   six   times   the   risk   of   mortality     P a g e    77     compared   to   minimally   invasive   procedures.   The   risk   of   mortality   also   appeared   to   have   significantly  reduced  over  time.       1800                       Number  of  Procedures     1600   1400   1200   1000   800   600   400   200   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   VATS  Sublobar   Open  Lobectomy   Open  Sublobar   Pneumonectomy   VATS  Lobectomy   Figure  4.2.4h:  Shift  in  lung  cancer  procedures  in  Canada  (2004-­‐12)     One  of  the  possible  reasons  for  this  reduction  in  mortality  rates  after  lung  cancer  resections   is   a   shift   towards   minimally   invasive   procedures   over   time,   which   is   associated   with   lower   mortality   rates   (Figure   4.2.4h).77   Alternatively,   it   is   also   possible   that   lower-­‐risk   sublobar   VATS   resections   were   able   to   be   selected   rather   than   higher   risk   lobectomy   and   pneumonectomy   procedures   due   to   an   increase   in   earlier   detection   of   smaller   tumours   during  this  time  period.   In   univariate   analysis,   both   hospital   and   surgeon   volumes   were   significantly   associated   with   the   mortality   rate.   Similarly,   thoracic   surgeons   had   a   lower   mortality   rate   (2%)   on   average   than   the   general   surgeons.   Figure   4.2.4i   demonstrates   the   crude   mortality   rate   reduction  for  annual  hospital  volume.  After  controlling  for  random  hospital  related  factors   or  provincial  variations,  hospital  volume  was  significantly  associated  with  a  reduced  risk  of   in-­‐hospital   mortality   (Appendix   4).   Surgeon   volume   and   specialty   significantly   predicted   the  risk  of  mortality  in  the  model  but  the  effect  disappeared  with  the  inclusion  of  hospital   volume.   Furthermore,   the   random   effects   of   hospital   were   also   significant   implying   that   factors  other  than  volume  and  specialty  also  predict  mortality.     P a g e    78   60 40 0 20 Mortality Rate (%) 80 100   0 100 200 Annual Hospital Volume 300   Figure  4.2.4i:  Annual  hospital  volume  and  mortality  rate  (%)  for  lung  cancer  surgeries  (2004-­‐12)   ! According   to   multivariate   analysis,   an   annual   increase   of   10   cases   in   a   given   hospital   predicted   a   3%   reduction   in   risk   of   in-­‐hospital   mortality,   after   controlling   for   the   influence   of   covariates   and   random   clustering   effects   of   individual  hospitals.       ! Assuming   that   the   quality   of   care   and   outcomes   are   the   same   across   all   the   hospitals   in   the   highest   volume   tercile,   theoretically   a   total   of   209   additional   lives  could  have  been  saved  after  lung  cancer  surgeries  in  Canada  (2004-­‐12).     Length  of  Stay     TABLE  4.2.4C:  MEDIAN  LENGTH  OF   The   median   length   of   stay   of   6   days   [IQR=5]   for   STAY       lung  cancer  surgeries  was  the  lowest  amongst  the   Province   Median  Length  of   Stay   five  cancer  types.  Over  the  course  of  the  nine-­‐year   [IQR]   study  period,  the  median  length  of  stay  decreased   AB   6  [5]   from   7   days   to   5   days   across   Canada   (Table   BC   7  [4]   4.2.4b).   There   was   a   median   difference   of   three   MB   5  [3]   days   between   Manitoba,   with   shortest   hospital   NB   8  [5]   NL   8  [4]   stay   (5   days),   and   Newfoundland   and   New   NS   6  [3]   Brunswick  with  the  longest  hospital  stay  (8  days)   ON   6  [4]   (Table  4.2.4c).  Strong  linear  associations  between   SK   7  [4]   higher   annual   facility   volume   and   higher   rates   of   discharge   6   days   after   surgery   are   evident   in   Figure   4.2.4j.   VATS   procedures   are   least     P a g e    79     invasive,  hence,  had  a  higher  discharge  rate  at  any  given  case  volume  relative  to  the  other   two  procedures.  Pneumonectomy  and  open  procedures  showed  the  greatest  variability  in   length  of  stay  at  a  volume  of  at  least  250  resections  a  year  (Figure  4.2.4k).   TABLE  4.2.4B:  LENGTH  OF  STAY  IN  CANADA  AFTER  LUNG  CANCER  SURGERY,  BY  YEAR     Median  Length  of  Stay   [IQR]   Fiscal  Year   2004   2005   2006   2007   2008   2009   2010   2011   2012   7   7   7   6   6   6   6   5   5   [4]   [4]   [5]   [4]   [4]   [4]   [4]   [4]   [4]     After  controlling  for  hospital  clustering  effect  and  other  factors  in  the  model,  higher  age,  co-­‐ morbidity  score,  and  male  patients  were  associated  with  higher  risk  of  longer  length  of  stay   (Appendix  5).  In  comparison  to  the  relatively  lower  risk  VATS  sub-­‐lobar  resection,  all  other   procedures   required   relatively   longer   hospitalization.   Yearly   reductions   in   number   of   hospital   days   were   significant.   Surgeries   in   Ontario   predicted   a   significantly   shorter   hospital  stay  in  comparison  with  those  resections  performed  elsewhere  in  Canada.  Both  the   hospital  and  surgeon  volume  significantly  predicted  shorter  length  of  stay  with  increasing   volume.  Interestingly,  surgeries  conducted  by  general  surgeons  predicted  a  shorter  length   of  stay  for  the  patients,  but  the  differences  were  not  significant.     1 .9 Discharge Rate (%) .8 .7 .6 .5 .4 .3 .2 .1 0 0 50 100 150 200 Annual Hospital Volume VATS Open 250 300 Pneumonectomy   Figure  4.2.4j:  Predicted  discharge  rate  6  days  after  lung  cancer  procedures  (2004-­‐12)     P a g e    80   1 1 .9 .9 .8 .8 .7 .7 Discharge Rate (%) Discharge Rate (%)   .6 .5 .4 .3 .6 .5 .4 .3 .2 .2 .1 .1 0 0 0 25 50 75 100 125 150 175 200 225 250 275 300 Annual Hospital Volume 4 Days 6 Days 8 Days 0 25 50 75 100 125 150 175 200 225 250 275 300 Annual Hospital Volume 4 Days 6 Days 8 Days   Figure  4.2.4k:  Predicted  discharge  rate  after  a.  Pneumonectomy,  b.  Open  Sub-­‐lobar  (2004-­‐12)     ! Assuming   that   the   quality   of   care   and   outcomes   are   the   same   across   all   the   hospitals   in   the   highest   volume   tercile,   a   total   of   3335   hospital   days   could   potentially  have  been  saved  annually  across  Canada  for  lung  cancer  surgeries.       4.2.5  OVARIAN  CANCER  SURGERY   Ovarian   cancer   resections   constituted   the   second   highest   case   volume,   with   16,949   surgeries  performed  in  232  institutions  across  Canada  (excluding  Quebec).  The  mean  age   of  patients  was  56.2  (SD=14.6)  years.  The  median  hospital  volume  was  68  resections  per   year  [IQR=101],  and  the  median  surgeon  volume  was  16  resections  per  year  [IQR=26].  The   annual   number   of   surgeries   increased   11%   from   1814   in   2004   to   2030   in   2012.   A   consistent   increase   in   the   annual   number   of   procedures   was   observed   in   Ontario,   Manitoba,  and  Nova  Scotia  (Table  4.5.2a;  Figure  4.2.5a).  Although  the  number  of  hospitals   performing  ovarian  resections  decreased  from  176  to  150  (Figure  4.2.5b)  over  the  9-­‐year   period,   this   reduction   appears   to   be   a   result   of   random   year-­‐to-­‐year   fluctuations   in   institutions  with  a  low  case  volume  (<2  surgeries  per  year).                   P a g e    81     TABLE  4.2.5A:  NUMBER  OF  OVARIAN  CANCER  SURGERIES,  BY  PROVINCE  (2004-­‐12)     AB   BC   MB   NB   NL   NS   ON   PEI   SK   Canada   2004   254   317   82   59   38   72   893   13   86   1814   2005   233   360   90   74   40   76   888   10   77   1848   2006   238   360   101   58   39   74   923   *   81   1877   2007   241   331   93   52   32   71   887   *   99   1808   2008   255   330   88   39   38   85   876   *   109   1825   2009   209   329   98   60   36   99   916   *   95   1847   2010   253   337   121   72   41   84   946   10   85   1949   2011   238   356   113   55   45   79   957   9   99   1951   2012   261   331   109   57   53   91   1054   *   69   2030   Total   2182   3051   895   526   362   731   8340   62   800   16949   *Number  suppressed  if  number  of  surgeries  <5       Ovarian   resection   types   were   classified   into   six   categories   by   an   expert   Gynecological   Oncologist   based   on   the   degree   of   surgical   intensity,   which   stratifies   the   associated   risks.   Procedure   types   in   order   of   increasing   complexity   are:   omentectomy,   minimally   invasive   oophorectomy/fallopian   tube   resection,   open   oophorectomy/fallopian   tube   resection,   minimally   invasive   simple   hysterectomy,   open   simple   hysterectomy,   and   radical   hysterectomy.   Gynecological   cancer   procedures   could   involve   any   combination   of   either   the   removal   of   ovaries,   fallopian   tube(s),   or   uterus   using   a   minimally   invasive   or   open   approach.   Procedures   were   classified   based   on   the   most   complex   surgical   intervention   in   cases   where   there   was   a   combination   of   interventions.   For   instance,   if   a   patient   underwent   omentectomy  and  open  radical  hysterectomy,  the  procedure  was  classified  as  open  radical   hysterectomy.  A  majority  of  procedures  were  categorized  as  an  open  simple  hysterectomy   (59%)  and  an  open  oophorectomy/fallopian  tube  resection  (29%).  It  is  not  uncommon  in   ovarian   cancer   surgeries   that   patients   undergo   further   debulking   or   another   associated   resection   at   a   later   point   in   time.   The   present   data   only   provided   records   for   one   patient   visit,   without   the   possibility   of   accounting   for   multiple   visits   by   the   same   patient   for   a   follow   up   procedures.   Therefore,   any   follow-­‐up   procedure(s)   would   be   treated   as   a   new   cancer  occurrence  and  a  new  surgical  case.           P a g e    82     Number  of  Surgeries   1200   1000   800   600   400   200   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Alberta   Atlantic   British  Columbia   Ontario   Prairies     Figure  4.2.5a:  Number  of  ovarian  cancer  surgeries,  by  province  (2004-­‐12)   80   Number  of  Hospitals   70   60   50   40   30   20   10   0   2004   2005   2006   2007   2008   2009   2010   2011   2012   Fiscal  Year   Alberta   Atlantic   British  Columbia   Ontario   Prairies       Figure  4.2.5b:  Number  of  institutions  performing  ovarian  cancer  surgeries,  by  province  (2004-­‐12)     P a g e    83     Travel  Distance     Ovarian  cancer  procedures  had  the  shortest  median  travel  times  in  all  provinces  for  the  five   high  risk  cancers  discussed  here.  This  trend  holds  true  for  all  the  provinces,  across  all  the   years   (Table   4.2.5b).   They   appear   to   be   least   centralized,   given   the   high   number   of   institutions   with   small   annual   case   load   (<5   resections).   Overall,   the   number   of   facilities   varied   from   year   to   year   within   the   same   province   due   to   volume   fluctuations   among   facilities  providing  <5  resections  per  year.  For  example,  102  institutions  performed  at  least   one  surgery  for  ovarian  cancer  in  Ontario  over  the  9-­‐year  period.  In  2004,  74  institutions   performed   at   least   one   surgery   as   compared   to   63   in   2012.   However,   if   only   institutions   with  more  than  one  surgery  in  a  given  year  are  counted,  the  reduction  from  2004  to  2012   was   58   to   53   hospitals.   Similar   trends   were   observed   in   other   provinces   as   well.     For   instance,   a   total   of   14   hospitals   performed   at   least   one   surgery   over   9   years   (10   in   2004   to   5   in   2012)   in   Manitoba.   Although   only   2   big   institutions   were   carrying   out   the   highest   caseload,  the  fluctuation  in  number  of  facilities  each  year  was  mainly  a  result  of  institutions   carrying  out  small  (1  to  5)  number  of  surgeries  one  year  and  nothing  the  next  year.       The   random   fluctuations   in   hospitals   performing   ovarian   cancer   surgeries   could   have   impacted  the  difference  in  median  travel  distance  and  proportion  of  individuals  within  an   hour  distance  of  surgical  centre.  Figures  4.2.5c  and  4.2.5d  demonstrate  the  distribution  of   institutions   performing   ovarian   cancer   surgeries   across   provinces   and   the   respective   changes   in   proportion   of   patients   within   1-­‐3   hours   commute.   As   opposed   to   the   other   cancer  surgeries,  2  centres  in  Prince  Edward  Island  performed  a  small  number  of  surgeries   throughout  the  9-­‐year  study  period.     TABLE  4.2.5B:  CHANGE  IN  MEDIAN  TRAVEL  DISTANCES  FROM  2004-­‐06  TO  2010-­‐12   Province     AB   BC   MB   NB   NL   NS   ON   SK         2004-­‐06   Median  Travel   Distance  in   kilometers   [IQR]   17  [106]   12  [34]   13  [96]   29  [79]   23  [136]   70  [186]   19  [47]   9  [122]   2010-­‐12   %  in  1   hour  travel   time   65   72   65   56   56   45   71   58   Median  Travel   Distance  in   kilometers   [IQR]   17  [96]   17  [47]   12  [53]   58  [177]   25  [149]   51  [187]   25  [54]   50  [161]   %  in  1  hour   travel  time   65   68   70   41   56   48   67   52   Change  from   2010-­‐12  to   2004-­‐06   Difference  in   Median  Travel   Distances   (kilometers)   0   +5   -­‐1   +29   +2   -­‐19   +6   +42   P a g e    84     Figure  4.2.5c:  Patient  travel  times  and  pan-­‐Canadian  distribution  of  hospitals  performing  ovarian  cancer  surgeries,  2004-­‐06             P a g e    85     Figure  4.2.5d:  Patient  travel  times  and  pan-­‐Canadian  distribution  of  hospitals  performing  ovarian  cancer  surgeries,  2010-­‐12     P a g e    86     Resection  Rate   Ovarian   cancer   surgeries   had   the   lowest   inter-­‐provincial   differences   in   resection   rates   amongst   all   cancers   of   interest.   Apart   from   Manitoba   with   an   increase   of   5   resections   per   capita,  the  remaining  provinces  had  minimal  changes  in  resection  rates  over  the  two  time   period.   In   2010-­‐12,   there   was   a   difference   of   over   5   resections   per   100,000   between   the   province   with   highest   number   of   ovarian   cancer   surgeries   per   capita   (Manitoba)   to   provinces  with  the  lowest  (Alberta,  Nova  Scotia)  (Figure  4.2.5e).  The  relatively  small  inter-­‐ provincial  differences  in  resection  rates  do  not  correlate  well  with  the  large  differences  in   in-­‐hospital  mortalities  in  Figure  4.2.5g.     30   Rate  (/100,000)   25   20   15   10   5   0   AB   BC   MB   NB   Resection  Rate  (2010-­‐12)   NL   NS   ON   SK   Resection  Rate  (2004-­‐06)     Figure  4.2.5e:  Age-­‐standardized  resection  rates  per  100,000  populations  for  ovarian  cancer  surgeries     In-­‐Hospital  Mortality     Ovarian  cancers  had  the  lowest  mortality  rate  amongst  the  five  cancers  under  investigation.   The  overall  in-­‐hospital  mortality  rate  for  the  9-­‐year  study  period  was  0.9%,  declining  from   1.3%   in   2004   to   0.69%   in   2012   (Figure   4.2.5f),   with   variations   is   mortality   rates   evident   over   the   9-­‐year   period.   Provincially,   there   was   apparent   variation.   Prince   Edward   Island   had   the   highest   in-­‐hospital   mortality   of   2.84%   (SD=1.3)   although   it   only   performed   60   resections  in  9  years  and  Newfoundland,  Nova  Scotia,  and  New  Brunswick  had  age-­‐adjusted   mortality   rates   that   were   considerably   higher   than   the   national   9-­‐year   average   of   0.9%   (Figure  4.2.4g).             P a g e    87     2   1.8   Mortality  Rate  (%)   1.6   1.4   1.2   1   0.8   0.6   0.4   0.2   0   2003   2004   2005   2006   2007   2008   2009   2010   2011   2012   2013   Fiscal  Year     Figure  4.2.5f:  Annual  age-­‐adjusted  mortality  rates  for  ovarian  cancer  surgeries  (2004-­‐12)   3   Mortality  Rate  (%)   2.5   1.84   1.72   2   1.44   1.5   1.05   1.08   1   0.70   0.76   BC   MB   0.79   0.5   0   AB   NB   NL   Province   NS   ON   SK     Figure  4.2.5g:  Age-­‐adjusted  mortality  rates  for  ovarian  cancer  surgeries  (2004-­‐12)   In   univariate   analysis,   hospital   volume   was   significantly   associated   with   mortality   rate   (Figure  4.2.4h).  These  outcomes  persisted  even  after  controlling  for  provincial  differences   unaccounted   for   in   the   model.   There   were   non-­‐significant   differences   between   obstetrics/gynecologists   and   gynecological   oncologists   for   in-­‐hospital   mortality,   while   general   surgeons   had   a   significantly   higher   risk   of   in-­‐hospital   mortality.   A   non-­‐significant   association  between  surgeon  volume  and  risk  of  in-­‐hospital  mortality  was  observed.  Higher   age,   co-­‐morbidity   score,   and   the   addition   of   bowel   resections   for   complete   resection   predicted   a   significantly   higher   risk   of   mortality.   There   were   no   significant   differences   in       P a g e    88     mortality  risk  found  over  time  (years),  between  different  procedures,  or  between  Ontario   and  rest  of  the  country.       ! After  adjusting  for  patient-­‐specific  factors,  procedure  type,  and  hospital  random   effects,  an  increase  of  10  ovarian  cancer  resections  per  institution  in  a  given  year   was  associated  with  7%  reduction  in  the  risk  of  in-­‐hospital  mortality.     60 40 0 20 Mortality Rate (%) 80 100 ! Assuming   that   the   quality   of   care   and   outcomes   are   the   same   across   all   the   hospitals  in  the  highest  volume  tercile,  theoretically  a  total  of  48  additional  lives   could  have  been  saved  after  ovarian  cancer  surgeries  in  Canada  (2004-­‐12).   0 50 100 150 Annual Hospital Volume 200 250   Figure  4.2.5h:  Annual  hospital  volume  and  mortality  rate  (%)  for  ovarian  cancer  surgeries  (2004-­‐12)                                 P a g e    89     Length  of  Stay         TABLE  4.2.5B:  MEDIAN  LENGTH   For   ovarian   cancer   surgery,   the   overall   median   length   OF  STAY       of  stay  was  4  days  [IQR=3-­‐6].  There  were  only  minute   Province   Median  Length  of   differences  between  the  provinces  for  length  of  hospital   Stay  [IQR]   AB   4  [3]   stay.   Nova   Scotia,   Saskatchewan,   New   Brunswick,   and   BC   3  [3]   Newfoundland   had   the   highest   (5   days),   whereas   MB   4  [3]   British   Columbia   had   the   lowest   median   length   of   stay   NB   5  [5]   (3  days)  (Table  4.2.5b).  No  variations  in  median  length   NL   5  [3]   of   stay   were   observed   over   time   (Table   4.2.5c).   NS   5[4]   ON   4  [3]   Omentectomy   and   open   procedures   (oophorectomy,   SK   5  [4]   fallopian   tube   resection,   simple   hysterectomy)   showed   a   strong   association   between   higher   facility   volume   and   increased   discharge   rate   (Figure   4.2.5i).     Patient-­‐related   factors,   bowel   resections   and   an   increasing   number   of   combined   gynecological  resections  during  the  same  operation  predicted  a  significantly  higher  length   of   stay.   Similarly,   compared   to   minimally   invasive   oophorectomy,   the   risk   of   longer   hospitalization   was   higher   for   all   other   resection   types.   A   higher   annual   hospital   volume   was  significantly  associated  with  lower  length  of  stay,  as  opposed  to  higher  surgeon  volume   that   was   associated   with   a   higher   risk   of   prolonged   hospitalization.   There   were   non-­‐ significant   differences   between   obstetrician/gynecologists   and   gynecological   oncologists,   however,  general  surgeons  had  significantly  higher  risk  of  longer  length  of  stay.  Moreover,   non-­‐significant   differences   existed   between   the   Ontario   and   rest   of   the   country   in   the   number  of  hospital  days  post  ovarian  cancer  resections.   TABLE  4.2.5C:  MEDIAN  LENGTH  OF  STAY  IN  CANADA,  BY  YEAR     Median  Length  of  Stay   [IQR]   Fiscal  Year   2004   2005   2006   2007   2008   2009   2010   2011   2012   4[4]   4[3]   4[3]         4[3]   4[3]   4[3]   4[2]   4[3]   4[2]   1 1 .9 .9 .8 .8 Discharge Rate (100%) Discharge Rate (100%) P a g e    90   .7 .6 .5 .4 .3 .7 .6 .5 .4 .3 .2 .2 .1 .1 0 0 0 25 50 3 Days 75 100 125 150 Annual Hospital Volume 4 Days 5 Days 175 200 225 7 Days 0 25 50 3 Days 75 100 125 150 Annual Hospital Volume 4 Days 5 Days 175 200 225 7 Days Figure  4.2.5i:  Predicted  discharge  rate  after  a.  Omentectomy,  b.  Open  Oophorectomy/Simple   Hysterectomy  (2004-­‐12)   ! There  were  no  differences  in  median  number  of  hospital  days  between  hospitals   in  lowest  and  highest  volume  categories.   P a g e    91     5.0:  REGIONALIZATION:  PROVINCIAL   CASE  STUDIES   CHAPTER  HIGHLIGHTS   " " " " " " " The  review  of  literature  and  published  policies  of  health  care  delivery  organizations  shows  that  at   the  present  time,  there  are  no  published  pan-­‐Canadian  standards  for  the  delivery  of  high  risk   cancer  surgeries   Discrepancies  were  found  between  the  provinces  with  regard  to  the  approaches  undertaken  to   provide  surgical  cancer  care.  Given  the  varying  geographic,  political,  and  economic  differences   between  the  different  provinces,  it  is  rather  impractical  to  have  one  set  of  standards  for  every   jurisdiction.   Passive  centralization  has  occurred  in  most  provinces  (other  than  Ontario)  despite  not  having  a   formal  regionalization  strategy   There  appears  to  be  minimal  regulation  at  the  provincial  level  for  organization  of  delivery   structure,  and  availability  of  published  guidelines  to  ensure  standardization  of  high  quality  care.     o Ontario  is  the  only  province  with  specific  externally  published  standards  and  guidelines   for  directing  high  risk  surgical  procedures  to  designated  centres   The  key  informant  interviews  revealed  a  lack  of  accountability  in  most  of  the  country;  inability  to   enact  change;  or  a  formal  process  to  track  and  evaluate  outcomes  of  surgical  care     The  most  striking  finding  of  this  pan-­‐Canadian  analysis  is  the  tremendous  variance  in  care  across   the  country.  In  many  of  the  surgeries  evaluated,  the  mortality  difference  between  the  provinces   with  the  lowest  and  highest  mortality  rates  shows  an  increase  of  mortality  of  three  to  four   times.  Similarly,  the  length  of  stay  can  vary  by  2-­‐3  days  on  average  per  patient  depending  on   where  the  procedure  is  performed.  By  opting  for  regionalization  policies,  not  only  could  we   influence  mortality  but  we  could  do  it  in  a  more  efficient  and  cost  effective  manner.  These   results  are  clinically  striking  and  statistically  significant  as  well,  which  further  emphasizes  their   significance.     Particularly  in  the  case  of  ovarian  cancer,  the  current  state  of  care  seems  to  be  very  much  in  need   of  a  regionalization  effort.  It  has  half  the  number  of  cases  nationally  as  lung  cancer,  but  three   times  the  number  of  institutions  performing  these  surgeries  with  many  of  the  centres  reporting  a   very  small  total  annual  case  volume.  Considering  the  demonstrated  improvements  in  mortality   rates  from  regionalization  to  high  volume  centres,  it  encourages  health  care  planners  to  consider   addressing  this  issue  along  with  topic  experts  and  regional  health  system  stakeholders  through   the  implementation  of  regionalization  policies.         P a g e    92     5.1  OVERVIEW     In   this   section,   an   insight   into   the   structures   of   care   for   the   delivery   of   cancer   surgical   procedures  is  presented  along  with  reports  of  outcomes  compared  to  the  national  average.   In  the  absence  of  available  reporting  on  policies  and  standards  in  place  for  the  delivery  of   complex   cancer   surgeries,   structured   interviews   were   conducted   with   experts   across   Canada   to   gain   their   perspective   on   the   delivery   and   organization   of   surgical   cancer   care   in   their  respective  province.  The  participants  included  practicing  surgeons,  heads  of  surgical   oncology,  or  stakeholders  who  are  involved  with  decision  making  for  cancer  surgical  care.   Provincial  report  cards  are  presented  to  allow  for  intra-­‐  and  inter-­‐provincial  comparisons   on   mortality,   length   of   stay,   and   travel   times   based   on   the   latest   available   data.   The   rates   of   mortality,   median   length   of   stay   and   travel   times   for   provinces   with   smaller   population   sizes  are  subject  to  large  variations.  For  instance,  if  twenty  esophagectomies  are  performed   in  a  province,  only  one  instance  of  mortality  can  increase  the  mortality  rate  from  0%  to  5%.   Therefore,   3-­‐year   crude   averages   for   the   primary   outcomes   are   presented   to   draw   fair   comparisons.   Similarly,   a   number   of   hospitals   across   Canada   performed   only   one   surgery   per   year,   which   affected   the   total   number   of   institutions   performing   surgeries   in   a   given   year.   For   the   purpose   of   addressing   the   large   range   of   institutional   volumes   and   random   variation  within  a  given  hospital,  a  count  of  facilities  with  an  annual  volume  of  >5  was  also   stated.   The   discussion   below   is   intended   to   be   used   by   stakeholders   to   broaden   their   understanding  of  current  organizational  structures  of  surgical  care  and  associated  patient-­‐ based  outcomes  in  their  respective  provinces.           5.2  ALBERTA   Alberta  Health  Services  directly  oversees  the  structure  of  the  entire  health  care  system  in   the   province,   including   the   delivery   of   the   five   cancer   surgeries   under   discussion   in   this   report.   The   province   is   further   divided   into   6   zones.   Cancer   Control   Alberta   oversees   the   cancer   centre-­‐based   delivery   of   chemotherapy   and   radiotherapy   oncological   services.   The   zones   are   authorized   to   oversee   cancer   care,   including   surgical   cancer   care   within   their   jurisdiction,  working  together  with  the  institutions  undertaking  the  procedures.  Alberta  is   described  as  having  an  asymmetric  bipolar  structure  of  care,  with  the  South  zone  managing   a   significant   portion   of   the   total   provincial   surgical   volume   and   equipped   with   a   broader   infrastructure  for  delivering  surgical  cancer  services.  Recognizing  the  need  for  a  provincial   body   for   policy-­‐related   decisions   concerning   cancer   surgical   care,   the   "Strategic   Clinical   Network"  has  been  established  in  association  with  Alberta  Health  Services.     Overall,   it   appears   that   Alberta   has   some   quality   improvement   initiatives   underway,   but   there  is  nothing  specifically  with  regards  to  ‘active’  regionalization  or  the  implementation   of   policy   guidelines   for   the   cancer   types   under   discussion   in   this   report.   Based   on   conversations   with   key   informants,   it   appears   that   there   has   been   passive   regionalization   for   some   surgical   procedures   as   a   result   of   steps   taken   by   individual   surgeons   to   open       P a g e    93     surgical  centres.  Table  5.1  provides  evidence  for  some  form  of  consolidation  of  esophagus,   lung,  and  liver  surgeries  with  an  attendant  increase  in  the  number  of  resections.     The  number  of  resections  carried  out  in  2012  was  higher  than  the  3-­‐year  provincial  average   for  all  procedures.  Ovarian  cancer  surgeries  had  the  second  highest  volume  in  the  province,   yet   they   appear   to   be   least   centralized.   Only   three   institutions   had   a   volume   greater   than   five  in  2012,  whereas  two  centres  undertook  greater  than  90  resections.  The  crude,  3-­‐year   mortality   rates   appear   to   be   higher   than   the   national   average   for   pancreatic   and   ovarian   cancer   procedures.   Similarly,   the   median   length   of   stay   was   higher   than   the   national   average  for  all  but  ovarian  cancer  resections,  which  indicates  room  for  improvement.  The   travel  distance  was  more  than  twice  longer  for  esophagectomies  as  compared  to  the  other   procedures,  with  the  lowest  (49%)  proportion  of  patients  within  one-­‐hour  commute.   TABLE  5.1:  STATE  OF  HIGH-­‐RISK  SURGERIES  IN  ALBERTA     #  Surgeries  in  2012   (Average  2010-­‐12)   #  Hospitals  in  2012   (#  with  Case  Volume   >5)   Mean  Hospital  Case   Volume  in  2012   (range)   %  Mortality  Rate   (2010-­‐12)   %  Mortality  Rate   National  Average   (2010-­‐12)   (Standard  Error)   Median  LOS  [IQR]   (2010-­‐12)   Median  LOS   National  Average   [IQR]  (2010-­‐12)   Median  Travel   Distance  (2010-­‐12)   Kilometers  [IQR]   %  of  patient  in  1-­‐ hour  travel  time   (2010-­‐12)   Esophagus   49   (44)   4   (2)   Pancreas   111   (96)   7   (2)   Liver   224   (184)   5   (3)   Lung   403   (370)   2   (2)   Ovarian   261   (251)   16   (3)   12   (2-­‐33)   16   (1-­‐60)   44   (1-­‐130)   202   (131-­‐272)   16   (1-­‐124)   2.29   4.17   2.35   1.62   1.46   4.74  (0.68)     2.90  (0.42)       2.63  (0.27)       1.63  (0.12)       0.74  (0.11)     15  [12]   11  [8]   8  [6.5]   6  [5]   4  [3]     13  [10]     10  [9]     7  [5]     5  [4]     4  [2]   47  [160]   22  [165]   18  [144]   18  [118]   17  [96]   49   57   62   62   65             P a g e    94     5.3  BRITISH  COLUMBIA   The  British  Columbia  Cancer  Agency  (BCCA)  does  not  have  a  formal  role  in  regulating  the   structure   and   delivery   of   cancer   surgical   care.   The   BCCA   plays   a   supporting   role   in   improving   the   quality   of   surgical   care   in   the   province.   For   instance,   a   ‘Surgical   Oncology   Network’   has   been   established   to   improve   coordination   and   communication   between   oncology   service   providers   and   the   surgeons,   in   addition   to   connecting   these   surgery   providers   with   the   BCCA.   The   Surgical   Oncology   Network   has   different   committees   that   aid   in   the   planning,   support,   and   implementation   of   its   strategic   initiatives,   including   the   development   and   implementation   of   clinical   practice   guidelines,   continuing   professional   development,  knowledge  transfer,  and  research  and  outcomes  evaluation.     The   key   informant   interviews   revealed   that   the   thoracic   cancer   surgeries   have   been   formally  regionalized  in  British  Columbia  with  minimum  direct  involvement  from  the  BCCA.   However,  the  BCCA  appears  to  have  recently  played  a  significant  role  in  the  centralization   of   gynecological   surgeries.   In   one   of   the   thoracic   surgical   cancer   care   centres   in   Kelowna,   the   initial   assessments   are   completed   via   video-­‐conferencing   with   surgical   teams   to   help   ameliorate   some   of   the   significant   travel   burden   that   would   otherwise   limit   access   to   surgical   care   for   remote   patients.   This   example   of   innovation   may   be   helpful   for   jurisdictions  looking  to  limit  the  impact  of  regionalization  policies  on  patients.       The  analysis  of  data  shows  that  British  Columbia  appears  to  have  managed  the  increasing   volume   of   lung,   liver,   and   esophageal   procedures   by   consolidating   surgical   care   into   four   large   volume   centres.   The   annual   number   of   procedures   for   pancreatic,   liver,   and   lung   cancer   was   higher   than   the   3-­‐year   provincial   average   implying   an   increasing   case   load.   Overall,   the   3-­‐year   mortality   rates   were   lower   than   the   national   average   for   all   the   procedures   under   investigation.   The   number   of   institutions   performing   less   than   5   cases   annually   may   present   with   potential   room   for   improvement.   For   instance,   a   total   of   31   institutions   carried   out   ovarian   cancer   surgeries   but   only   eight   of   these   centres   were   conducting  more  than  five  resections  in  2012  (Table  5.2).  This  number  does  not  represent   the   vast   case   volume   differences   across   these   eight   institutions,   which   had   a   case   load   ranging   from   6   to   174   surgeries.   Similarly,   five   of   nine   institutions   performed   a   small   number   of   pancreatic   procedures.   Although   the   3-­‐year   crude   mortality   rate   was   over   5   times  less  than  the  national  average,  the  median  length  of  stay  was  higher  than  the  national   average  for  pancreatic  cancer  procedures.  Similarly,  the  mortality  rates  were  roughly  50%   less  than  the  national  average  for  ovarian  and  esophageal  procedures,  with  a  median  length   of   stay   lower   than   the   national   average.   Each   of   the   procedures   was   generally   within   an   hour   travel   time   for   a   majority   of   the   patients.   The   higher   number   of   institutions   performing  ovarian  procedures  could  account  for  shortest  travel  burden  on  the  patients  for   these  resections.           P a g e    95      TABLE  5.2:  STATE  OF  HIGH-­‐RISK  SURGERIES  IN  BRITISH  COLUMBIA     #  Surgeries  in  2012   (Average  2010-­‐12)   #  Hospitals  in  2012   (#  with  Case  Volume   >5)   Mean  Hospital  Case   Volume  in  2012   (range)   %  Mortality  Rate   (2010-­‐12)   %  Mortality  Rate   National  Average   (2010-­‐12)   (Standard  Error)   Median  LOS  [IQR]   (2010-­‐12)   Median  LOS   National  Average   [IQR]  (2010-­‐12)   Median  Travel   Distance  (2010-­‐12)   Kilometers  [IQR]   %  of  patient  in  1-­‐ hour  travel  time   (2010-­‐12)   Esophagus   66   (64)   Pancreas   92   (77)   Liver   221   (208)   Lung   614   (550)   Ovarian   331   (341)   6   (4)   9   (4)   5   (4)   4   (4)   31   (8)   11   (1-­‐21)   10   (1-­‐53)   44   (1-­‐144)   154   (102-­‐224)   11   (1-­‐174)   2.60   0.43   2.40   1.09   0.39   4.74  (0.68)     2.90  (0.42)       2.63  (0.27)       1.63  (0.12)       0.74  (0.11)     11  [9]   12  [10]   7  [5]   5  [4]   3  [2]     13  [10]     10  [9]     7  [5]     5  [4]     4  [2]   28  [88]   28  [59]   20  [60]   25  [93]   17  [47]   57   57   62   58   68   5.4  MANITOBA   Health  care  delivery  in  Manitoba  has  been  organized  by  the  five  Regional  Health  Authorities   (RHA)   since   2012,   with   Cancer   Care   Manitoba   (CCMB)   responsible   for   overseeing   cancer   services   within   the   province.   Cancer   care   consists   of   a   primary   facility   in   Winnipeg   and   a   secondary   institution   located   in   Brandon.   Given   the   presence   of   two   major   cancer   treatment   centres   in   the   province,   the   surgical   care   seems   to   be   ‘passively’   centralized.   There   is   one   institution   in   Winnipeg   managing   the   highest   case   load   for   all   five   high-­‐risk   cancer   surgeries.   Based   on   our   discussions   with   expert   surgeons   in   the   region,   there   is   evidence   of   planned   consolidation   of   thoracic   cancer   surgeries.   In   2007,   all   thoracic   surgeries  were  moved  to  a  single  centre  in  Winnipeg,  with  only  a  small  number  of  thoracic   surgeries  still  carried  out  in  Brandon.   The  number  of  surgeries  for  all  five  cancers  did  not  show  much  deviance  from  the  3-­‐year   provincial   average.   A   strong   trend   for   consolidation   of   surgical   care   to   a   small   number   of   institutions  is  evident  for  all  five  cancer  procedures.  For  instance,  only  one  institution  was   performing   liver   resections   in   2012,   down   from   four   in   2004.   Likewise,   the   number   of       P a g e    96     institutions  performing  ovarian  cancer  surgeries  was  down  from  ten  to  five  over  the  study   period.   In   spite   of   apparent   centralization   of   surgical   care,   the   median   travel   distances   remained   close   to   15   kilometers   for   all   surgeries.   There   were   small   variations   in   the   proportion   of   patients   who   travelled   less   than   an   hour   for   surgical   care   across   all   the   procedures.  About  70%  of  the  patients  for  ovarian  cancer  surgeries  travelled  less  than  an   hour,  compared  to  53%  for  esophageal  cancer  procedures.   The   case   study   of   Manitoba   presents   with   an   interesting   scenario.   The   3-­‐year   crude   mortality   rate   for   esophageal   cancer   resections   was   more   than   three   times   that   of   the   national  average,   whereas   the   median  length  of  stay  was  2  days  higher  (Table  5.3).  Some  of   the   variation   in   mortality   rate   could   be   explained   by   the   small   total   provincial   case   load   leading   to   significant   year   over   year   variance;   however   it   may   also   emphasize   the   significance   of   factors   other   than   hospital   volume   on   patient   outcomes   (i.e.   surgeon   specialty,   staff   training).   A   lack   of   available   data   on   such   measures   limits   our   ability   to   comment   on   the   observed   differences   in   the   rate   of   mortality   for   esophageal   resections.   Pancreatic   resection   data   reported   a   mortality   rate   and   length   of   stay   higher   than   the   national   average   in   spite   of   the   fact   that   these   resections   were   carried   out   in   only   2   institutions.       TABLE  5.3:  STATE  OF  HIGH-­‐RISK  SURGERIES  IN  MANITOBA     #  Surgeries  in  2012   (Average  2010-­‐12)   #  Hospitals  in  2012   (#  with  Case  Volume   >5)   Mean  Hospital  Case   Volume  in  2012   (range)   %  Mortality  Rate   (2010-­‐12)   %  Mortality  Rate   National  Average   (2010-­‐12)  (Standard   Error)   Median  LOS  [IQR]   (2010-­‐12)   Median  LOS  National   Average  [IQR]  (2010-­‐ 12)   Median  Travel   Distance  (2010-­‐12)   Kilometers  [IQR]   %  of  patient  in  1-­‐hour   travel  time  (2010-­‐12)       Esophagus   16   (12)   Pancreas   31   (27)   Liver   55   (55)   Lung   226   (233)   Ovarian   109   (114)   2   (1)   2   (2)   1   (1)   2   (2)   5   (3)   8   (2-­‐14)   16   (8-­‐23)   55   (55)   113   (13-­‐213)   21   (1-­‐62)   16.67   5.0   2.42   1.29   0.58   4.74  (0.68)     2.90  (0.42)       2.63  (0.27)       1.63  (0.12)       0.74  (0.11)     15  [13]   13  [11.5]   7  [4]   4  [3]   4  [3]     13  [10]     10  [9]     7  [5]     5  [4]     4  [2]   14  [104]   16  [94]   13  [85]   13  [94]   12  [53]   53   59   64   63   70   P a g e    97     5.5  NEW  BRUNSWICK     New   Brunswick   is   divided   into   2   Regional   Health   Authorities   (RHAs),   which   serve   the   Anglophone   and   Francophone   populations.   Although   the   various   cancer   services   are   coordinated   by   the   New   Brunswick   Cancer   Network,   the   delivery   and   organization   of   surgical   care   seems   to   be   the   responsibility   of   the   RHAs   and   the   provider   institution.   By   nature   of   relatively   small   geography,   surgery   in   New   Brunswick   operates   out   of   a   small   number  of  centres  and  has  a  small  total  surgical  case  load.  Overall,  higher  volume  centres   were   located   in   Moncton,   Fredericton,   St.   John,   and   Edmundston.   There   is   an   obvious   overlap  between  institutions  performing  resections  for  all  five  cancer  types.  Based  on  our   interviews   with   key   informants,   there   does   not   appear   to   be   any   formal   centralization   initiatives  directed  specifically  towards  cancer  surgeries.         New  Brunswick  had  smallest  number  of  resections  of  all  the  provinces.  Therefore,  year  over   year   fluctuations   in   the   number   of   surgeries   also   affected   the   number   of   institutions   performing   surgeries   annually.   Although   a   variation   in   number   of   facilities   was   observed   over   time,   it   could   mainly   be   a   result   of   year-­‐over-­‐year   fluctuations   in   the   number   of   surgeries.  For  instance,  the  number  of  institutions  for  pancreatic  cancer  reduced  from  six  in   2004   to   three   in   2012   (Table   5.4).   However,   it   is   unclear   whether   the   reduction   in   the   number   of   surgical   institutions   was   a   planned   move   or   a   result   of   random   fluctuations   in   the  number  of  resectable  cases  within  that  facility  catchment  area.  Small  variations  in  the   number   of   surgeries   were   observed   over   the   most   recent   3-­‐year   period.   None   of   the   institutions  for  liver,  esophagus,  and  pancreatic  cancer  had  a  case  volume  greater  than  five   in   2012.   The   3-­‐year,   crude   mortality   rate   was   smaller   than   the   national   average   for   all   cancer   resections   except   lung.   The   median   length   of   stay   was   smaller   than   the   national   average   only   for   esophageal   cancer.   However,   small   case   volumes   limit   our   ability   to   present   a   valid   comparison   between   the   national   average   and   New   Brunswick   data.   The   province  might  be  better  compared  with  other  provinces  with  small  case  load  and  relatively   large  geographic  catchment  areas.  Similarly,  the  travel  times  show  vast  differences  between   the   different   cancer   types,   but   the   large   interquartile   ranges   demand   caution   when   interpreting  these  results.     TABLE  5.4:  STATE  OF  HIGH-­‐RISK  SURGERIES  IN  NEW  BRUNSWICK     #  Surgeries  in  2012   (Average  2010-­‐12)   #  Hospitals  in  2012   (#  with  Case  Volume   >5)   Mean  Hospital  Case   Volume  in  2012   (range)   %  Mortality  Rate       Esophagus   13   (10)   Pancreas   9   (9)   Liver   6   (7)   Lung   213   (200)   Ovarian   57   (61)   5   (0)   3   (0)   4   (0)   6   (6)   7   (2)   2   (1-­‐5)   3   (2-­‐4)   1.5   (1-­‐3)   36   (6-­‐63)   8   (1-­‐38)   3.23   0   0   2.16   0.54   P a g e    98     (2010-­‐12)   %  Mortality  Rate   National  Average   (2010-­‐12)   (Standard  Error)   Median  LOS  [IQR]   (2010-­‐12)   Median  LOS   National  Average   [IQR]  (2010-­‐12)   Median  Travel   Distance  (2010-­‐12)   Kilometers  [IQR]   %  of  patient  in  1-­‐ hour  travel  time   (2010-­‐12)   4.74  (0.68)     2.90  (0.42)       2.63  (0.27)       1.63  (0.12)       0.74  (0.11)     12  [6]   11  [14]   8  [3]   7  [5]   5  [3]   13  [10]   10  [9]   7  [5]   5  [4]   4  [2]   20  [41]   102  [146]   33  [124]   48  [115]   58  [177]   65   35   55   45   41   5.6  NEWFOUNDLAND  AND  LABRADOR   Newfoundland   and   Labrador’s   health   care   system   is   governed   by   the   Department   of   Health   and  Community  Services.  There  are  four  Regional  Health  Authorities  in  the  province,  who   are   responsible   for   health   care   delivery   within   their   jurisdictions.   Similar   to   New   Brunswick,   Newfoundland   has   a   small   case   volume   for   all   five   cancer   resections.   These   procedures  appear  to  be  performed  in  one  central  location  in  St.  John's,  except  for  a  small   number   of   resections   performed   by   other   institutions   for   ovarian   cancer.   Therefore,   it   appears   that   there   has   been   an   informal   process   of   consolidation   of   services   within   the   province.  According  to  our  experts,  hepatic  and  thoracic  resections  are  performed  only  in   St.   John's.   Although   Eastern   Health   was   the   biggest,   stand-­‐alone   provincial   body,   it   was   mostly  involved  with  the  coordination  of  medical  and  radiation  oncology  services,  with  no   comparable  organization  with  a  provincial  mandate  to  oversee  cancer  surgical  care  across   the   four   regional   health   authorities.   The   hospital(s)   undertaking   the   surgeries   were   the   groups   primarily   responsible   for   making   decisions   concerning   surgical   care,   including   the   scope  of  practice  for  surgeons.     According  to  the  analysis  of  provincial  data,  the  3-­‐year  mortality  rates  were  higher  than  the   national   average   for   esophagus   and   lung   resections   but   the   small   number   of   cases   does   not   present  a  fair  comparison  (Table  5.5).  The  median  length  of  stay  was  3  days  more  than  the   national   average   for   pancreatic   and   lung   resections.   The   potential   impact   of   the   consolidation   of   care   to   one   large   centre   is   visible   in   longer   patient   travel   distances   for   esophageal,  pancreatic,  liver  and  lung  cancers.  Six  institutions  performing  ovarian  resection   do  not  appear  to  have  resulted  in  higher  mortality  rates  and  have  retained  a  short  patient   travel  distance  in  Newfoundland  and  Labrador.               P a g e    99     TABLE  5.5:  STATE  OF  HIGH-­‐RISK  SURGERIES  IN  NEWFOUNDLAND  AND  LABRADOR     #  Surgeries  in  2012   (Average  2010-­‐12)   #  Hospitals  in  2012   (#  with  Case  Volume   >5)   Mean  Hospital  Case   Volume  in  2012   (range)   %  Mortality  Rate   (2010-­‐12)   %  Mortality  Rate   National  Average   (2010-­‐12)   (Standard  Error)   Median  LOS  [IQR]   (2010-­‐12)   Median  LOS   National  Average   [IQR]  (2010-­‐12)   Median  Travel   Distance  (2010-­‐12)   Kilometers  [IQR]   %  of  patient  in  1-­‐ hour  travel  time   (2010-­‐12)   Esophagus   3   (4)   Pancreas   13   (9)   Liver   29   (21)   Lung   49   (59)   Ovarian   53   (46)   2   (0)   1   (1)   1   (1)   1   (1)   6   (1)   1.5   (1-­‐2)   13   (13)   29   (29)   49   (49)   8   (1-­‐43)   8.33   0   0   1.69   0     4.74  (0.68)     2.90  (0.42)       2.63  (0.27)       1.63  (0.12)       0.74  (0.11)     13  [11]   13  [22]   7  [5]   8  [4]   4  [2]     13  [10]     10  [9]     7  [5]     5  [4]     4  [2]   146  [422]   272  [405]   95  [286]   292  [539]   25  [149]   33   25   40   30   56   5.7  NOVA  SCOTIA     Cancer  Care  Nova  Scotia  is  the  provincial  cancer  agency  that  is  involved  with  setting  overall   standards   for   cancer   care,   except   in   the   delivery   of   surgical   cancer   care.   For   more   than   a   decade,   surgical   cancer   care   has   primarily   been   organized   by   each   of   the   nine   District   Health  Authorities  who  are  responsible  for  health  care  service  planning  and  allocation  for   residents   within   the   region   as   well   as   neighbouring   Atlantic   provinces.   Except   for   the   reporting  of  surgical  wait  times,  the  provincial  oversight  for  surgical  care  is  minimal.  It  is   also   unclear   how   the   planned   move   to   single   provincial   health   authority   (in   April   2015)   for   adult   clinical   care   will   impact   delivery   of   cancer   surgical   care.   The   major   cancer   surgical   centres   in   Nova   Scotia   are   located   in   Halifax   and   Sydney.   The   available   data   and   expert   opinion   supports   passive   regionalization   of   surgical   care   to   1   or   2   major   surgical   centres   in   Halifax  and  Sydney.     The   annual   case   volume   did   not   show   much   deviance   from   the   3-­‐year   provincial   average   (2010-­‐2012),   and   neither   did   the   number   of   institutions   from   2004-­‐2012   (Table   5.6).   Except   for   ovarian   cancer,   a   majority   of   surgeries   were   carried   out   in   two   major   centres.       P a g e    100     High   3-­‐year   crude   mortality   rates   for   liver,   pancreatic,   and   ovarian   cancers   present   with   an   interesting   discussion   and   an   opportunity   for   improvement.   Outcomes   of   esophageal   and   lung   surgeries   were   better   than   the   national   average,   although   the   number   of   cases   for   were  too  small  for  a  meaningful  comparison.  The  added  burden  of  travel  with  higher  risk  of   mortality  for  pancreatic  and  liver  resections  may  point  towards  a  need  for  improvements  in   care  processes.   TABLE  5.6:  STATE  OF  HIGH-­‐RISK  SURGERIES  IN  NOVA  SCOTIA     #  Surgeries  in  2012   (Average  2010-­‐12)   #  Hospitals  in  2012   (#  with  Case  Volume   >5)   Mean  Hospital  Case   Volume  in  2012   (range)   %  Mortality  Rate   (2010-­‐12)   %  Mortality  Rate   National  Average   (2010-­‐12)   (Standard  Error)   Median  LOS  [IQR]   (2010-­‐12)   Median  LOS   National  Average   [IQR]  (2010-­‐12)   Median  Travel   Distance  (2010-­‐12)   Kilometers  [IQR]   %  of  patient  in  1-­‐ hour  travel  time   (2010-­‐12)   Esophagus   14   (16)   Pancreas   16   (19)   Liver   71   (68)   Lung   176   (192)   Ovarian   91   (85)   2   (1)   1   (1)   3   (1)   2   (2)   10   (1)   7   (4-­‐10)   16   (16)   24   (1-­‐69)   88   (13-­‐163)   9   (1-­‐71)   0   5.36   3.45   0.87   1.18   4.74  (0.68)     2.90  (0.42)       2.63  (0.27)       1.63  (0.12)       0.74  (0.11)     12  [9]   12  [17]   7  [5]   5  [3]   4  [2]     13  [10]     10  [9]     7  [5]     5  [4]     4  [2]   93  [282]   115  [292]   193  [280]   71  [148]   51  [187]   44   36   31   45   48   5.8  ONTARIO   In   Ontario,   health   care   planning   and   delivery   is   coordinated   by   fourteen   Local   Health   Integration  Networks,  with  cancer  care  coordinated  by  Cancer  Care  Ontario  (CCO).  Through   initiatives   led   by   CCO,   Ontario   has   taken   the   most   formal,   active,   and   clearly   defined   approach   to   the   regionalization   of   high-­‐risk   complex   cancer   surgery   in   Canada.   With   respect   to   the   cancers   evaluated   in   this   report,   CCO   has   implemented   evidence-­‐based   guidelines   and   standards   for   thoracic   and   hepatopancreatobiliary   cancers,   while   the   guideline   development   process   for   ovarian   cancer   was   initiated   in   2010   and   implementation  is  currently  under  way.           P a g e    101     Pancreatic  cancer  surgical  regionalization  started  in  1999  and  continued  through  the  early   2000s,   while   thoracic   surgery   initiated   the   regionalization   process   in   Ontario   starting   in   2004-­‐2005.   This   complex   multifaceted   undertaking   was   not   instantaneous,   but   over   time   led   to   the   consolidation   of   case   volume   through   the   successful   implementation   of   explicit   and   financially   supported   regionalization   policies   as   outlined   by   CCO   (www.cancercare.on.ca).   To   support   this   process   of   regionalization,   practice   standards   were   established   in   an   evidence-­‐based   manner   to   clearly   delineate   definitions   of   high   volume   centres,   minimum   institutional   requirements   such   as   the   availability   of   surgical   step   down   units   or   training   requirements   for   surgeons,   and   other   key   components   to   promoting  high  quality  care.  Communities  of  practice  for  each  specialty  were  encouraged  to   participate   in   the   development   and   implementation   of   these   standards,   but   also   to   set   national  priorities  and  facilitate  the  transfer  of  knowledge.   This   regionalization   work   is   reflected   in   the   observed   lower   number   of   institutions  offering   these   cancer   surgeries   and   higher   mean   surgeon   case   volumes   relative   to   those   reported   in   much   of   Canada.   Despite   this   active   management   of   regionalization   in   Ontario,   there   are   still   a   number   of   smaller   centres   and   presently   unregionalized   cancers   such   as   ovarian   cancer   that   persist.   This   speaks   to   the   great   challenges   facing   governing   bodies   as   they   implement  quality  improvement  initiatives.     With   regards   to   the   province-­‐specific   outcomes   and   travel   times   in   Ontario,   most   metrics   are  comparable  to  the  rest  of  Canada.  The  travel  times  for  most  cancers  are  reasonable  for   most  patients,  with  the  vast  majority  of  patients  able  to  access  surgical  care  within  the  hour   threshold,   although   esophageal   cancer   care   sees   times   that   tend   to   exceed   one   hour.   The   length  of  stay  for  all  cancers  is  at  or  below  the  national  averages.  The  3-­‐year  mortality  rates   for   most   operations   are   near   the   national   average,   but   this   may   be   a   result   of   a   sampling   bias   as   Ontario   represents   roughly   50%   of   all   the   total   surgical   cases   performed   across   Canada  in  most  of  the  cancers  evaluated  (Table  5.7).   A   better   evaluation   of   regionalization   in   Ontario   may   be   to   compare   mortality   rates   over   time   relative   to   the   rest   of   the   country,   which   largely   had   minimal   formal   regionalization   policies  (although  province-­‐specific  undertakings  are  outlined  accordingly  in  this  section).     This  is  reported  in  Figure  5.7.  At  face  value,  the  pancreatic  mortality  is  comparable  to  the   non-­‐regionalized   cases   over   time.   When   evaluating   this   number,   one   has   to   keep   in   mind   that   the   surgeries   in   Ontario   were   largely   regionalized   by   this   period   of   time   and   interestingly,   there   is   a   disproportionate   rise   in   case   volume   in   Ontario   compared   to   the   rest   of   the   country.   The   mortality   rates   may   be   influenced   by   a   greater   number   of   cases,   which   would   naturally   result   in   a   larger   raw   number   of   deaths,   although   it   is   difficult   to   ascertain  for  certain  with  administrative  data.  The  results  for  lung  cancer  surgery,  liver  and   esophagus   are   more   complex   in   nature.   The   crossing   curves   make   statistical   analysis   difficult.   This   analytical   challenge   notwithstanding,   it   does   appear   that   the   results   are       P a g e    102     inferior   to   the   national   statistics   although   multivariate   regression   analysis   does   not   show   significance.     For   lung   cancer,   the   most   regionalized   cancer   type   in   the   time   period   studied,   the  mortality  in  Ontario  seems  to  converge  with  the  rates  for  the  rest  of  Canada.     Generally   speaking,   it   is   difficult   to   measure   differences   when   a   series   of   measures   are   undertaken   over   time.   Very   different   results   could   be   found   if   Ontario   was   evaluated   the   day  after  the  policies  were  published  or  if  some  duration  of  time  was  allowed  to  pass  before   initiating   evaluation   to   allow   processes   to   become   streamlined   and   fully   integrated   into   the   health   care   system.   Overarching   this   evaluation   issue,   there   has   been   a   considerable   number   of   other   changes   in   the   health   system   that   could   be   confounding   the   results.     Another   plausible   explanation   for   differing   mortality   observations   is   that   there   is   an   increased   number   of   patients   receiving   curative   resections   in   Ontario   (See   Chapter   4).   Resection   rate   is   a   combination   of   cancer   stage   at   time   of   presentation,   indications   for   surgery   and   technical   considerations.   Regionalization,   in   theory,   facilitates   collaboration   between   highly   trained   surgeons   while   also   encouraging   the   development   of   specialized   staff   and   infrastructure   that   may   allow   patients   with   a   higher   stage   disease   or   more   complex   co-­‐morbidities   to   undergo   resection   in   Ontario.   The   higher   risk   of   mortality   may   be   resultant   from   this   increased   patient   complexity.   With   this   said,   it   is   important   to   remember  that  this  is  just  a  hypothetical  explanation  for  the  observed  findings.     In   this   context,   mortality   and   length   of   stay   have   both   fallen   over   time,   and   the   attributable   component   of   regionalization   to   this   is   undeterminable.   Perhaps   the   greatest   impact   of   regionalization  can  be  found  in  more  subjective  measures  that  are  difficult  to  measure  such   as  adherence  to  guidelines,  appropriate  surgical  decision  making,  or  effective  utilization  of   medical  and  radiation  oncology  to  name  a  few.    Over  time,  an  improvement  in  care  across   many  domains  of  cancer  care  has  been  observed,  but  there  continues  to  be  a  tremendous   variability  in  outcomes.  In  the  Ontario  example,  there  has  been  a  tightening  of  this  variance,   which   allows   for   more   consistent   care.   Overall,   these   outcomes   provide   a   complex   and   unanswerable   series   of   questions.   What   does   appear   to   be   missing   in   the   Ontario   experience  is  quality  feedback  loops  on  facility  specific  outcomes.   The   Ontario   regionalization   experience   serves   as   a   robust,   if   not   complex,   model   of   implementation   of   regionalization   for   the   rest   of   Canada.   The   large   population   of   Ontario,   the   geographic   density   and   infrastructure   are   not   necessarily   transferrable,   but   the   lessons   learned   from   successfully   implementing   these   significant   structural   changes   despite   the   challenges   can   serve   to   provide   lessons   for   other   future   adopters.   Given   that   Ontario   has   utilized  a  multi-­‐focal  approach  to  actively  implementing  regionalization,  it  is  impossible  to   dissect  out  the  impact  of  regionalization  on  the  quality  of  patient  care  in  isolation,  but  as  a   whole,  Ontario’s  regionalization  activities  appear  to  have  positively  affected  the  delivery  of   care.         P a g e    103     8   Pancrea'c  Cancer  Resec'ons Liver  Cancer  Resec'ons   6   5   6   4   4   3   2   2   1   0   2004   12   2006   2008   2010   2012   ONTARIO   REST  OF  CANADA   Esophageal  Cancer  Resec'ons 0   2004   2006   2008   2010   2012   ONTARIO   REST  OF  CANADA   Lung  Cancer  Resec'ons 5   10   4   8   3   6   2   4   1   2   0   2004   2006   2008   2010   2012   ONTARIO   REST  OF  CANADA   0   2004   2006   2008   2010   2012   ONTARIO   REST  OF  CANADA   Figure  5.7    Age-­‐adjusted  Mortality  Rates  Ontario  vs.  Rest  of  Canada  (2004-­‐12)   TABLE  5.7:  STATE  OF  HIGH-­‐RISK  SURGERIES  IN  ONTARIO     #  Surgeries  in  2012   (Average  2010-­‐12)   #  Hospitals  in  2012   (#  with  Case  Volume   >5)   Mean  Hospital  Case   Volume  in  2012   (range)   %  Mortality  Rate   (2010-­‐12)   %  Mortality  Rate   National  Average   (2010-­‐12)   (Standard  Error)   Median  LOS  [IQR]   (2010-­‐12)   Median  LOS       Esophagus   161   (165)   Pancreas   316   (280)   Liver   638   (583)   Lung   1964   (1819)   Ovarian   1054   (986)   15   (9)   13   (9)   20   (11)   24   (17)   63   (21)   11   (1-­‐32)   24   (1-­‐93)   32   (1-­‐149)   81   (1-­‐292)   17   (1-­‐222)   5.65   2.98   2.86   1.83   0.68   4.74  (0.68)     2.90  (0.42)       2.63  (0.27)       1.63  (0.12)       0.74  (0.11)     12  [11]   9  [7]   7  [4]   5  [3]   4  [2]             P a g e    104     National  Average   [IQR]  (2010-­‐12)   Median  Travel   Distance  (2010-­‐12)   Kilometers  [IQR]   %  of  patient  in  1-­‐ hour  travel  time   (2010-­‐12)   13  [10]   10  [9]   7  [5]   5  [4]   4  [2]   31  [79]   32  [76]   31  [72]   26  [69]   25  [54]   61   61   57   63   67   5.9  PRINCE  EDWARD  ISLAND  AND  TERRITORIES     The   health   system   in   Prince   Edward   Island   is   operated   by   Health   PEI,   a   single   subsidiary   of   the   provincial   government.     There   is   a   single   cancer   centre   on   the   island,   although   the   majority  of  relevant  cancer  surgeries  were  transferred  to  institutions  in  New  Brunswick  or   Nova  Scotia.  The  territories  also  follow  a  similar  pattern  of  referral  to  other  provinces  for   their   surgical   cancer   cases.   Surgeries   for   residents   in   Yukon   were   performed   in   either   British  Columbia  or  Alberta  for  all  five  cancers  except  for  esophageal  cancer  surgeries  that   were  performed  in  British  Columbia.  Ontario,  Manitoba  and  Alberta  undertook  surgeries  for   residents  in  Nunavut  and  Northwest  Territories.  All  four  jurisdictions  have  extremely  small   caseloads,  so  it  would  be  inappropriate  to  perform  any  comparisons.   5.10  SASKATCHEWAN       The  cancer  surgical  care  in  Saskatchewan  is  overseen  by  the  "Surgical  Initiative  Branch"  of   Saskatchewan  Ministry  of  Health.  Saskatchewan  Cancer  Agency  (SCA)  does  not  have  a  role   in  the  organization  of  oversight  of  surgical  care  in  the  province.  Our  best  efforts  could  not   capture   any   information   regarding   regionalization,   control   of   surgical   volume,   and/or   creation   of   designated   surgical   centres.   However,   there   are   reports   of   initiatives   taken   in   2009   to   improve   surgical   care   and   reduce   wait   times   for   certain   surgical   procedures,   including   cancer   surgery.   As   part   of   this   initiative,   a   specialist   directory   has   been   made   available  online  that  provides  information  on  types  of  surgical  procedure,  the  case  volume,   and   wait   times   for   last   12   months   at   each   of   the   provider   sites   in   their   respective   health   regions.   This   is   an   example   of   the   initiation   of   transparency   of   information   policies   that   are   aimed  at  improving  system  performance.   The  annual  case  volume  for  four  of  the  five  cancer  types  seems  to  be  decreasing,  and  along   with  it,  the  number  of  facilities  providing  surgical  care  (Table  5.8).  For  instance,  the  number   of  institutions  providing  liver  cancer  surgery  dropped  from  4  centres  to  2,  and  from  3  to  2   for  lung  resections  from  2004  to  2012.  These  institutions  were  concentrated  in  Saskatoon   and   Regina.   It   was   unclear   if   the   reduction   in   the   number   of   institutions   was   related   to   random   variations   associated   with   the   small   number   of   cases,   or   an   intent   to   consolidate   these   services.   Other   institutions   either   had   very   low   volume   or   there   is   an   evidence   of   consolidation.   Still,   there   is   more   room   for   institutions   in   Saskatchewan   to   switch   to   one       P a g e    105     centre  as  illustrated  by  the  case  volume  for  ovarian  procedures  being  less  than  half  of  lung   resections,   yet   nine   institutions   performed   such   operations   as   compared   to   two   for   lung   cancer.   A   majority   of   institutions   had   a   case   volume   less   than   five   per   year.   Interestingly,   the   3-­‐year   average   mortality   rates   were   higher   than   the   national   average   for   all   the   procedures,   except   pancreatic   cancer.   The   median   length   of   stay   over   the   same   time   period   was   higher   than   national   average   for   all   procedures,   however.   Higher   travel   times   for   esophagus   and   pancreatic   procedures   might   be   a   result   of   consolidation   of   such   services   to   a   few   centres,   but   same   was   not   true   for   liver   cancer   resections.   A   large   number   of   institutions   for   ovarian   resections   could   explain   shorter   travel   distances   for   these   procedures  in  Saskatchewan.   TABLE  5.8:  STATE  OF  HIGH-­‐RISK  SURGERIES  IN  SASKATCHEWAN     #  Surgeries  in  2012   (Average  2010-­‐12)   #  Hospitals  in  2012   (#  with  Case  Volume   >5)   Mean  Hospital  Case   Volume  in  2012   (range)   %  Mortality  Rate   (2010-­‐12)   %  Mortality  Rate   National  Average   (2010-­‐12)   (Standard  Error)   Median  LOS  [IQR]   (2010-­‐12)   Median  LOS   National  Average   [IQR]  (2010-­‐12)   Median  Travel   Distance  (2010-­‐12)   Kilometers  [IQR]   %  of  patient  in  1-­‐ hour  travel  time   (2010-­‐12)               Esophagus   12   (15)   Pancreas   11   (12)   Liver   20   (24)   Lung   150   (135)   Ovarian   69   (84)   2   (1)   3   (0)   2   (1)   2   (2)   9   (3)   6   (1-­‐11)   4   (2-­‐5)   10   (2-­‐18)   75   (53-­‐97)   8   (1-­‐33)   6.67   2.70   2.78   1.97   0.79   4.74  (0.68)     2.90  (0.42)       2.63  (0.27)       1.63  (0.12)       0.74  (0.11)     14.5  [9]   14  [12.5]   9  [7]   6  [4]   5  [5]     13  [10]     10  [9]     7  [5]     5  [4]     4  [2]   162  [229]   155  [204]   72  [216]   73  [186]   50  [161]   29   35   50   46   52   P a g e    106     6.0:  CITIZEN  ENGAGEMENT   CHAPTER  HIGHLIGHTS     " Regionalization  of  high-­‐risk,  complex  cancer  services  is  fraught  with  potential  impacts  to  patients   and   caregivers;   therefore,   it   is   necessary   to   consider   the   thoughts   of   these   potential   users   of   the   system.         Principal Findings " The   citizen   panels   highlighted   that   the   regionalization   of   complex   cancer   surgeries   into   regionalized,  high-­‐volume  centres  may  be  the  most  promising  approach  to  improving  outcomes   related  to  the  delivery  of  care.  Indeed,  the  number  of  panel  members  who  ranked  regionalization   as   the   most   important   option   increased   consistently   across   all   three   panels   from   pre-­‐   to   post-­‐ deliberations,   which   suggests   an   increased   strength   in   their   preference   for   this   option   as   a   result   of  developing  a  clearer  understanding  of  regionalization  through  discussions.  Figure  6.0  visually   demonstrates  this  trend.   Regionalize  complex  cancer   surgeries  into  designated   surgical  centres  of  excellence   90%   80%   70%   57%   62%   60%   64%   79%   83%   82%   50%   40%   29%   27%   20%   Alberta   14%   21%   Post-­‐panel   17%   Pre-­‐panel   0%   23%   Post-­‐panel   10%   Ontario   9%   9%   9%   Post-­‐panel   15%   Pre-­‐panel   30%   Pre-­‐panel   Percentage  of  participants   100%   P.E.I.   Implement  province-­‐wide   quality  improvement  initiatives   to  improve  the  delivery  of   complex  cancer  surgeries  where   they  are  now  being  provided   Encourage  the  local  adoption  of   quality  improvement  initiatives   to  improve  the  delivery  of   complex  cancer  surgeries  where   they  are  now  being  provided   Figure  6.0:  What  is  the  most  favoured  option  to  improve  the  delivery  of  complex  cancer  surgeries  in   Canada?   "       Panel   deliberations   revealed   that   in   an   optimal   situation,   patients   would   receive   regionalized   surgical  cancer  care  in  centres  of  excellence,  yet  complete  pre-­‐  and  post-­‐  operative  investigations   closer   to   home   in   either   local   centres   or   via   telemedicine   programs   to   allow   for   access   with   highly  trained  expertise  while  minimizing  the  burden  of  travel.     P a g e    107     It  is  important  to  emphasize  that,  despite  this  general  preference  towards  regionalization,  there   will   always   be   a   small   proportion   of   patients   (and   families)   who   will   favour   locally-­‐delivered   treatment   options.   Some   of   the   discussed   concerns   included   added   burden   of   travel,   a   loss   of   family  and  community  support  due  to  increased  distances,  and  a  potential  shift  of  pre-­‐and  post-­‐ operative   care   away   from   the   community.   This   situation   illustrates   the   importance   of   support   infrastructure,   as   well   as   travel   distances,   in   influencing   decisions   regarding   treatment   options   and  where  to  obtain  care.   " Certain  institutional  level  factors  were  also  identified  that  could  hinder  the  efforts  to  regionalize   surgical   care.   These   include   the   loss   of   local   expertise,   polarization   of   surgical   care   structure   within   a   province,   difficulties   in   changing   provider   behaviour,   and   complexities   associated   with   developing  unanimous  guidelines  and  standards.   " In   consideration   of   aforementioned   patient   and   system-­‐level   challenges,   it   is   strongly   recommended   that   policy   makers   and   health   care   planners   be   prepared   to   address   various   stakeholder  perspectives  when  developing  policies  for  implementing  regionalization  of  complex   cancer  surgical  services.     Strengths " The   best   available   research   evidence   was   paired   with   a   robust   deliberative   process   that   provided   a  voice  to  the  real-­‐world  views  and  experiences  of  a  diverse  group  of  citizens.     " Panel   deliberations   identified   of   a   set   of   core   values,   and   a   comparison   of   three   options   (ranging   from  quality-­‐improvement  strategies  to  structural  changes)  that  can  guide  the  efforts  of  health-­‐ system  leaders  in  improving  the  delivery  of  complex  cancer  surgeries.     " Convening  the  citizen  panels  in  three  provinces  allowed  an  examination  of  views  and  experiences   of   citizens   residing   in   provinces   with   different   degrees   of   consolidation   for   complex   cancer   surgeries  and  access  to  surgical  centres  of  excellence.     Limitations " The   first   limitation   of   our   qualitative   analysis   is   that   the   findings   remain   exploratory   in   nature.   They  are  drawn  from  a  small  number  of  sites,  which  may  not  reflect  the  views  and  experiences  of   citizens  across  all  Canadian  jurisdictions.     " The   second   limitation   is   related   to   the   challenges   of   engaging   citizens   in   in-­‐depth   deliberations   over  a  short  period  of  time.  This  requires  intense  efforts  from  panel  members  to  learn  from  the   evidence  and  from  others’  views  and  experiences,  and  share  their  newly  informed  views  about   the  issue  and  how  to  address  it.  We  addressed  this  limitation  to  some  extent  by  not  aiming  for   consensus  among  panel  members,  but  instead  to  gather  a  range  of  perspectives  on  the  topic.     Implications for policy " Regionalization  policies  were  generally  supported  by  citizens  in  the  sampled  jurisdictions.  While   they  were  more  concerned  with  ensuring  that  they  received  the  highest  quality  of  care  possible   despite  a  potential  restriction  of  local  care,  the  citizens  preferred  the  provision  of  local  services   when   possible.   Therefore,   policies   should   be   implemented   with   an   eye   for   integrating   care   pathways   that   utilize   local   pre-­‐   and   post-­‐operative   care   where   appropriate   and/or   high-­‐quality   telemedicine  systems  to  increase  access  to  care.     "     P a g e    108     6.1  OBJECTIVES   Quality-­‐improvement  strategies  that  focus  on  the  process  of  care  and  structural  changes  to   health   systems   may   have   important   implications   for   the   experiences   of   patients   and   their   families.  One  approach  to  explore  such  issues  is  to  convene  a  citizen  panel  bringing  together   10-­‐14   citizens   from   all   walks   of   life.   Panel   members,   informed   by   a   pre-­‐circulated   citizen   brief,  share  their  ideas  and  experiences  on  an  issue,  and  learn  from  research  evidence  and   from   the   views   of   others.   The   deliberations   of   a   citizen   panel   can   reveal   new   understandings  about  an  issue  and  spark  insights  about  how  it  should  be  addressed.     To   foster   such   efforts,   the   McMaster   Health   Forum   (www.mcmasterhealthforum.org)   convened  a  citizen  panel  in  each  of  three  Canadian  jurisdictions  to  deliberate  about  how  to   improve   the   delivery   of   complex   cancer   surgeries.   In   this   section,   we   highlight   the   views   and   experiences   of   panel   members   about   the   problem,   three   options   (among   many)   to   address   the   problem,   and   barriers   and   facilitators   to   implementing   the   options.   In   particular,   we   examined   areas   of   common   ground   and   differences   of   opinions   among   panel   members  and  (where  possible)  identify  the  values  underlying  different  positions.  The  study   protocol   was   approved   by   the   Hamilton   Integrated   Research   Ethics   Board   at   McMaster   University  and  all  panel  members  provided  voluntary,  informed  consent.  Further  details  on   participant  recruitment  are  presented  in  Appendix  6.   The   main   objective   of   this   section   is   to   provide   insights   into   the   citizen’s   preferences   in   organization   of   cancer   surgical   care,   along   with   the   debate   addressing   the   potential   challenges   in   access   to   quality   care.   This   information   can   inform   policymakers   and   health   care   administrators   regarding   citizen’s   perspective   on   access   to   care,   travel   challenges,   and   potential  approaches  to  improve  quality  of  cancer  surgical  care.  The  features  of  the  problem   and   a   summary   of   the   factors   contributing   to   the   problem   are   outlined   in   Appendix   7,   as   well  as  the  three  options  are  summarized  in  Appendix  8.   6.2  DATA  COLLECTION  AND  ANALYSIS   An  overarching  qualitative  approach  was  used  to  synthesize  the  key  themes  that  emerged   from  the  face-­‐to-­‐face  deliberations.  This  approach  was  combined  with  pre-­‐  and  post-­‐event   questionnaires   to   document   participants’   views   regarding   different   facets   of   the   problem,   as  well  as  three  proposed  options)  to  improve  the  delivery  of  complex  cancer  surgeries.  The   deliberations   were   recorded   and   non-­‐participant   observers   took   extensive   notes.   The   recordings   and   notes   were   used   to   draft   summaries   for   each   panel   that   highlighted   the   views   and   experiences   of   panel   members,   along   with   values-­‐related   themes   that   emerged   during   the   deliberations.   Pre-­‐   and   post-­‐meeting   questionnaires   were   developed   by   the   research   team   to   assess   the   relative   importance,   assigned   by   panel   members   before   and   after   the   deliberations,   to   the   three   options   for   addressing   the   problem   of   shortfalls   with   the   delivery   of   complex   cancer   surgeries   in   Canada.   Questions   were   also   included   about       P a g e    109     participants’  preferred  location  for  undergoing  complex  cancer  surgery,  the  importance  of   travel  distance  to  their  decision  to  have  the  surgery  or  not,  and  concerns  about  having  to   travel  a  long  distance  to  undergo  complex  cancer  surgery.  Descriptive  statistics  were  used   to   analyze   the   data   collected   from   the   questionnaires.   The   participant   profiles   are   summarized  in  Appendix  9.     6.3  MAIN  FINDINGS     The  summary  of  the  key  themes  that  emerged  during  the  deliberations  of  the  three  citizen   panels  are  presented  below,  with  a  specific  focus  on  their  views  and  experiences  regarding:   1)  the  problem;  2)  three  potentially  viable  options  (among  many)  to  address  the  problem;   and   3)   the   implementation   considerations.   The   section   concludes   with   an   examination   of   results  from  the  pre-­‐  and  post-­‐questionnaires.     The problem – why is it challenging to improve the delivery of complex cancer surgeries? All  three  panels  discussed  at  great  length  the  challenges  of  making  an  informed  decision  in   the   midst   of   a   cancer   diagnosis.   The   panel   members   highlighted   six   key   challenges   with   some   consistency,   which   represented   deeper   reflections   on   the   features   of   the   problem   presented  in  the  brief  based  on  their  experiences  and  values  (see  Table  6.1  below).   TABLE  6.1:  KEY  FEATURES  OF  THE  PROBLEM  OF  DELIVERING  COMPLEX  CANCER  SURGERIES   (AND  ITS  CAUSES)  ACCORDING  TO  PANEL  MEMBERS   Panel  members  highlighted  six  key  challenges  with  some  consistency:   1. Making  decisions  in  the  midst  of  a  cancer  diagnosis  is  difficult;   2. Inequities  exist  in  access  to  optimal  surgical  care,  as  well  as  palliative  care  and  system-­‐level   support;   3. Patients  and  their  informal/family  caregivers  lack  support;   4. The  cancer  patient  journey  is  marked  by  care  coordination  problems  (e.g.,  communication   breakdowns  with  and  between  healthcare  providers,  and  out-­‐of-­‐province  care  processes   being  complex  and  not  optimally  coordinated);   5. Current  financial  arrangements  (i.e.,  global  budgets  for  hospitals)  limit  our  capacity  to   improve  the  delivery  of  complex  cancer  surgeries;  and   6. Governance  arrangements  (e.g.,  lack  of  regulation  for  hospitals  and  surgeons,  and  lack  of   arrangements  to  optimally  address  cross-­‐jurisdictional  issues)  limit  our  capacity  to  improve   the  delivery  of  complex  cancer  surgeries.   1) It  is  difficult  to  make  an  informed  decision  in  the  midst  of  a  cancer  diagnosis     Participants   generally   agreed   that   a   cancer   diagnosis   triggers   very   emotional   reactions,   which  affect  the  capacity  of  patients  and  families  to  make  decisions.  They  also  pointed  out   the  rapid  pace  at  which  decisions  must  be  made,  often  without  complete  information  (e.g.,   information  about  their  condition,  the  full  scope  of  treatment  options,  and  the  availability  of       P a g e    110     high-­‐quality   surgical   and   post-­‐operative   care)   and   without   the   support   needed   to   make   these   decisions,   which   nurtured   uncertainty   about   which   course   of   action   to   take.   As   one   Ontario   panel   member   said:   “[you   must   make]   rushed   decisions   that   could   mean   life   and   death.”   Participants   also   emphasized   that   such   decisions   are   made   more   challenging   for   patients   and   families   from   rural   and   remote   areas   (as   well   as   those   from   provinces   where   such   surgeries   are   not   performed)   because   they   have   to   weigh   the   significant   emotional,  financial  and  practical  implications  of  traveling  far  away  from  home  to  undergo   lengthy  and  complex  surgical  procedures  against  the  potential  benefits.   2) Inequities   exist   in   access   to   optimal   surgical   care,   palliative   care,   and   other   system-­‐level  support     All   three   panels   discussed   inequities   of   access   to   optimal   surgical   care,   and   in   some   instances   inequities   of   access   to   palliative   care,   and   other   system-­‐level   support.   Several   participants  from  the  Ontario  and  Alberta  panels  acknowledged  that  they  were  lucky  to  live   near   a   regional   centre   of   excellence   for   cancer   (in   Hamilton   or   Edmonton),   but   they   also   agreed  that  many  people  living  in  rural  and  remote  areas  could  face  very  difficult  decisions.   Some  examples  included  decision  regarding  whether  to  choose  a  treatment  option  offered   at   a   hospital   close   to   home   (and   potentially   undergoing   surgery   in   a   local,   low-­‐volume   hospital),  or  travelling  far  away  from  home  to  receive  the  surgery  at  a  high-­‐volume  hospital   with   a   concentration   of   expertise   (with   the   hope   of   better   outcomes).   A   few   participants   pointed   out   that   these   inequities   had   been   an   ongoing   issue   given   the   vast   Canadian   landscape  and  its  low-­‐density  population,  which  require  that  many  patients  travel  to  obtain   specialized   care   only   available   in   urban   centres.   However,   they   emphasized   that   it   was   essential   to   collectively   find   ways   to   overcome   all   the   barriers   that   may   restrict   access   to   care  and  to  alleviate  the  burden  on  patients  and  families  who  must  travel.   3) Patients  and  their  informal/family  caregivers  lack  support     Several   panel   members   indicated   that   patients   and   their   informal/family   caregivers   lack   support.   They   emphasized   the   crucial   role   that   caregivers   play   along   the   cancer   journey.   As   one   panel   member   from   Alberta   indicated:   “It   is   a   daunting   journey.   If   I   didn’t   have   people   helping   me,   I   would   say   forget   it   [undergoing   complex   cancer   surgery].”   A   second  panel  member  from  Alberta  emphasized  that:  “As  Canadians,  we  deserve  the  right   to   have   the   proper   support   system   in   place   when   going   through   cancer.”   This   issue   was   also   discussed   at   great   length   among   PEI   panel   members.   Given   the   need   to   travel   long   distances  for  specialist  care,  and  the  lack  of  supports  on  the  island  for  both  pre-­‐  and  post-­‐ operative  care,  several  panel  members  noted  that  it  was  the  informal  caregivers  who  played   the   most   significant   support   role.   Yet,   many   informal   and   family   caregivers   faced   many   challenges  in  their  roles,  particularly  those  with  lower  incomes  and  those  living  in  remote   and  northern  communities  who  must  travel  to  urban  centres  to  accompany  their  loved  ones   undergoing  complex  cancer  surgeries.  Participants  from  all  three  panels  called  for  greater       P a g e    111     financial   support   to   alleviate   the   burden   on   caregivers,   who   can   face   significant   out-­‐of-­‐ pocket   expenses,   as   well   as   greater   practical   support,   including   accommodations   like   the   Ronald  McDonald  Houses  that  provide  a  ‘home  away  from  home.’   4) The  cancer  patient  journey  is  marked  by  care  coordination  problems       Several   panel   members   emphasized   that   the   cancer   patient   journey   is   marked   by   care   coordination   problems.   For   instance,   a   few   Ontario   panel   members   experienced   difficult   interactions  with  their  family  physicians  and  specialists  about  their  initial  symptoms,  which   may   have   delayed   their   cancer   diagnosis.   Other   Ontario   panel   members   provided   examples   of   communication   lapses   between   the   regional   cancer   centre   and   their   local   Community   Care  Access  Centre  (i.e.,  local  agencies  that  have  responsibility  for  connecting  people  with   the  care  they  need  at  home  and  in  their  communities),  which  affected  the  coordination  of   post-­‐operative   care.   As   one   Ontario   panel   member   indicated:   “When   patients   leave   the   hospital,   there   is   a   risk   of   getting   into   a   void.”   Such   coordination   problems   were   particularly  salient  among  PEI  panel  members,  who  highlighted  that  having  to  travel  out-­‐of-­‐ province   to   undergo   complex   cancer   surgery   can   complicate   the   process   of   care,   which   is   then   often   poorly   coordinated.   In   addition,   they   emphasized   that   that   most   residents   of   PEI   were  “blind  to  the  way  the  system  works  [on  the  island].”  Thus,  they  face  challenges  in   getting   information   related   to   how   the   system   works,   their   likely   care   pathway,   and   the   supports  available  to  them.     5) Current  financial  arrangements  may  limit  our  capacity  to  improve  the  delivery   of  complex  cancer  surgeries     All   three   panels   talked   to   a   lesser   extent   about   how   current   financial   arrangements   may   limit   our   capacity   to   improve   the   delivery   of   complex   cancer   surgeries.   While   some   participants   were   generally   concerned   about   the   rising   costs   associated   with   cancer   care   and  the  overall  financial  sustainability  of  the  health  system,  others  expressed  concern  that   the  predominant  funding  model  for  hospitals  (i.e.,  global  budgets)  may  not  create  incentives   for   hospitals   to   improve   the   delivery   of   complex   cancer   surgeries,   or   to   increase   surgical   volumes   for   certain   complex   cancer   surgeries.   PEI   panel   members   were   particularly   worried   that   global   hospital   budgets   may   create   disincentives   for   providers   in   other   provinces   to   take   on   patients   requiring   complex   cancer   surgeries   from   their   province,   as   they  could  be  more  expensive  to  treat.     6) Existing   governance   arrangements   may   limit   our   capacity   to   improve   the   delivery  of  complex  cancer  surgeries     All   three   panels   also   talked   (albeit   to   a   lesser   extent)   about   how   current   governance   arrangements  may  limit  our  capacity  to  improve  the  delivery  of  complex  cancer  surgeries.   Ontario  and  Alberta  panel  members  both  expressed  concern  about  the  minimal  regulation   regarding   which   procedures   surgeons   can   perform   within   their   specialty   area,   or   how       P a g e    112     frequently  they  need  to  perform  these  procedures  to  ensure  their  surgical  skills  remain  up   to   date.   They   were   similarly   concerned   about   the   lack   of   regulation   about   which   surgical   procedures  hospitals  can  perform  or  how  frequently  they  need  to  perform  them  to  ensure   that  quality  remains  high.  One  panel  member  from  Alberta  indicated  that  there  was  a  need   to   raise   public   awareness   about   this.   “That   scares   me.   It   is   very   scary   and   Canadians   need   to   pay   attention   to   that.”   Other   participants   were   also   concerned   that   ‘solo’   surgeons   could   be   allowed   to   perform   these   high-­‐risk   and   resource-­‐intensive   cancer   surgeries  in  low-­‐volume  hospitals.  PEI  panel  members  focused,  for  instance,  on  the  absence   of   explicit   mechanisms   to   ensure   that   the   patients   referred   out   of   province   receive   the   highest  quality  care  possible  in  a  timely  way.  PEI  panel  members  also  argued  that  one  of  the   main  issues  underpinning  these  challenges  was  the  complex  political  context.  In  particular,   the   dynamics   of   Canada’s   federalist   structure   of   government   and   interprovincial   politics   were   thought   to   complicate   the   situation,   blurring   lines   of   accountability   and   making   it   challenging  to  pursue  changes  to  the  status  quo  in  a  consolidated  and  coordinated  way.   Participants   were   asked   to   rank   each   of   the   options   based   on   preferences   after   considering   their   personal   beliefs   and   values.  Figure   6.1   summarizes   the   values   associated   with   each   of   the   options.   Regionalizing   complex   cancer   surgeries   into   designated   surgical   centres   of   excellence   was   ranked   number   one   by   a   vast   majority   of   participants   in   all   three   panels,   both   before   and   after   the   deliberations.   This   number   one   ranking   increased   consistently   across   all   three   panels   over   the   pre-­‐   and   post-­‐panel   period,   which   indicates   that   as   participants  learned  more  about  regionalization,  their  strength  of  preference  increased  as  a   result.   The   implementation   of   province-­‐wide   quality   improvement   initiatives   to   improve   the   delivery   of   complex   cancer   surgeries   where   they   are   now   being   provided   was   most   often   ranked   second   in   the   Ontario   and   Edmonton   panels   (both   before   and   after   deliberations),   but   PEI   participants   ranked   this   option   third   before   the   deliberations   and   then   second   after   the   deliberations.   Lastly,   the   concept   of   encouraging   the   local   adoption   of   quality   improvement   initiatives   in   order   to   improve   the   delivery   of   complex   cancer   surgeries  where  they  are  now  being  provided  was  most  often  ranked  third  in  the  Ontario   and   Edmonton   panels   (both   before   and   after   deliberations),   but   PEI   participants   ranked   this  option  second  before  the  deliberations  and  then  third  after  the  deliberations.  It  is  quite   likely   that   this   decrease   in   the   strength   of   preference   for   this   option   is   as   a   result   of   the   education  process  facilitated  by  the  deliberations.   The options – how can we address the problem? After   discussing   the   challenges   that   reflect   or   contribute   to   shortfalls   in   the   delivery   of   complex  cancer  surgeries  in  Canada,  participants  discussed  three  options  (among  many)  for   making   improvements   (see   Appendix   8   for   a   detailed   description   of   each   option).   In   this   section,  we  briefly  describe  the  views  of  panel  members  regarding  each  option  and  (where       P a g e    113   possible)  identify  the  values  underlying  different  positions.  Table  6.2  presents  a  summary   of  the  value-­‐related  themes  that  emerged  for  each  option.   Option   1   -­‐   Encourage   the   local   adoption   of   quality-­‐improvement   initiatives   to   improve  the  delivery  of  complex  cancer  surgeries  where  they  are  now  being  provided   While   participants   from   all   three   panels   acknowledged   that   this   option   was   aimed   to   continuously  improve  the  local  delivery  of  complex  cancer  surgeries,  and  potentially  local   care  processes,  most  of  them  agreed  that  the  inherent  challenges  and  drawbacks  associated   with   this   option   made   it   the   least   promising   among   the   three   considered   to   improve   the   delivery   of   complex   cancer   surgeries.   When   discussing   the   limitations   of   option   1,   three   values-­‐related  themes  emerged  with  some  consistency:   • • • Stewardship  -­‐  All  three  panels  indicated  that  this  option  may  lack  the  clear  direction and   incentives   needed   to   implement   quality-­‐improvement   initiatives.   Most   panel members   saw   this   option   as   a   ‘weak’   approach   to   improving   the   delivery   of   complex cancer   surgeries.   As   a   panel   member   from   Ontario   illustrated:   “It’s   like   telling   a child,  ‘please  be  good’.”  A  panel  member  from  Edmonton  emphasized  that  without proper   stewardship,   local   quality-­‐improvement   initiatives   will   be   “moving   targets that  could  be  easily  side-­‐tracked.” Policies   based   on   data   and   evidence   -­‐   The   Alberta   and   Ontario   panels   pointed   out that  this  option  may  lead  to  policies  that  are  not  aligned  with  what  is  known  about the  relationship  between  surgical  volumes  and  outcomes.  An  Ontario  panel  member indicated   that   trying   to   provide   full   access   to   these   complex   cancer   surgeries   in   all hospitals  was  impossible.  “Full  access  and  fairness,  it’s  a  utopia.  Until  we  reach population  density  allowing  that,  [it  will  not  be  possible].” Excellent   health   outcomes   -­‐   The   Alberta   and   Ontario   panels   suggested   that   this option   may   improve   the   local   delivery   of   care,   but   not   necessarily   improve   patient outcomes.   Ontario   panel   members   emphasized   the   lack   of   available   expertise   for these   complex   cancer   surgeries.   As   one   panel   member   pointed   out:   “There   is   a   lack of   expertise   to   go   around.”   Thus,   implementing   this   option   does   not   solve   the problem   of   not   having   enough   surgeons   and   health   care   providers   available   to achieve  excellent  health  outcomes  in  all  hospitals. A  few  other  values-­‐related   themes   emerged   in   some   panels   that   may   explain   the   reluctance   of  participants  to  support  this  option:   • Choice  -­‐  The  Ontario  panel  pointed  out  that  while  this  option  may  allow  patients  to choose   local   treatment   options,   it   may   not   allow   them   to   choose   the   most   optimal treatment  option  (i.e.,  going  to  a  high-­‐volume  hospital). P a g e    114     • • • Cost-­‐effectiveness  -­‐  The  Ontario  panel  indicated  that  this  option  may  not  make  the   best  use  of  the  limited  financial  resources  and  expertise  available.   Collaboration  -­‐  The  Ontario  panel  highlighted  that  this  option,  by  focusing  efforts  on   local  hospitals,  may  increase  fragmentation  within  the  system.   Sustainability   -­‐   The   PEI   panel   emphasized   that   there   may   be   a   lack   the   resources   in   small   provinces   like   theirs   to   implement   and   sustain   locally   driven   quality-­‐ improvement  initiatives.   Option  2   -­‐  Implement  province-­‐wide  quality-­‐improvement  initiatives  to  improve  the   delivery  of  complex  cancer  surgeries  where  they  are  now  being  provided   Participants   in   all   three   panels   generally   agreed   that   this   option   was   preferred   to   the   previous   one,   although   significant   challenges   still   existed.   Several   values-­‐related   themes   highlighted  the  potential  benefits  of  province-­‐wide  quality-­‐improvement  initiatives,  which   include  but  are  not  limited  to:   • • • • • Continuously   improve   -­‐   This   option   may   support   province-­‐wide   continuous-­‐ improvement   efforts,   facilitate   the   dissemination   and   uptake   of   successful   local   initiatives   across   the   province,   but   also   facilitate   patient   coordination   across   jurisdictions.   Stewardship   -­‐   This   may   provide   the   necessary   direction   and   incentives   to   implement  these  local  quality-­‐improvement  initiatives.   Accountability  -­‐  This  option  may  increase  public  accountability  by  requiring  public   reporting  of  surgeons’  and  hospitals’  performance,  and  establishing  clear  provincial   benchmarks.   Fairness  -­‐  This  option  may  standardize  the  quality  of  care  across  the  province,  and   could   be   extended   to   include   the   development   of   pan-­‐Canadian   guidelines   and   standards  to  ensure  that  all  Canadians  have  access  to  optimal  surgical  care.   Excellent   patient   and   family   experience   –   This   option   may   ensure   that   care   pathways   are   more   attuned   to   the   needs   of   patients   and   families   who   must   travel   outside  the  province  to  undergo  surgery.   However,   when   discussing   the   limitations   of   this   option,   two   values-­‐related   themes   emerged  with  some  consistency:   •     Fairness   -­‐   Some   elements   of   this   option,   more   specifically   implementing   pay-­‐for-­‐ performance  for  hospitals  and  establishing  requirements  for  reporting  to  the  public   about   quality   indicators   and   other   performance   measures,   may   have   some   unintended   consequences.   Several   participants   expressed   concerns   that   such   quality-­‐improvement   initiatives   could   lead   providers   and   hospitals   to   ‘cherry-­‐pick’   patients   who   may   help   them   score   well,   or   to   avoid   those   who   may   cause   them   to   P a g e    115     • score   poorly,   in   order   to   make   their   statistics   look   better.   As   one   panel   member   from   Alberta  said:  “I  have  concerns  that  [they  wouldn’t]  want  to  take  people  like  me   because  the  success  rate  is  not  there.  My  chances  of  my  survival  are  not  high.”   Excellent   health   outcomes   -­‐   By   not   making   structural   changes,   the   impact   of   this   option  would  most  likely  be  limited  in  improving  patient  outcomes.     Option  3  –  Regionalize  complex  cancer  surgeries  into  designated  surgical  centres  of     excellence   Overall,  this  option  generally  resonated  the  most  with  participants  in  all  three  panels  and   was   identified   as   the   one   more   likely   to   significantly   improve   the   delivery   of   complex   cancer   surgeries.   Several   values-­‐related   themes   highlighted   the   potential   benefits   of   regionalization,  which  include  but  are  not  limited  to:   • • • • • • • Continuously   improve   -­‐   This   option   may   yield   continuous   improvements   in   the   quality  of  both  surgical  and  post-­‐operative  care.   Excellent   health   outcomes   -­‐   This   option   is   more   likely   to   improve   patient   outcomes.   Policies   based   on   data   and   evidence   -­‐   This   option   may   lead   to   policies   that   are   aligned   with   what   is   known   about   the   relationship   between   surgical   volumes   and   outcomes.     Cost-­‐effectiveness  -­‐  This  option  could  make,  as  a  panel  member  from  Alberta  said,   the  most  “cost-­‐effective  use  of  trained  personnel.”   Expertise   -­‐   This   option   may   create   a   concentration   of   highly   skilled   surgeons   and   health   care   providers   to   deliver   these   very   complex   and   high-­‐risk   surgical   procedures.   Safety  -­‐  This  option  may  offer  safer  surgical  care  by  having  a  critical  mass  of  highly-­‐ trained  personnel  who  can  support  complex  cancer  surgeries.   Innovation   -­‐   This   option   offers   an   environment   more   likely   to   facilitate   or   trigger   innovation.   While   regionalization   resonated   the   most   with   panel   members,   it   still   generated   some   concerns  that  were  embodied  in  the  following  values-­‐related  themes:   •     Excellent   patient-­‐   and   family-­‐experience   –   Regionalizing   complex   cancer   surgeries   should   not   neglect   that   care   must   be   attuned   to   the   values,   needs   and   preferences   of   patients   and   families.   As   a   panel   member   from   Alberta   said:   “[regionalization]   needs   to   be   looked   at   in   a   holistic   way,   with   the   family   taken   into   consideration.”   In   addition,   regionalization   must   be   attuned   to   the   unique   aspects  of  different  cancer  care  pathways.  PEI  panel  members  highlighted  that  not  all   cancers  are  the  same,  meaning  that  regionalization  across  the  full  spectrum  of  care   P a g e    116     • • • (pre-­‐operative   care   to   surgical   care   to   post-­‐operative   care)   should   not   be   considered   a  ‘one-­‐size  fits  all’  approach.   Fairness   –   Regionalizing   complex   cancer   surgeries   should   not   create   additional   barriers   to   accessing   optimal   surgical   care   and   should   include   interventions   to   mitigate  the  negative  consequences  of  traveling  to  undergo  complex  cancer  surgery   in  a  regional  surgical  centre  of  excellence.   Adaptability  –  The  health  system  should  provide  flexible  care  pathways  that  allow   every  patient  to  receive  optimal  surgical  care  in  a  regional  centre  of  excellence,  with   the  remaining  care  provided  locally.   Proximity  –  While  the  regionalization  of  complex  cancer  surgeries  may  be  the  most   optimal  way  forward,  panel  members  still  greatly  valued  receiving  cancer  care  close   to  home  whenever  possible.   These   concerns   led   several   panel   members   to   propose   enhancements   to   the   third   option.   More  specifically,  PEI  panel  members  proposed  three  additional  features:   • Implementing   local   post-­‐care   recovery   centres   –   PEI   panel   members   emphasized   the   importance   of   balancing   regionalized   surgical   care   with   strong,   local   post-­‐operative   care.   These   panel   members   acknowledged   that   the   island   (along   with   other   regions   across  Canada)  may  not  have  the  capacity  to  implement  surgical  centres  of  excellence,   but  they  proposed  the  implementation  of  local  post-­‐care  recovery  centres  of  excellence.   • Introducing  telemedicine  initiatives  –  PEI  panel  members  called  for  effective  use  of   advances   in   telemedicine   or   other   technologies   that   enable   remote   consultations   and   thus  reduce  the  burden  associated  with  travelling  to  a  regional  centre  of  excellence  for   routine  consultations.   • Increasing  the  role  of  ‘patient  navigators’  –  PEI  panel  members  called  for  a  greater   role  of  ‘patient  navigators’  who  can  provide  support  and  coordination  for  patients  and   their  informal/family  caregivers  throughout  the  entire  care  pathway  (especially  in  the   context  of  those  who  must  travel  to  undergo  complex  cancer  surgery).       Ontario   panel   members   went   further   and   proposed   a   fourth,   stand-­‐alone   option:   introducing   flexible   care   pathways,   whereby   every   patient   could   receive   optimal   surgical   care   in   a   regional   centre   of   excellence,   but   with   the   remaining   care   provided   close   to   home   whenever  possible.  This  option  resonated  with  the  proposal  from  PEI  panel  members  who   emphasized  the  importance  of  balancing  regionalized  surgical  care  with  strong,  local  post-­‐ operative  care.  Indeed,  several  Ontario  panel  members  were  hesitant  about  the  full-­‐blown   regionalization  of  cancer  care,  especially  given  the  significant  burden  that  it  would  put  on   the   shoulders   of   patients   and   families   from   rural   and   remote   communities.   These   participants  indicated  that  regionalization  might  be  a  good  option  for  the  surgeries,  which   have   a   high   risk   of   complications,   are   resource   intensive,   and   require   highly   skilled       P a g e    117     surgeons   and   providers.   However,   they   questioned   whether   all   aspects   of   cancer   care   needed  to  be  regionalized  (e.g.,  chemotherapy,  radiotherapy,  and  ancillary  cancer  care)  and   proposed   the   development   of   more   flexible   care   pathways,   with   the   surgery   being   regionalized,   but   other   cancer   care   being   offered   locally   whenever   possible.   This   would   ensure  that  patients  and  families  are  close  to  home  for  as  long  as  possible  during  the  cancer   journey,  which  was  perceived  as  an  environment  more  favourable  to  recovery.     The implementation considerations- What are concerns with regard to regionalizing care? After   discussing   the   options   for   improving   the   delivery   of   complex   cancer   surgeries   in   Canada,  participants  examined  potential  barriers  and  facilitators  for  moving  forward.  These   can  be  classified  as  either  institutional/government  level  considerations  or  patient-­‐impact   related  concerns.       Institutional-­‐level  concerns  of  the  implementation  of  regionalization     In  general,  five  broad  groupings  of  health  care  system  institutional  barriers  emerged  during   deliberations:     1. The   public   resistance   to   regionalization   if   it   leads   to   a   loss   of   local   expertise   or   if   the   regionalization   model   is   based   on   only   the   a   very   limited   number   of   ‘poles’   (i.e.,   centres   of  expertise)  in  each  province   2. The   difficulties   in   changing   provider   behavior   and   organizational   culture   to   embrace   quality   improvement   (and   the   long   time   required   to   assess   the   full   impact   of   these   options,  which  could  fuel  resistance  towards  such  quality-­‐improvement  efforts)   3. The   lack   of   human   and   financial   resources   to   achieve   desired   changes   in   the   system   (e.g.,  developing  a  critical  mass  of  highly  skilled  surgeons  and  health  care  providers  to   meet   the   demands   for   complex   cancer   surgeries   or   having   the   financial   resources   to   implement  new  and  sustainable  regional  infrastructures)   4. The   complexity   of   developing   commonly   agreed   provincial   standards   and   regional   infrastructures,  which  could  face  a  ‘push  back’  from  certain  health-­‐system  stakeholders     5. The   separation   of   powers   between   provinces   that   exists   given   Canada’s   federalist   structure,  which  makes  it  difficult  to  coordinate  health  services  across  jurisdictions  (a   barrier   particularly   salient   for   those   living   in   provinces   and   territories   that   are   not   providing  complex  cancer  surgeries)   When  considering  potential  facilitators  for  overcoming  these  barriers  and  moving  forward   with   implementing   regionalization,   participants   from   all   three   panels   emphasized   the   opportunity   of   building   on   positive   things   already   happening   in   the   health   system   as   a   foundation   for   future   efforts.   In   Alberta   and   Ontario,   panel   participants   highlighted   current   efforts   to   regionalize   certain   complex   cancer   surgeries   and   to   establish   province-­‐wide   standards.  As  for  participants  in  PEI,  many  acknowledged  the  reality   was  that  it  probably     P a g e    118   TABLE  6.2.  KEY  VALUES-­‐RELATED  THEMES  THAT  EMERGED  ABOUT  EACH  OPTION   Option 1 - Encourage the local adoption of quality-improvement initiatives to improve the delivery of complex cancer surgeries where they are now being provided • May allow for continuous improvement of LOCAL care delivery Values-related themes reflecting the strengths of each option Option 2 - Implement province-wide qualityimprovement initiatives to improve the delivery of complex cancer surgeries where they are now being provided • May allow for continuous improvement of care delivery at a PROVINCIAL level via dissemination of successful initiatives and encouraging patient coordination across jurisdictions • May facilitate stewardship: Provides leadership and incentivization of locally-based quality improvement initiatives • May increase public accountability • May standardize the quality of care across the jurisdictions to promote fairness • May ensure improved patient and family experience by tailoring care pathways to the needs of the regional users Option 3 - Regionalize complex cancer surgeries into designated surgical centres of excellence • • • • • • • • • Values-related themes reflecting the limitations of or concerns with each option • • • • May not lead to better patient outcomes May not lead to the development of data-driven evidence-based policies since there is no way to observe a possible volume-outcome relationship May lack stewardship opportunities to provide large-scale leadership and incentivization of locally-based quality improvement initiatives May reduce patient and family choice to select the most optimal treatment options May not make the most cost-effective use of the limited financial resources and expertise available May promote fragmentation and discourage collaboration within the system May not be a sustainable approach due to limited available resources in smaller provinces to sustain a multitude of initiatives • • May not lead to better patient outcomes May have unintended consequences, such as hospitals ‘cherry-picking’ less complex cases thus promoting unequitable access to health care services for patients. • • • May allow for continuous improvement of care delivery on PROVINCIAL AND PANCANADIAN levels Most likely option for broadly improving patient health outcomes May facilitate the development of data-driven evidence-based policies based on observed changes in outcomes due to higher volumes May promote more cost-effective use of trained personnel May promote a concentration of expertise of highly skilled surgeons and health care providers. This may also increase the safety of surgical care given that a critical mass of trained personnel will be assembled. May facilitate innovation through close proximity of expertise May reduce the quality of the patient and family experience given that surgical care may be delivered away from home and associated caregiving and community resources May create a disconnect between the surgical care and locally-based primary physicians and adjuvant oncology treatment: To counteract this, care pathways may be broadened to include preand post-operative surgical care closer to home May create unfair barriers to accessing care due to a need to travel P a g e    119     wasn’t   feasible   to   bring   complex   cancer   surgeries   to   local   hospitals   in   PEI.   Given   this   reality,   participants   were   more   accepting   of   the   current   situation   in   which   much   of   this   care  is  regionalized  to  centres  of  excellence  in  other  Atlantic  provinces.  As  such,  efforts  to   improve  the  existing  processes  of  care  within  and  outside  of  PEI,  rather  than  make  major   structural  changes  to  the  existing  system,  were  seen  as  possible.     An  additional  barrier  not  noted  in  the  deliberations  is  that  of  surgeon  willingness  to  adopt   new  practices  or  to  modify  existing  ones  in  order  to  enact  change.  Regionalization  requires   that   surgeons   accept   the   evidence   that   higher   volume   centres   tend   to   result   in   better   outcomes,   and   to   be   willing   to   change   practice   patterns   in   order   to   benefit   their   patients.   A   designated,   well-­‐placed   opinion   leader   within   the   community   of   practice   can   champion   regionalization  policies  within  a  province  to  minimize  the  effect  of  this  barrier  to  change.           Patient-­‐level  concerns  of  the  implementation  of  regionalization   Participants   were   also   asked   to   consider   the   impact   of   regionalization   on   their   personal   experience  with  the  health  care  system  when  accessing  surgical  cancer  services.  They  were   also   asked   to   consider   whether   these   concerns   would   then   affect   their   preference   for   regionalization,  although  the  preference  for  surgical  centres  of  excellence  endured.           1)  Travel  time     Speaking  to  the  known  increase  in  travel  times  due  to  the  regionalization  of  cancer  services   as   demonstrated   in   the   mapping   presented   earlier   in   this   report,   panel   members   were   asked  to  explore  the  trade-­‐offs,  preferences  and  concerns  associated  with  travel  to  undergo   complex   cancer   surgeries   through   three   questions.   First,   panel   members   were   asked   whether  they  would  prefer  to  undergo  complex  cancer  surgery  at  a  hospital  close  to  their   home,  even  if  that  hospital  did  not  have  a  lot  of  experience  performing  the  type  of  surgery   that  they  needed  (i.e.,  it  was  a  low-­‐volume  hospital)  or  at  a  hospital  where  many  of  these   types   of   surgeries   are   performed   (i.e.,   a   high-­‐volume   hospital),   even   if   that   meant   they   would   need   to   travel   far   away   from   home   to   receive   the   surgery   and   for   related   medical   appointments  before  and  after  the  surgery.  The  majority  of  participants  in  all  three  panels   indicated   a   preference   to   have   surgery   at   a   high-­‐volume   hospital   even   if   that   meant   they   would   need   to   travel   far   away   from   home   to   receive   the   surgery   and   for   related   medical   appointments  before  and  after  the  surgery.  On  the  whole,  these  views  changed  very  little   following  the  panel  deliberations  (see  Figure  6.1)  and  resonated  with  the  general  tenor  of   discussions,  which  emphasized  participants’  willingness  to  do  everything  possible  to  access   a  high-­‐volume  centre,  even  if  it  required  travelling  to  another  region.     Although  willingness  to  travel  to  get  the  best  care  possible  was  a  dominant  theme,  at  least   one  panel  participant  (from  Ontario)  viewed  the  burden  to  be  too  great.  This  person  clearly   indicated  that,  whatever  the  cancer  diagnosis  and  treatment  options  available,  she  would       P a g e    120     always   choose   local   treatment   options   and   “hope   for   the   best”,   because   she   “couldn’t   put   that  burden  [of  travelling  to  get  surgical  care]”  on  her  family.  Thus,  personal  and  familial   circumstances  are  also  important  factors  that  will  influence  decisions  regarding  treatment   options   and   where   to   obtain   care.   This   difference   in   opinion   was   stable   across   the   pre-­‐   and   post-­‐panel  questionnaires,  suggesting  that  there  will  always  be  a  small  proportion  who  will   favour  local  delivery  despite  downsides.     We  also  asked  panel  members  how  important  the  distance  to  travel  would  be  to  making  the   decision  to  have  surgery  or  not,  on  a  scale  of  1  (very  unimportant)  to  7  (very  important).  In   general,   panel   members   from   Ontario   and   Alberta   found   that   the   travel   distance   was   slightly  more  important  than  PEI  panel  members  (see  Table  6.3).  This  result  likely  reflects   the   reality   for   PEI   residents   that   they   must   travel   to   neighbouring   provinces   to   obtain   health  care  services  not  available  locally.  These  concerns  would  manifest  more  often  after   regionalization   of   cancer   care   services,   and   should   be   taken   into   consideration   when   making  catchment  area  decisions  at  the  health  care  planning  levels.     100%   80%   70%   62%   60%   69%   77%   79%   82%   50%   90%   40%   31%   Alberta   23%   21%   18%   Ontario   10%   At  a  hospital  where  many  of  these   types  of  surgeries  are  performed   (i.e.,  a  high-­‐volume  hospital),  even  if   that  meant  I  would  need  to  travel  far   away  from  home  to  receive  the   surgery  and  for  related  medical   appointments  before  and  after  the   surgery   At  a  hospital  close  to  my  home,  even   if  that  hospital  did  not  have  a  lot  of   experience  performing  the  type  of   surgery  that  I  needed  (i.e.,  it  was  a   low-­‐volume  hospital)   Post-­‐panel   0%   Pre-­‐panel   10%   Pre-­‐panel   38%   Pre-­‐panel   20%   Post-­‐panel   30%   Post-­‐panel   Percentage  of  participants   90%   P.E.I.   Figure  6.1:  Where  would  you  prefer  to  undergo  complex  cancer  surgery  if  you  needed  it?                   P a g e    121     TABLE  6.3:  IF  YOU  NEEDED  COMPLEX  CANCER  SURGERY,  HOW  IMPORTANT  WOULD   THE  DISTANCE  THAT  YOU  WOULD  HAVE  TO  TRAVEL  BE  TO  YOUR  DECISION  TO  HAVE   THE  SURGERY  OR  NOT?   Panels   Ontario   Alberta   PEI   Pre-­‐panel   Post-­‐panel   Pre-­‐panel   Post-­‐panel   Pre-­‐panel   Post-­‐panel   Rating  on  a  scale  of  1  (very  unimportant)  to  7  (very  important)   Mean   SD   Median   Range   4.1   2.2   4.5   1-­‐7   4.4   2.4   5   2-­‐7   4   2.1   4   1-­‐7   3.8   2.0   4.5   1-­‐7   3.2   1.3   3   2-­‐5   3   1.6   3   1-­‐6     2)  Lack  of  Local  Access  to  Post-­‐operative  Care       Lastly,  we  asked  panel  members  to  identify  the  two  things  that  concerned  them  most  about   having  to  travel  a  long  distance  to  undergo  complex  cancer  surgery.  Being  able  to  receive   good  post-­‐operative  care  once  they  go  back  home  emerged  as  the  most  important  concerns   in   all   three   panels   (both   pre-­‐   and   post-­‐deliberations   in   Ontario   and   Alberta,   and   post-­‐ deliberations  in  PEI).  Other  sources  of  concerns  that  emerged  include  being  able  to  cover   travel  costs,  being  away  from  family  and  friends  and  safety  concerns  (see  Figure  6.2).   Figure  6.2:  What  are  two  things  that  would  concern  you  the  most  about  having  to  travel  a  long   distance  to  undergo  complex  cancer  surgery?       P a g e    122     Implications for policy Despite  slight  variances  across  citizen  panels,  a  set  of  core  values  was  identified  that  should   be   considered   in   improving   the   delivery   of   complex   cancer   surgeries.   Several   of   these   values  clearly  suggest  that  any  quality-­‐improvement  initiative  or  structural  change  should   pay   close   attention   to   its   impact   on   patients   and   families.   Indeed,   panel   deliberations   indicated   that   interventions   must   be   put   in   place   to   provide   enhanced   support   for   patients   and  families  throughout  their  cancer  journey,  and  especially  to  overcome  barriers  to  access   optimal   surgical   care.   In   addition,   participants   from   all   three   panels   emphasized   the   opportunity   of   building   on   positive   things   already   happening   in   the   health   system   as   a   foundation   for   future   efforts,   including   current   efforts   to   regionalize   certain   complex   cancer  surgeries  and  to  establish  province-­‐wide  standards.  Panel  deliberations  also  suggest   the   need   to   develop   governance   arrangements   that   are   favourable   to   support   coordination   across   jurisdictions,   which   are   essential   to   supporting   patients   and   families   residing   in   a   province  or  territory  where  complex  cancer  surgeries  are  not  available.                                               P a g e    123     7.0:  SUMMARY  AND  KEY  RECOMMENDATIONS   The   high   burden   of   disease   associated   with   cancer   on   both   the   healthcare   system   and   patients  presses  the  need  for  novel  quality  improvement  initiatives  to  improve  outcomes  of   surgical   cancer   care.   This   report   summarized   the   pan-­‐Canadian   current   state   of   surgical   cancer  care  for  high-­‐risk,  resource  intensive  surgery  for  esophageal,  pancreatic,  liver,  lung   and  ovarian  cancers.    The  findings  observed  a  significant  variability  in  outcomes  between   provinces   and   heterogeneous   approaches   to   healthcare   system   organization  and  oversight.   There   has   been   very   little   progress   on   the   implementation   of   active   regionalization   nationally,   despite   evidence   indicating   that   regionalization   policies   can   help   create   an   environment   for   quality   improvement.   Lastly,   the   report   established   that   patients   are   willing  to  accept  the  increased  travel  time  to  surgical  cancer  care  and  apparent  loss  of  local   resources   if   it   means   that   they   can   expect   improved   outcomes.   This   evidence   provides   support   to   the   implementation   of   regionalization   on   a   broader   scale.   While   further   study   is   required,   we   recommend   that   provinces   consider   thoughtfully   implementing   regionalization  policies  in  ways  that  best  serve  the  needs  of  their  population  and  by  doing   so,  improving  the  overall  quality  of  patient  care  and  health  care  system  efficiency.       Evaluating   the   present   state   of   regionalization   and   surgical   cancer   care   outcomes   is   challenging.   The   impacts   of   regionalization   will   be   different   for   each   surgery   type,   due   partially   to   the   difference   in   associated   risks   and   frequency   of   procedures.   Some   cancers   are  diagnosed  later  in  life  and  the  proportion  of  resectable  cases  is  low.  Since  there  are  no   prospective  disease-­‐specific  registries  in  Canada  to  date,  the  breadth  of  the  available  data   for   review   was   limited   beyond   mortality,   length   of   stay,   and   resection   rates.   There   are   a   number   of   ‘soft’   outcomes   of   interest   such   as   complications,   the   number   of   salvage   procedures,   rates   of   readmission   that   could   help   inform   the   quality   of   surgical   cancer   care,   but  we  could  not  measure  this  given  the  data  limitations  in  this  report.  We  do  expect  to  see   an  improvement  in  such  outcomes  with  regionalization  as  well.  Despite  these  limitations,   the   potential   impact   of   regionalization   to   high   volume   centres   remains   evident   and   the   evidence   can   only   be   strengthened   through   the   establishment   of   systematic   data   collection   with   formalized   reporting   along   with   national,   disease-­‐specific   standards   of   care   so   that   one  regionalized  centre  is  comparable  to  another.       While   challenges   are   to   be   expected   with   implementing   any   sort   of   quality   improvement   program,   establishing   a   program   that   requires   such   considerable   coordination   between       P a g e    124     multiple  levels  of  the  health  care  system,  communities  and  citizens  is  likely  to  face  barriers   along  the  way.  It  will  require  a  strong  coordinated  effort  to  create  a  change  that  considers   the   consequences   and   develops   innovative   approaches   that   overcome   the   perceived   barriers   in   order   to   realize   the   potential   benefits   to   care   for   Canadians.   This   report   concludes   with   a   series   of   recommendations   that   will   assist   policy   makers,   health   care   planners,   opinion   leaders   within   the   surgical   communities   of   practice   and   administrators   with  working  towards  the  desired  future  state  of  active  regionalization.       Recommendation  #1:  Surgical  cancer  care  should  be  integrated  into  the  overall     spectrum   of   provincial   cancer   services   with   the   capacity   for   establishing   systematic  evaluation  and  the  provision  of  sufficient  resources  to  enact  change       A   common   theme   noted   during   the   provincial   expert   interviews   was   that   there   is   a   distinct   lack  of  leadership  for  surgical  cancer  services.  In  medical  and  radiation  oncology,  there  is   often   an   appointed   Lead   who   is   responsible   for   directing   the   implementation   of   clinical   guidelines,  manpower  allocation,  advocacy  at  policymaker  and  budget  allocation  tables  and   finally   the   ability   to   facilitate   quality   assurance   initiatives.   In   most   provinces,   there   is   no   such   role   for   the   surgical   portion   of   the   cancer   treatment   process,   despite   the   fact   that   surgery   is   associated   with   better   survival.   The   role   should   be   created   and   filled   by   a   key   opinion   leader   within   the   surgical   cancer   community   in   each   province,   with   a   clearly   defined   mandate   and   objectives.   This   common   role   across   the   provinces   will   allow   provinces   to   have   the   flexibility   to   integrate   this   position   into   their   existing   cancer   care   organization  schemes  while  still  maintaining  the  central  purpose  of  the  role.           Recommendation   #2:   Purposeful   regionalization   of   cancer   surgical   services     above   and   beyond   simple   consolidation   is   necessary   to   improve   health   care     quality  and  patient  outcomes     The   intent   of   this   recommendation   is   to   facilitate   a   formal   leadership   position   in   each   province  that  will  allow  initiatives  such  as  regionalization,  quality  improvement  reporting   and   the   development   and   uptake   of   clinical   guidelines   to   gain   traction   and   proceed   to   implementation.   It   is   important   to   differentiate   between   passive   consolidation   of   health   care   services   to   centralized   locations   and   active   regionalization   approaches   that   encompass   a   range   of   quality   improvement   initiatives   in   addition   to   amassing   larger   volumes.    The  quantitative  analysis  undertaken  in  this  report  indicates  that  higher  volume   centres   are   associated   with   improvements   in   patient   mortality   rates   after   resections,   yet   volume   alone   is   not   wholly   responsible   for   these   improvements.   Even   in   the   highest   volume   centres,   the   introduction   of   evidence-­‐based   surgical   standards   and   clinical   pathways,   the   utilization   of   dedicated,   trained   staff   training   or   the   adoption   of   optimal       P a g e    125     surgical   techniques   have   the   potential   to   further   reduce   mortality.   Regionalized,   higher-­‐ volume   centres   are   more   likely   to   possess   the   infrastructure   and   manpower   to   provide   multidisciplinary   care,   a   key   quality   standard   indicated   by   clinical   guidelines   released   by   organizations  such  as  the  American  National  Comprehensive  Cancer  Network  (NCCN)167,168   and   the   European   Society   for   Medical   Oncology   (ESMO)169,170   for   thoracic   malignancies   and   by  NCCN  for  hepatobiliarypancreatic  cancers171.172.  Lastly,  feedback  obtained  via  the  citizen   panel   work   undertaken   for   this   report   indicated   that   the   public   is   willing   to   trade   off   a   certain   degree   of   convenience   for   optimal   outcomes.   This   provides   the   justification   needed   on  a  user-­‐basis  to  support  the  regionalization  of  surgical  cancer  services.         The  intent  of  this  recommendation  is  promote  the  adoption  of  active  regionalization  on  a   provincial   basis,   as   it   has   shown   promise   as   a   good   approach   to   improve   the   quality   of   surgical  cancer  services  that  is  also  palatable  to  patients  and  the  general  public.         Recommendation  #3:  Regionalization  policies  should  be  tailored  to  meet  unique   provincial  needs     During  the  course  of  consultations  with  experts  across  the  country,  it  was  very  clear  that   each   province   has   unique   needs   that   will   impact   the   implementation   of   quality   improvement   policies   within   existing   health   care   system   organization.   Each   province   possesses   different   healthcare   system   governance   and   oversight   structures,   geographic   realities,   and   access   to   care   expectations   that   shape   how   policies   can   be   implemented.   This   is   best   demonstrated   by   the   case   of   Prince   Edward   Island,   where   complex   surgeries   are   often  referred  to  neighbouring  provinces.  This  policy  certainly  limits  access  to  care,  but  the   low   population   of   this   province   means   that   meaningful   volumes   could   never   be   met.   In   order  for  residents  of  PEI  to  realize  the  benefit  of  regionalization  policies,  this  population   would  be  best  suited  to  continue  to  be  referred  to  other  provinces  for  complex  care.  Since   no  one  province  shares  similar  infrastructure  with  another,  it  is  not  feasible  to  establish  a   single,   national   regionalization   policy.     Regardless,   the   volume-­‐outcome   effect   should   be   considered   when   allocating   surgical   centres   of   excellence   where   appropriate.   Similarly,   while   it   would   be   ideal   to   be   able   to   define   a   threshold   for   each   cancer   type   to   be   incorporated   into   provincial   regionalization   policies,   the   relative   difference   in   provincial   populations  means  that  it  would  be  truly  impractical  to  recommend  a  single  volume  cut-­‐off.       The   intent   of   this   recommendation   is   not   enforce   a   single,   national   policy   for   regionalization   activities,   but   to   instead   acknowledge   that   each   province  should  integrate   this   policy   into   their   current   organizational   infrastructure.   Each   province   should   be   responsible  for  initiating  a  dialogue  within  each  community  of  practice,  led  by  the  Surgical       P a g e    126     Cancer   Care   Lead,   to   establish   provincial   needs   and   to   define   reasonable   volume   standards   that  consider  population  distribution  and  available  resources.         Recommendation   #4:   Surgical   cancer   care   should   embrace   flexible   implementation   of   regionalization  policies   based   on   access   to   care   concerns   and     patient  preferences     A   primary   concern   of   patients   contemplating   undergoing   surgical   cancer   care   in   regionalized   centres   of   excellence   is   that   some   patients   would   be   removed   from   their   communities  and  support  networks  and  would  face  longer  travel  times  to  access  care.  Over   the   course   of   the   citizen   panel   discussions,   participants   indicated   that   innovative   healthcare   delivery   approaches   can   help   overcome   the   burden   of   reduced   access   to   care   that  may  come  with  adopting  regionalization  policies.  These  approaches  can  take  the  form   of   the   development   of   collaborations   with   local   institutions   in   smaller   centres   with   oversight  and  direction  from  the  regional  centre,  where  patients  can  attend  for  diagnostic   assessment   programs,   clinical   testing   or   post-­‐surgical   examinations   closer   to   home   while   still   realizing   the   benefits   of   regionalized   care.   Alternately,   telemedicine   may   be   utilized,   where   patients   can   present   at   their   local   healthcare   facility   for   a   consultation   with   their   surgeon   and   care   team   at   the   regional   centre.   The   objective   of   these   adapted   approaches   would   be   to   allow   patients   to   minimize   costs   and   travel   while   still   receiving   high   quality   care.       The   intent   of   this   recommendation   is   to   encourage   provinces   considering   the   implementation  of  regionalization  to  seek  innovative  approaches  that  will  allow  patients  to   see   the   benefits   of   such   quality   improvement   programs   while   also   ameliorating   the   impact   on   patients   and   their   families.   This   flexibility   is   intended   to   promote   implementation   given   the  different  provincial  realities.         Recommendation   #5:   Nationally-­‐implemented   standards   of   care   should   be     developed  for  each  cancer  surgery  type     The   volume-­‐outcome   relationship   has   a   significant   impact   on   patient   mortality   and   likely   other  less  concrete  outcomes,  but  does  not  account  for  all  the  improvements  in  outcomes.   Regionalized  centres  will  naturally  have  a  higher  volume  of  patients  than  they  would  have   in   a   non-­‐regionalized   setting,   but   there   are   other   quality   factors   such   as   highly   trained   personnel,  the  creation  of  specialized  surgical  stepdown  units  or  the  hiring  of  sufficiently   trained  surgeons  among  many  other  possibilities.  One  such  approach  to  ensuring  a  degree   of   uniformity   as   to   what   defines   a   regionalized   centre   would   be   to   develop   nationally-­‐ implemented   standards   of   care   specific   to   each   cancer   type.   Each   specialty-­‐based       P a g e    127     community   of   practice   should   be   charged   with   defining   minimal   standards   for   practitioner   certification   or   training   to   ensure   only   highly   trained   surgeons   are   undertaking   these   complex   resections,   the   use   of   multi-­‐disciplinary   cancer   conferences   to   ensure   the   most   appropriate   care   is   selected   for   patients,   the   establishment   of   minimal   institutional   and/or   surgeon  case  volumes  to  guide  policy  decision  making,  and  the  mandatory  participation  of   centres   in   national   quality   improvement   registries   to   facilitate   the   evaluation   of   regionalization   efforts.   The   standards   should   also   consider   the   integration   of   specialized   nursing,   critical   care   and   anaesthesia   services,   and   adequate   radiology   and   pathology   service   support   to   ensure   that   the   appropriate   personnel   and   infrastructure   are   in   place   to   deliver   care   according   to   clinical   guidelines.   These   standards   need   to   be   evidence-­‐based   using   current   research,   similar   to   those   that   have   been   developed   internationally,   to   encourage  the  uptake  of  best  practices  on  both  provincial  and  national  levels.  The  adoption   of  standards  nationally  will  also  help  to  address  the  disparities  in  patient  outcomes  noted   between   the   provinces,   both   by   establishing   what   constitutes   quality   care,   but   also   by   allowing   outcomes   to   be   compared   between   institutions   in   different   provinces   to   better   define   success.     Without   defined   national   standards   of   care,   there   is   too   much   inherent   variability  in  care  practices.  The  evaluation  of  adherence  to  these  surgical  standards  of  care   should   be   integrated   into   existing   national   evaluation   structures   in   order   to   measure   province-­‐level  improvements  and  inter-­‐provincial  variance.     The   intent   of   this   recommendation   is   for   national-­‐level   specialty   communities   of   care   to   review   existing   evidence   and   to   undergo   a   consensus   generating   exercise   to   establish   suitable  informed  standards  of  care  centred  on  regionalizing  surgical  cancer  care  services.   These  standards  are  then  intended  to  be  applied  in  each  province,  with  regular  auditing  for   compliance  and  quality  improvement.       Recommendation   #6:   A   structured   benchmarking   process   for   each   specialty     should  be  supported  to  improve  surgical  outcomes  and  inform  policy  decisions     When   specialty-­‐based   communities   of   care   are   asked   to   establish   nationally-­‐defined   standards   of   care,   it   is   expected   that   standardized   data   collection   with   explicit   public   reporting  will  be  a  key  component  to  measure  the  impact  of  regionalization  policies.  As  per   the  defined  standards  of  care,  all  surgical  care  providers  should  be  expected  to  participate   in   these   national   disease-­‐specific   registries.   The   utilization   of   registries   with   formal   reporting   procedures   can   improve   the   quality   of   patient   care   independently   of   other   interventions,  as  observed  with  the  American  College  of  Surgeons’  National  Surgical  Quality   Improvement   Program   (NSQIP).   This   program   utilizes   systematic   data   collection   and   reporting   for   patients   undergoing   surgery   for   any   indication,   and   has   reported   reduced   mortality   and   morbidity   variability   by   9%   and   30%,   respectively173.   While   there   is   apparent   value   to   generic   databases,   considerably   more   valuable   information   can   be       P a g e    128     derived  from  disease-­‐specific  registries  that  capture  a  standardized  core  dataset  consisting   of   patient   demographics   and   risk   factors,   cancer   and   procedure-­‐specific   details   and   eventual   outcomes,   including   surgery-­‐related   adverse   events   and   oncologic   outcomes.     The   richness  of  this  data  will  allow  for  a  valid  evaluation  of  improvements  in  patient  outcomes   and  reductions  in  inter-­‐provincial  variability  as  a  result  of  volume  intensification  and  the   establishment  of  standards  of  care.     It  is  necessary  for  each  community  of  practice  to  define  and  standardize  these  data  fields  to   ensure   that   data   collection   is   clinically   relevant   and   that   the   results   can   be   compared   between  institutions.  The  standardization  of  data  collection  across  sites  will  allow  surgeons   and  institutions  to  be  benchmarked  against  their  peers  with  trends  tracked  overtime  both   within   and   between   participants.   This   continuous   monitoring   facilitates   the   detection   of   poor   outcomes,   informing   decision   makers   and   health   care   planners   of   areas   requiring   additional   attention   or   resources.   Such   monitoring   also   has   the   potential   to   allow   for   the   creation   and   dissemination   of   best   practices,   where   institutions   with   excellent   outcomes   can   share   innovative   strategies   with   other   institutions,   resulting   in   a   system-­‐wide   improvement   in   outcomes   according   to   the   principles   of   positive   deviance.   Finally,   the   collection  and  monitoring  of  surgical  data  encourages  surgeon  accountability  to  evidence-­‐ based  guidelines  and  the  established  standards  of  care.             The  recommendations  of  this  report  were  selected  as  they  indicate  important  factors  that   are  missing  in  the  present  state  of  surgical  cancer  care  across  Canada,  leading  to  large  gaps   in   knowledge   and   inconsistencies   in   surgical   cancer   care   delivery.   Provinces   implementing   active   regionalization   should   consider   these   recommendations   starting   from   the   policy   development   stages   and   continuing   through   to   ongoing   evaluation   and   monitoring.   Ideally,   adoption   of   a   nationally   standardized,   provincially   implemented   approach   to   regionalization   will   place   Canada   as   a   world   leader   in   quality   surgical   cancer   care   for   high-­‐ risk,  complex  cancer  surgery.       P a g e    129     References   1.  Constitution  of  the  World  Health  Organization.  Geneva:  World   Health  Organization.  1948.   standards.  Toronto  (ON).  Cancer  Care  Ontario.  Program  in   Evidence-­‐based  Care  Practice  Guideline  Report;  2006.   2.  Donabedian  A.  Evaluating  the  quality  of  medical  care.   Milbank  Mem  Fund  Q.  1966;  44(3  Suppl):  166–206   3.  Matula  SR,  Mercado  C,  Ko  CY,  Tomlinson  JS.  Quality  of  Care  in   Surgical  Oncology.  Cancer  Control.  2009;  14(6):  303-­‐11.   18.  Yasunaga  H,  Horiguchi  H,  Matsuda  S,  et  al.  Relationship   between  hospital  volume  and  operative  mortality  for  liver   resection:  Data  from  the  Japanese  diagnosis  procedure   combination  database.  Hepatology  Research.  2012;  42(11):   1073-­‐1080.   4.  Canadian  Cancer  Society’s  Advisory  Committee  on  Cancer   Statistics.  Canadian  Cancer  Statistics  2014.  Toronto,  ON:   Canadian  Cancer  Society;  2014.   19.  McColl  RJ,  You  X,  Ghali  WA,  Kaplan  G,  Myers  R,  Dixon  E.   Recent  trends  of  hepatic  resection  in  Canada:  1995-­‐2004.   Journal  of  Gastrointestinal  Surgery  .2008;  12(11):  1839-­‐1846.     5.  Canadian  Institute  for  Health  Information,  National  Health   Expenditure  Trends,  1975  to  2014  (Ottawa,  Ontario:  CIHI,   2014).   20.  Dimick  JB,  Wainess  RM,  Cowan  JA,  Upchurch  GR  Jr,  Knol   JA,  Colletti  LM.  National  trends  in  the  use  and  outcomes  of   hepatic  resection.  Journal  of  the  American  College  of  Surgeons.   2004;  199(1):  31-­‐38.   6.  Canadian  Institute  for  Health  Information,  Health  Care  Cost   Drivers:  The  Facts.  (Ottawa,  Ontario:  CIHI,  2011).     7.  Canadian  Cancer  Society’s  Advisory  Committee  on  Cancer   Statistics.  Canadian  Cancer  Statistics  2013.  Toronto,  ON:   Canadian  Cancer  Society;  2013.   8.  Public  Health  Agency  of  Canada.  Leading  cause  of  death  and   hospitalization  in  Canada.  Ottawa  (ON):  The  Agency;  2008.   Available:  www.phac-­‐aspc.gc.ca/publicat/lcd-­‐pcd97/index-­‐ eng.php  (accessed  2014  Apr,6)   9.  Fritz  A,  Percy  C,  Jack  A,  et  al.    International  Classification  of   Diseases  for  Oncology.  3rd  ed.  Geneva,  Switzerland:  World   Health  Organization;  2000.   10.  Canadian  Institute  for  Health  Information,  Surgical  Volume   Trends,  2009—Within  and  Beyond  Wait  Time  Priority  Areas   (Ottawa,  Ontario:  CIHI,  2009).   11.  Feo  CV,  Villaflor  VM,  Patti  MG.  Should  esophageal   resections  for  cancer  be  performed  in  high-­‐volume  centers   only?  Updates  in  Surgery.  2011;  63(3):  147-­‐150.   21.  Choti  MA,  Bowman  HM,  Pitt  HA,  et  al.  Should  hepatic   resections  be  performed  at  high-­‐volume  referral  centers?   Journal  of  Gastrointestinal  Surgery:  Official  Journal  of  the  Society   for  Surgery  of  the  Alimentary  Tract.  1998;  2(1):  11-­‐20.   22.  Park  BJ,  Altorki  NK.  Diagnosis  and  management  of  early   lung  cancer.  Surg  Clin  North  Am.  2002;  82(3):457-­‐76.   23.  Vernooij  F,  Heintz  AP,  Coebergh  JW,  Massuger   LF,  Witteveen  PO,  van  der  Graaf  Y.  Specialized  and  high-­‐volume   care  leads  to  better  outcomes  of  ovarian  cancer  treatment  in   the  Netherlands.  Gynecologic  Oncology.  2009;  112(3):  455-­‐461.   24.  Olaitan  A,  McCormack  M.  Centralisation  of  services  for  the   management  of  ovarian  cancer:  Arguments  for.  BJOG:  An   International  Journal  of  Obstetrics  and  Gynaecology.  2007;   114(10):  1188-­‐1190.   25.  Elit  L,  Bondy  SJ,  Paszat  L,  Przybysz  R,  Levine  M.  Outcomes   in  surgery  for  ovarian  cancer.  Gynecologic  Oncology.  2002;   87(3):  260-­‐267.   12.  McPhee  JT,  Hill  JS,  Whalen  GF,  et  al.  Perioperative  mortality   for  pancreatectomy:  A  national  perspective.  Annals  of  Surgery.   2007;  246(2):  246-­‐253.   26.  de  Oliveira  C,  Bremmer  K,  Pataky  R,  et  al.  Under-­‐standing   the  costs  of  cancer  care  before  and  after  diagnosis  for  the  21   most  common  cancers  in  Ontario:  A  population-­‐based   descriptive  study.  CMAJ  Open.  2013;  1:  E1-­‐8.   13.  Nienhuijs  SW,  Rutten  HJ,  Luiten  EJ,  et  al.  Reduction  of  in-­‐ hospital  mortality  following  regionalisation  of  pancreatic   surgery  in  the  south-­‐east  of  The  Netherlands.  European  Journal   of  Surgical  Oncology.  2010;  36(7):  652-­‐656.   27.  Kocher  KE,  Sklar  DP,  Mehrotra  A,  Tayal  VS,  Gausche-­‐Hill  M,   Myles  Riner  R.  Categorization,  designation,  and  regionalization   of  emergency  care:  Definitions,  a  conceptual  framework,  and   future  challenges.  Acad  Emerg  Med.  2010;  17:  1306–1311.   14.  Simunovic  M,  Urbach  D,  Major  D,  et  al.  Assessing  the   volume-­‐outcome  hypothesis  and  region-­‐level  quality   improvement  interventions:  Pancreas  cancer  surgery  in  two   Canadian  provinces.  Annals  of  Surgical  Oncology.  2010;  17(10):   2537-­‐2544.   28.  Church  J,  Barker  P.  Regionalization  of  health  services  in   Canada:  a  critical  perspective.  Int  J  Health  Serv.  1998;  28(3):   467–486.     15.  Young  J,  Thompson  A,  Tait  I,  Waugh  L,  McPhillips  G.   Centralization  of  services  and  reduction  of  adverse  events  in   pancreatic  cancer  surgery.  World  Journal  of  Surgery.  2013;   37(9):  2229-­‐2233.   16.  Nordback  L,  Parviainen  M,  Raty  S,  Kuivanen  H,  Sand  J.   Resection  of  the  head  of  the  pancreas  in  Finland:  Effects  of   hospital  and  surgeon  on  short-­‐term  and  long-­‐term  results.   Scandinavian  Journal  of  Gastroenterology.  2002;  37(12):  1454-­‐ 1460.   17.  Marcaccio  M,  Langer  B,  Rumble  B,  Hunter  A.  Hepatic,   Pancreatic,  and  Biliary  Tract  (HPB)  surgical  oncology       29.  Raval  MV,  Bilimoria  KY,  Talamonti  MS.  Quality   improvement  for  pancreatic  cancer  care:  Is  regionalization  a   feasible  and  effective  mechanism?  Surgical  Oncology  Clinics  of   North  America.  2010;  19(2):  371-­‐390.   30.  Halm  EA,  Lee  C,  Chassin  MR.  Is  volume  related  to  outcome   in  health  care?  A  systematic  review  and  methodologic  critique   of  the  literature.  Ann  Intern  Med.  2002;  137(6):  511-­‐520.   31.  Metreveli  RE,  Sahm  K,  Denstman  F,  Abdel-­‐Misih  R,  Petrelli   NJ.  Hepatic  resection  at  a  major  community-­‐based  teaching   hospital  can  result  in  good  outcome.  Annals  of  Surgical   Oncology.  2005;  12(2):  133-­‐137.   P a g e    130     32.  Woo  YL,  Kyrgiou  M,  Bryant  A,  Everett  T,  Dickinson  HO.   Centralisation  of  services  for  gynaecological  cancers-­‐  A   Cochrane  systematic  review.  Gynecologic  Oncology.  2012;   126(2):  286-­‐290.   47.  Stavrou  PE,  Smith  GS,  Baker  DF.  Surgical  outcomes   associated  with  oesophagectomy  in  New  South  Wales:  An   investigation  of  hospital  volume.  Journal  of  Gastrointestinal   Surgery.  2010;  14(6):  951-­‐957.   33.  Urbach  DR,  Bell  CM,  Austin  PC.  Differences  in  operative   mortality  between  high-­‐and  low-­‐volume  hospitals  in  Ontario   for  5  major  surgical  procedures:  estimating  the  number  of  lives   potentially  saved  through  regionalization.  Canadian  Medical   Association  Journal.2003;  168(11):  1409-­‐1414.   48.  Lin  HC,  Lin  CC.  Surgeon  volume  is  predictive  of  5-­‐year   survival  in  patients  with  hepatocellular  carcinoma  after   resection:  A  population-­‐based  study.  Journal  of  Gastrointestinal   Surgery.  2009;  13(12):  2284-­‐2291.   34.  Birkmeyer  JD,  Siewers  AE,  Finlayson  EV,  et  al.  Hospital   volume  and  surgical  mortality  in  the  United  States.  N  Engl  J   Med.  2002;  346:  1128–37.   49.  Jarnagin  WR,  Gonen  M,  Fong  Y,  et  al.  Improvement  in   perioperative  outcome  after  hepatic  resection:  analysis  of   1,803  consecutive  cases  over  the  past  decade.  Annals  of   Surgery.  2002;  236(4):  397-­‐406.   35.  Birkmeyer  JD,  Finlayson  EV,  Birkmeyer  CM.  Volume   standards  for  high-­‐risk  surgical  procedures:  potential  benefits   of  the  Leapfrog  initiative.  Surgery.  2001;  130:  415–22.   50.  Fan  ST,  Lo  CM,  Liu  CL,  et  al.  Hepatectomy  for  hepatocellular   carcinoma:  toward  zero  hospital  deaths.  Annals  of  Surgery.   1999;  229(3):  322-­‐330.   36.  Begg  CB,  Cramer  LD,  Hoskins  WJ,  Brennan  MF.  Impact  of   hospital  volume  on  operative  mortality  for  major  cancer   surgery.  Journal  of  the  American  Medical  Association.  1998;   280(20):  1747-­‐1751.   51.  Fong  Y,  Fortner  J,  Sun  RL,  Brennan  MF,  Blumgart  LH.   Clinical  score  for  predicting  recurrence  after  hepatic  resection   for  metastatic  colorectal  cancer:  Analysis  of  1001  consecutive   cases.  Annals  of  Surgery.  1999;  230(3):  309-­‐318.   37.  Glasgow  RE,  Showstack  JA,  Katz  PP,  Corvera  CU,  Warren  RS,   Mulvihill  SJ.  The  relationship  between  hospital  volume  and   outcomes  of  hepatic  resection  for  hepatocellular  carcinoma.   Archives  of  Surgery.  1999;  134(1):  30-­‐35.   52.  Gayowski  TJ,  Iwatsuki  S,  Madariaga  JR,  Selby  R,  Todo  S,   Irish  W,  Starzl  TE.  Experience  in  hepatic  resection  for   metastatic  colorectal  cancer:  Analysis  of  clinical  and  pathologic   risk  factors.  Surgery.  1994;  116(4):  703-­‐710.   38.  Dimick  JB,  Cowan  JA  Jr,  Knol  JA,  Upchurch  GR  Jr.  Hepatic   resection  in  the  United  States:  Indications,  outcomes,  and   hospital  procedural  volumes  from  a  nationally  representative   database.  Archives  of  Surgery.  2003;  138(2):  185-­‐191.     53.  Nordlinger  B,  Guiguet  M,  Vaillant  JC,  et  al.  Surgical  resection   of  colorectal  carcinoma  metastases  to  the  liver:  A  prognostic   scoring  system  to  improve  case  selection,  based  on  1568   patients.  Cancer.  1996;  77(7):  1254-­‐1262.   39.  Jin  LX,  Pitt  SC,  Hall  BL,  Pitt  HA.  Aggressive  surgical   management  of  gallbladder  cancer:  At  what  cost?  Surgery   (United  States).  2013;  154(2):  266-­‐273.   54.  Scheele  J,  Stang  R,  Altendorf-­‐Hofmann  A,  Paul  M.  Resection   of  colorectal  liver  metastases.  World  journal  of  surgery.  1995;   19(1):  59-­‐71.   40.  Wouters  MW,  Gooiker  GA,  van  Sandick  JW,  Tollenaar  RA.   The  volume-­‐outcome  relation  in  the  surgical  treatment  of   esophageal  cancer:  A  systematic  review  and  meta-­‐analysis.   Cancer.  2012;  118(7):  1754-­‐1763.   55.  Bristow  RE,  Zahurak  ML,  Diaz-­‐Montes  TP,  Giuntoli  RL,   Armstrong  DK.  Impact  of  surgeon  and  hospital  ovarian  cancer   surgical  case  volume  on  in-­‐hospital  mortality  and  related   short-­‐term  outcomes.  Gynecologic  Oncology.  2009;  115(3):   334-­‐338.   41.  Gillison  EW,  Powell  J,  McConkey  CC,  Spychal  RT.  Surgical   workload  and  outcome  after  resection  for  carcinoma  of  the   oesophagus  and  cardia.  Br  J  Surg.  2002;  89:  344–348.   42.  Rouvelas  I,  Lindblad  M,  Zeng  W,  Viklund  P,  Ye  W,  Lagergren   J.  Impact  of  hospital  volume  on  long-­‐term  survival  after   esophageal  cancer  surgery.  Archives  of  Surgery.  2007;  142(2):   113-­‐117.   43.  Thompson  AM,  Rapson  T,  Gilbert  FJ,  Park  KG,  Scottish  Audit   of  Gastric  and  Oesophageal  Cancer.  Hospital  volume  does  not   influence  long-­‐term  survival  of  patients  undergoing  surgery  for   oesophageal  or  gastric  cancer.  Br  J  Surg.  2007;  94(5):  578-­‐84.   44.  Verhoef  C,  van  de  Weyer  R,  Schaapveld  M,  Bastiaannet  E,   Plukker  JT.  Better  survival  in  patients  with  esophageal  cancer   after  surgical  treatment  in  university  hospitals:  A  plea  for   performance  by  surgical  oncologists.  Ann    Surg    Oncol.  2007;   14:  1678–1687.     45.  Derogar  M,  Sadr-­‐Azodi  O,  Johar  A,  Lagergren  P,  Lagergren  J.   Hospital  and  surgeon  volume  in  relation  to  survival  after   esophageal  cancer  surgery  in  a  population-­‐based  study.  Journal   of  Clinical  Oncology.  2013;  31(5):  551-­‐557.   46.  Dikken  JL,  Dassen  AE,  Lemmens  VE,  et  al.  Effect  of  hospital   volume  on  postoperative  mortality  and  survival  after   oesophageal  and  gastric  cancer  surgery  in  the  Netherlands   between  1989  and  2009.  European  Journal  of  Cancer.  2012;   48(7):  1004-­‐1013.       56.  Fago-­‐Olsen  CL,  Hogdall  C,  Kehlet  H,  Christensen  IJ,  Ottesen   B.  Centralized  treatment  of  advanced  stages  of  ovarian  cancer   improves  survival:  A  nationwide  Danish  survey.  Acta  Obstet   Gynecol  Scand.  2011;  90(3):  273-­‐79.     57.  Bristow  RE,  Palis  BE,  Chi  DS,  Cliby  WA.  The  national  cancer   database  report  on  advanced-­‐stage  epithelial  ovarian  cancer:   Impact  of  hospital  surgical  case  volume  on  overall  survival  and   surgical  treatment  paradigm.  Gynecologic  Oncology.  2010;   118(3):  262-­‐267.   58.  Schrag  D,  Earle  C,  Xu  F,  et  al.  Associations  between  hospital   and  surgeon  procedure  volumes  and  patient  outcomes  after   ovarian  cancer  resection.  Journal  of  the  National  Cancer   Institute.  2006;  98(3):  163-­‐171.   59.  Wright  JD,  Herzog  TJ,  Siddiq  Z,  et  al.  Failure  to  rescue  as  a   source  of  variation  in  hospital  mortality  for  ovarian  cancer.   Journal  of  Clinical  Oncology.  2012;  30(32):  3976-­‐3982.   60.  Bristow  R,  Chang  J,  Ziogas  A,  Anton-­‐Culver  H.  NCCN   treatment  guidelines  for  ovarian  cancer:  A  population-­‐based   validation  study  of  structural  and  process  quality  measures.   Gynecologic  Oncology.  2013;  130(1):  e18.   61.  Dimick  JB,  Pronovost  PJ,  Cowan  JA  Jr,  Lipsett  PA,  Stanley  JC,   Upchurch  GR  Jr.  Variation  in  postoperative  complication  rates   after  high-­‐risk  surgery  in  the  United  States.  Surgery.  2003;  134:   534–40.   P a g e    131     62.  Dimick  JB,  Pronovost  PJ,  Cowan  JA,  Lipsett  PA.  Surgical   volume  and  quality  of  care  for  esophageal  resection:  do  high-­‐ volume  hospitals  have  fewer  complications?  Ann  Thorac  Surg.   2003;  75:  337-­‐41.   63.  Viklund  P,  Lindblad  M,  Lu  M,  Ye  W,  Johansson  J,  Lagergren  J.   Risk  factors  for  complications  after  esophageal  cancer   resection:  A  prospective  population-­‐based  study  in  Sweden.   Annals  of  Surgery.  2006;  243(2):  204-­‐211.       64.  Riall  TS,  Eschbach  KA,  Townsend  CM  Jr,  Nealon   WH,  Freeman  JL,  Goodwin  JS.  Trends  and  disparities  in   regionalization  of  pancreatic  resection.  Journal  of   Gastrointestinal  Surgery.  2007;  11(10):  1242-­‐1252.   65.  Birkmeyer  JD,  Siewers  AE,  Marth  NJ,  Goodman  DC.   Regionalization  of  high-­‐risk  surgery  and  implications  for   patient  travel  times.  JAMA.  2003;  290(20):  2703-­‐2708.   66.  Ghaferi  AA,  Birkmeyer  JD,  Dimick  JB.  Hospital  volume  and   failure  to  rescue  with  high-­‐risk  surgery.  Med  Care.  2011;  49:   1076-­‐1081.     67.  Kim  C,  Kwak  EK,  Lee  S.  The  relationship  between  hospital   volume  and  outcome  of  gastrointestinal  cancer  surgery  in   Korea.  Journal  of  Surgical  Oncology.  2011;  104(2):  116-­‐123.     68.  Belghiti  J,  Hiramatsu  K,  Benoist  S,  Massault  PP,  Sauvanet  A,   Farges  O.  Seven  hundred  forty-­‐seven  hepatectomies  in  the   1990s:  An  update  to  evaluate  the  actual  risk  of  liver  resection.   Journal  of  the  American  College  of  Surgeons.  2000;  191(1):  38-­‐ 46.   69.  Birkmeyer  JD,  Stukel  TA,  Siewers  AE,  Goodney  PP,   Wennberg  DE,  Lucas  FL.  Surgeon  volume  and  operative   mortality  in  the  united  states.  New  England  Journal  of  Medicine.   2003;  349(22):  2117-­‐2127.   70.  Hannan  EL,  Radzyner  M,  Rubin  D,  Dougherty  J,  Brennan  MF.   The  influence  of  hospital  and  surgeon  volume  on  in-­‐hospital   mortality  for  colectomy,  gastrectomy,  and  lung  lobectomy  in   patients  with  cancer.  Surgery.  2002;  131(1):  6-­‐15.   71.  Lien  Y,  Huang  M,  Lin  H.  Association  between  surgeon  and   hospital  volume  and  in-­‐hospital  fatalities  after  lung  cancer   resections:  The  experience  of  an  Asian  country.  Annals  of   Thoracic  Surgery.  2007;  83(5):  1837-­‐1843.   72.  Pecorelli  N,  Balzano  G,  Capretti  G,  Zerbi  A,  Di  Carlo  V,  Braga   M.  Effect  of  surgeon  volume  on  outcome  following   pancreaticoduodenectomy  in  a  high-­‐volume  hospital.  Journal  of   Gastrointestinal  Surgery.  2012;  16(3):  518-­‐523.   73.  Rosemurgy  AS,  Bloomston  M,  Serafini  FM,  Coon  B,  Murr   MM,  Carey  LC.  Frequency  with  which  surgeons  undertake   pancreaticoduodenectomy  determines  length  of  stay,  hospital   charges,  and  in-­‐hospital  mortality.  Journal  of  Gastrointestinal   Surgery.  2001;  5(1):  21-­‐26.   74.  Bachmann  MO,  Alderson  D,  Edwards  D,  et  al.  Cohort  study   in  south  and  west  England  of  the  influence  of  specialization  on   the  management  and  outcome  of  patients  with  oesophageal   and  gastric  cancers.  British  Journal  of  Surgery.  2002;  89(7):   914-­‐922.   75.  Miller  JD,  Jain  MK,  de  Gara  CJ,  Morgan  D,  Urschel  JD.  Effect   of  surgical  experience  on  results  of  esophagectomy  for   esophageal  carcinoma.  J  Surg  Oncol.  1997;  65:  20–21.   76.  Sundelof,  M,  Lagergren,  J,    Ye  W.  Surgical  factors  influencing   outcomes  in  patients  resected  for  cancer  of  the  esophagus  or       gastric  cardia.  World  Journal  of  Surgery.  2008;  32(11):  2357-­‐ 2365.   77.  Whitson  BA,  Groth  SS,  Duval  SJ,  Swanson  SJ,  Maddaus  MA.   Surgery  for  Early-­‐Stage  Non-­‐Small  Cell  Lung  Cancer:  A   Systematic  Review  of  the  Video-­‐Assisted  Thoracoscopic   Surgery  versus  Thoracotomy  Approaches  to  Lobectomy.  Ann   Thorac  Surg.  2008;  86:  2008-­‐2018.   78.  Goodney  PP,  Lucas  FL,  Stukel  TA,  Birkmeyer  JD.  Surgeon   specialty  and  operative  mortality  with  lung  resection.  Annals  of   Surgery.  2005;  241(1):  179-­‐184.   79.  Martin-­‐Ucar  AE,  Waller  DA,  Atkins  JL,  Swinson  D,  O’Byrne   KJ,  Peake  MD.  The  beneficial  effects  of  specialist  thoracic   surgery  on  the  resection  rate  for  non-­‐small-­‐cell  lung  cancer.   Lung  Cancer.  2004;  46:  227–232.   80.  Silvestri  GA,  Handy  J,  Lackland  D,  Corley  E,  Reed  CE.   Specialists  achieve  better  outcomes  than  generalists  for  lung   cancer  surgery.  Chest.  1998;  114:  675–680.   81.  Carney  ME,  Lancaster  JM,  Ford  C,  Tsodikov  A,  Wiggins  CL.  A   population-­‐based  study  of  patterns  of  care  for  ovarian  cancer:   Who  is  seen  by  a  gynecologic  oncologist  and  who  is  not?   Gynecol  Oncol.  2002;  84:  36–42.     82.  Tingulstad  S,  Skjeldestad  FE,  Hagen  B.  The  effect  of   centralization  of  primary  surgery  on  survival  in  ovarian  cancer   patients.  Obstet  Gynecol.  2003;  102:  499–505.     83.  Junor  EJ,  Hole  DJ,  Gillis  CR.  Management  of  ovarian  cancer:   referral  to  a  multidisciplinary  team  matters.  Br  J  Cancer.  1994;   70:  363–370.     84.  Eisenkop  SM,  Spirtos  NM,  Montag  TW,  Nalick  RH,  Wang  HJ.   The  impact  of  subspecialty  training  on  the  management  of   advanced  ovarian  cancer.  Gynecol  Oncol.  1992;  47:  203–209.     85.    Junor  EJ,  Hole  DJ,  McNulty  L,  Mason  M,  Young  J.  Specialist   gynaecologists  and  survival  outcome  in  ovarian  cancer:  A   Scottish  national  study  of  1866  patients.  Br  J  Obstet  Gynaecol.   1999;  106:  1130–1136.     86.  Engelen  MJ,  Kos  HE,  Willemse  PHB,  et  al.  Surgery  by   consultant  gynecologic  oncologists  improves  survival  in   patients  with  ovarian  carcinoma.  Cancer.  2006;  106:  589–598.     87.  Harmon  JW,  Tang  DG,  Gordon  TA,  et  at.  Hospital  volume   can  serve  as  a  surrogate  for  surgeon  volume  for  achieving   excellent  outcomes  in  colorectal  resection.  Annals  of  surgery.   1999;  230(3):  404-­‐413.   88.  Finley  CJ,  Bendzsak  A,  Tomlinson  G,  Keshavjee  S,  Urbach   DR,  Darling  GE.  The  effect  of  regionalization  on  outcome  in   pulmonary  lobectomy:  A  Canadian  national  study.  J  Thorac   Cardiovasc  Sur.  2010;  140:  757–763.   89.  ouvelas  I,  Lagergren  J.  The  impact  of  volume  on  outcomes   after  oesophageal  cancer  surgery.  ANZ  Journal  of  Surgery.   2010;  80(9):  634-­‐641.   90.  Casson  AG,  Van  Lanschot  JB.  Improving  outcomes  after   esophagectomy:  The  impact  of  operative  volume.  Journal  of   Surgical  Oncology.  2005;  92(3):  262-­‐266.   91.  Metzger  R,  Bollschweiler  E,  Vallbohmer  D,  Maish  M,   DeMeester  TR,  Holscher  AH.  High  volume  centers  for   esophagectomy:  What  is  the  number  needed  to  achieve  low   postoperative  mortality?  Diseases  of  the  Esophagus.  2004;   17(4):  310-­‐314.   P a g e    132     92.  Crawford  R,  Greenberg  D.  Improvements  in  survival  of   gynecological  cancer  in  the  Anglia  region  of  England:  are  these   am  effect  of  centralisation  of  care  and  use  of  multidisciplinary   management?  BJOG.  2011;  119(2);  160-­‐165.   93.  Wouters  MW,  Krijnen  P,  Le  Cessie  S,  et  al.  Volume-­‐  or   outcome-­‐based  referral  to  improve  quality  of  care  for   esophageal  cancer  surgery  in  the  Netherlands.  Journal  of   Surgical  Oncology.  2009;  99(8):  481-­‐487.   94.  Chang  DC,  Zhang  Y,  Mukherjee  D,  et  al.  Variations  in  referral   patterns  to  high-­‐volume  centers  for  pancreatic  cancer.  Journal   of  the  American  College  of  Surgeons.  2009;  209(6):  720-­‐726.   95.  Von  Meyenfeldt  EM,  Gooiker  GA,  Van  Gijn  W,  et  al.  The   relationship  between  volume  or  surgeon  specialty  and   outcome  in  the  surgical  treatment  of  lung  cancer:  A  systematic   review  and  meta-­‐analysis.  Journal  of  Thoracic  Oncology.  2012;   7(7):  1170-­‐1178.   96.  Birkmeyer  JD,  Dimick  JB,  Birkmeyer  NJ.  Measuring  the   quality  of  surgical  care:  structure,  process,  or  outcomes?  J  Am   Coll  Surg.  2004;  198(4):  626–632.   97.  Langer  B.  Role  of  volume  outcome  data  in  assuring  quality   in  HPB  surgery.  Hpb.  2007;  9(5):  330-­‐334.   98.  Sundaresan  S,  Langer  B,  Oliver  T,  Schwartz  F,  Brouwers  M,   Stern  H.  Standards  for  thoracic  surgical  oncology  in  a  single-­‐ payer  health  care  system.  Annals  of  Thoracic  Surgery.  2007;   84(2):  693-­‐701.   99.  Abu  Hilal  M,  Jain  G,  Kasasbeh  F,  Zuccaro  M,  Elberm  H.   Laparoscopic  distal  pancreatectomy:  Critical  analysis  of   preliminary  experience  from  a  tertiary  referral  centre.  Surgical   Endoscopy  and  Other  Interventional  Techniques.  2009;  23(12):   2743-­‐2747.   100.  Westgaard  A,  Laronningen  S,  Mellem  C,  et  al.  Are  survival   predictions  reliable?  Hospital  volume  versus  standardisation  of   histopathologic  reporting  for  accuracy  of  survival  estimates   after  pancreatoduodenectomy  for  adenocarcinoma.  European   Journal  of  Cancer.  2009;  45(16):  2850-­‐2859.   107.  Teh  SH,  Diggs  BS,  Deveney  CW,  Sheppard  BC.  Patient  and   hospital  characteristics  on  the  variance  of  perioperative   outcomes  for  pancreatic  resection  in  the  United  States:  A  plea   for  outcome-­‐based  and  not  volume-­‐based  referral  guidelines.   Archives  of  Surgery.  2009;  144(8):  713-­‐721.   108.  Cunningham  JD,  O'Donnell  N,  Starker  P.  Surgical  outcomes   following  pancreatic  resection  at  a  low-­‐volume  community   hospital:  Do  all  patients  need  to  be  sent  to  a  regional  cancer   center?  American  Journal  of  Surgery.  2009;  198(2):  227-­‐230.   109.  Chan  C,  Franssen  B,  Rubio  A,  Uscanga  L.   Pancreaticoduodenectomy  in  a  Latin  American  country:  The   transition  to  a  high-­‐volume  center.  Journal  of  Gastrointestinal   Surgery.  2008;  12(3):  527-­‐533.   110.  Afsari  A,  Zhandoug  Z,  Young  S,  Ferguson  L,  Silapaswan  S,   Mittal  V.  Outcome  analysis  of  pancreaticoduodenectomy  at  a   community  hospital.  The  American  Surgeon.  2002;  68(3):  281-­‐ 284.   111.  Treasure  T,  Utley  M,  Bailey  A.  Assessment  of  whether  in-­‐ hospital  mortality  for  lobectomy  is  a  useful  standard  for  the   quality  of  lung  cancer  surgery:  Retrospective  Study.  BMJ.  2003;   327(7406):  73.   112.  Urbach  DR,  Baxter  NN.  Does  it  matter  what  a  hospital  is   "high  volume"  for?  Specificity  of  hospital  volume-­‐outcome   associations  for  surgical  procedures:  Analysis  of  administrative   data.  Quality  &  Safety  in  Health  Care.  2004;  13(5):  379-­‐383.   113.  Finlayson  EVA,  Birkmeyer  JD.  Effects  of  hospital  volume   on  life  expectancy  after  selected  cancer  operations  in  older   adults:  A  decision  analysis.  Journal  of  the  American  College  of   Surgeons.  2003;  196(3):  410-­‐417.   114.  Finlayson  EVA,  Goodney  PP,  Birkmeyer  JD,  Davies  RJ.   Hospital  volume  and  operative  mortality  in  cancer  surgery:  A   national  study.  Archives  of  Surgery.2003;   138(7):  721-­‐726.   115.  van  Meerbeeck  JP,  Damhuis  RA,  Vos  de  Wael  ML.  High   postoperative  risk  after  pneumonectomy  in  elderly  patients   with  right-­‐sided  lung  cancer.  Eur  Respir  J.  2002;  19:  141–5.   101.  Balzano  G,  Zerbi  A,  Capretti  G,  Rocchetti  S,  Capitanio  V,  Di   Carlo  V.  Effect  of  hospital  volume  on  outcome  of   pancreaticoduodenectomy  in  Italy.  The  British  Journal  of   Surgery.  2008;  95(3):  357-­‐362.   116.  Bach  PB,  Cramer  LD,  Schrag  D,  Downey  RJ,  Gelfand  SE,   Begg  CB.  The  influence  of  hospital  volume  on  survival  after   resection  for  lung  cancer.  New  England  Journal  of  Medicine.   2001;  345(3):  181-­‐188.   102.  Bilimoria  KY,  Bentrem  DJ,  Ko  CY,  et  al.  Multimodality   therapy  for  pancreatic  cancer  in  the  U.S.:  Utilization,  outcomes,   and  the  effect  of  hospital  volume.  Cancer.  2007;  110(6):  1227-­‐ 1234.     117.  Khuri  SF,  Daley  J,  Henderson  W,  et  al.  Relation  of  surgical   volume  to  outcome  in  eight  common  operations:  results  from   the  VA  National  Surgical  Quality  Improvement  Program.  Ann   Surg.  1999;  230:  414–29.   103.  Simunovic  M,  To  T,  Theriault  M,  Langer  B.  Relation   between  hospital  surgical  volume  and  outcome  for  pancreatic   resection  for  neoplasm  in  a  publicly  funded  health  care  system.   CMAJ.  1999;  160(5):  643-­‐648.   118.  Birkmeyer  JD,  Sun  Y,  Wong  SL,  Stukel  TA.  Hospital  volume   and  late  survival  after  cancer  surgery.  Annals  of  Surgery.  2007;   245(5):  777-­‐783.   104.  Birkmeyer  JD,  Finlayson  SRG,  Tosteson  ANA,  Sharp  SM,   Warshaw  AL,  Fisher  ES.  Effect  of  hospital  volume  on  in-­‐hospital   mortality  with  pancreaticoduodenectomy.  Surgery.  1999;   125(3):  250-­‐256.   105.  Schell  MT,  Barcia  A,  Spitzer  AL,  Harris  HW.   Pancreaticoduodenectomy:  Volume  is  not  associated  with   outcome  within  an  academic  health  care  system.  HPB  Surgery.   2008.   106.  Topal  B,  Van  de  Sande  S,  Fieuws  S,  Penninckx  F.  Effect  of   centralization  of  pancreaticoduodenectomy  on  nationwide   hospital  mortality  and  length  of  stay.  British  Journal  of  Surgery.   2007;  94(11):  1377-­‐1381.       119.  Sundaresan  S,  McLeod  R,  Irish  J,  et  al.  Early  results  after   regionalization  of  thoracic  surgical  practice  in  a  single-­‐payer   system.  Annals  of  Thoracic  Surgery.  2013;  95(2):  472-­‐479.   120.  Kozower  BD,  Stukenborg  GJ.  The  relationship  between   hospital  lung  cancer  resection  volume  and  patient  mortality   risk.  Annals  of  Surgery.  2011;  254(6):  1032-­‐1037.   121.  Otake  H,  Yasunaga  H,  Horiguchi  H,  Matsutani  N,  Matsuda   S,  Ohe  K.  Impact  of  hospital  volume  on  chest  tube  duration,   length  of  stay,  and  mortality  after  lobectomy.  Annals  of   Thoracic  Surgery.  2011;  92(3):  1069-­‐1074.   122.  Cheung  MC,  Hamilton  K,  Sherman  R,  et  al.  Impact  of   teaching  facility  status  and  high-­‐volume  centers  on  outcomes   P a g e    133     for  lung  cancer  resection:  an  examination  of  13,469  surgical   patients.  Annals  of  surgical  oncology.  2009;  16(1):  3-­‐13. 123.  Sioris  T,  Sihvo  E,  Sankila  R,  Salo  J.  Effect  of  surgical  volume   and  hospital  type  on  outcome  in  non-­‐small  cell  lung  cancer   surgery:  A  Finnish  population-­‐based  study.  Lung  Cancer.  2008;   59(1):  119-­‐125.   124.  Osada  H,  Yamakoshi  E.  Hospital  volume  and  surgical   outcomes  of  lung  cancer  in  Japan.  General  Thoracic  and   Cardiovascular  Surgery.  2007;  55(9):  360-­‐365.   125.  Stukenborg  GJ,  Wagner  DP,  Harrell  J,  FE.  Temporal  order   and  nonlinearity  in  the  relationship  between  lung  cancer   resection  volume  and  in-­‐hospital  mortality.  Health  Services  and   Outcomes  Research  Methodology.  2004;  5(1):  59-­‐73.     126.  Strand  TE,  Rostad  H,  Damhuis  RA,  Norstein  J.  Risk  factors   for  30-­‐day  mortality  after  resection  of  lung  cancer  and   prediction  of  their  magnitude.  Thorax.  2007;  62(11):  991-­‐997.   127.  Freixinet  JL,  Julia-­‐Serda  G,  Rodriguez  PM,  et  al.  Hospital   volume:  Operative  morbidity,  mortality  and  survival  in   thoracotomy  for  lung  cancer.  A  Spanish  multicenter  study  of   2994  cases.  European  Journal  of  Cardio-­‐Thoracic  Surgery.  2006;   29(1):  20-­‐25.   138.  Dimick  JB,  Cattaneo  SM,  Lipsett  PA,  Pronovost  PJ,   Heitmiller  RF.  Hospital  volume  is  related  to  clinical  and   economic  outcomes  of  esophageal  resection  in  Maryland.   Annals  of  Thoracic  Surgery.  2001;  72(2):  334-­‐339.   139.  Swisher  SG,  Deford  L,  Merriman  KW,  et  al.  Effect  of   operative  volume  on  morbidity,  mortality,  and  hospital  use   after  esophagectomy  for  cancer.  J  Thorac  Cardiovasc  Surg.   2000;  119:  1126–32.   140.  Patti  MG,  Corvera  CU,  Glasgow  RE,  Way  LW.  A  hospital’s   annual  rate  of  esophagectomy  influences  the  operative   mortality  rate.  J  Gastro  Surg.1998;2:186–92.   141.  Dikken  JL,  Van  Sandick  JW,  Allum  WH,  et  al.  Differences  in   outcomes  of  oesophageal  and  gastric  cancer  surgery  across   Europe.  British  Journal  of  Surgery.  2013;  100(1):  83-­‐94.   142.  Kozower  BD,  Stukenborg  GJ.  Hospital  esophageal  cancer   resection  volume  does  not  predict  patient  mortality  risk.   Annals  of  Thoracic  Surgery.  2012;  93(5):  1690-­‐1698.   143.  Anderson  O,  Ni  Z,  Moller  H,  et  al.  Hospital  volume  and   survival  in  oesophagectomy  and  gastrectomy  for  cancer.   European  Journal  of  Cancer.  2011;  47(16):  2408-­‐2414.   128.  Little  AG,  Rusch  VW,  Bonner  JA,  et  al.  Patterns  of  surgical   care  of  lung  cancer  patients.  Annals  of  Thoracic  Surgery.  2005;   80(6):  2051-­‐2056.   144.  Fujita  H,  Ozawa  S,  Kuwano  H,  Ueda  Y,  Hattori  S,  Yanagawa   T.  Esophagectomy  for  cancer:  Clinical  concerns  support   centralizing  operations  within  the  larger  hospitals.  Diseases  of   the  Esophagus.  2010;  23(2):  145-­‐152.   129.  Rouvelas  I,  Jia  C,  Viklund  P,  Lindblad  M,  Lagergren  J.   Surgeon  volume  and  postoperative  mortality  after   oesophagectomy  for  cancer.  European  Journal  of  Surgical   Oncology.  2007;  33(2):  162-­‐168.   145.  Wouters  MW,  Wijnhoven  BP,  Karim-­‐Kos  HE,  et  al.  High-­‐ volume  versus  low-­‐volume  for  esophageal  resections  for   cancer:  The  essential  role  of  case-­‐mix  adjustments  based  on   clinical  data.  Annals  of  Surgical  Oncology.  2008;  15(1):  80-­‐87.   130.  Migliore  M,  Choong  CK,  Lim  E,  Goldsmith  KA,  Ritchie  A,   Wells  FC.  A  surgeon's  case  volume  of  oesophagectomy  for   cancer  strongly  influences  the  operative  mortality  rate.   European  Journal  of  Cardio-­‐Thoracic  Surgery.  2007;  32(2):  375-­‐ 380.   146.  Al-­‐Sarira  AA,  David  G,  Willmott  S,  Slavin  JP,  Deakin  M,   Corless  DJ.  Oesophagectomy  practice  and  outcomes  in  England.   British  Journal  of  Surgery.  2007;  94(5):  585-­‐591.   131.  Jeganathan  R,  Kinnear  H,  Campbell  J,  et  al.  A  surgeon's   case  volume  of  oesophagectomy  for  cancer  does  not  influence   patient  outcome  in  a  high  volume  hospital.  Interact  Cardiovasc   Thorac  Surg.  2009;  9(1):  66-­‐69.   132.  Rutegard  M,  Lagergren  J,  Rouvelas  I,  Lagergren  P.  Surgeon   volume  is  a  poor  proxy  for  skill  in  esophageal  cancer  surgery.   Annals  of  Surgery.  2009;  249(2):  256-­‐261.   133.  Leigh  Y,  Goldacre  M,  McCulloch  P.  Surgical  specialty,   surgical  unit  volume  and  mortality  after  oesophageal  cancer   surgery.  European  Journal  of  Surgical  Oncology.  2009;  35(8):   820-­‐825.   147.  Jensen  LS,  Bendixen  A,  Kehlet  H.  Organisation  and  early   outcomes  of  major  upper  gastrointestinal  cancer  surgery  in   Denmark  1996-­‐2004.  Scandinavian  Journal  of  Surgery.  2007;   96(1):  41-­‐45.   148.  Wenner  J,  Zilling  T,  Bladstrom  A,  Alvegard  TA.  The   influence  of  surgical  volume  on  hospital  mortality  and  5-­‐year   survival  for  carcinoma  of  the  oesophagus  and  gastric  cardia.   Anticancer  Research.  2005;  25(1B):  419-­‐424.   149.  Fumagalli  U,  Bersani  M,  Russo  A,  Melis  A,  de  Pascale  S,   Rosati  R.  Volume  and  outcomes  after  esophageal  cancer   surgery:  The  experience  of  the  region  of  lombardy-­‐Italy.   Updates  in  Surgery.  2013;  65(4):  271-­‐275.   134.  Christian  CK,  Gustafson  ML,  Betensky  RA,  Daley  J,  Zinner   MJ.  The  Leapfrog  volume  criteria  may  fall  short  in  identifying   high-­‐quality  surgical  centers.  Ann  Surg.  2003;  238:  447–55.   150.  Boddy  AP,  Williamson  JM,  Vipond  MN.  The  effect  of   centralisation  on  the  outcomes  of  oesophagogastric  surgery-­‐A   fifteen  year  audit.  International  Journal  of  Surgery.  2012;  10(7):   360-­‐363.   135.  Dimick  JB,  Cowan  JA  Jr,  Ailawadi  G,  Wainess  RM,  Upchurch   GR  Jr.  National  variation  in  operative  mortality  rates  for   esophageal  resection  and  the  need  for  quality  improvement.   Arch  Surg.  2003;  138:  1305–9.   151.  Funk  LM,  Gawande  AA,  Semel  ME,  et  al.  Esophagectomy   outcomes  at  low-­‐volume  hospitals:  The  association  between   systems  characteristics  and  mortality.  Annals  of  Surgery.  2012;   253(5):  912-­‐917.   136.  Kuo  EY,  Chang  Y,  Wright  CD.  Impact  of  hospital  volume  on   clinical  and  economic  outcomes  for  esophagectomy.  Ann   Thorac  Surg.  2001;  72:  1118–24.   152.  Milne  AA,  Skinner  J,  Browning  G.  Centralisation  of   oesophageal  cancer  services;  the  view  from  the  periphery.   Journal  of  the  Royal  College  of  Surgeons  of  Edinburgh.  2000;   45(3):  164-­‐167.   137.  van  Lanschot  JJ,  Hulscher  JB,  Buskens  CJ,  Tilanus  HW,  ten   Kate  FJ,  Obertop  H.  Hospital  volume  and  hospital  mortality  for   esophagectomy.  Cancer.  2001;  91(8):  1574-­‐1578.           P a g e    134     153.  Staal  EF,  van  Coevorden  F,  Cats  A,  et  al.  Outcome  of  low-­‐ volume  surgery  for  esophageal  cancer  in  a  high-­‐volume   referral  center.  Annals  of  Surgical  Oncology.  2009;  16(12):   3219-­‐3226.   154.  Beenen  E,  Jao  W,  Coulter  G,  Roberts  R.  The  high  volume   debate  in  a  low  volume  country:  Centralisation  of  oesophageal   resection  in  New  Zealand.  The  New  Zealand  Medical  Journal.   2013;  126(1374):  34-­‐45.   155.  Santin  BJ,  Price  P.  Laparoscopic  transhiatal   esophagectomy  at  a  low-­‐volume  center.  Journal  of  the  Society  of   Laparoendoscopic  Surgeons.  2011;  15(1):  41-­‐46.   156.  Kumpulainen  S,  Kuoppala  T,  Leminen  A,  et  al.  Surgical   treatment  of  ovarian  cancer  in  different  hospital  categories-­‐  A   prospective  nation-­‐wide  study  in  Finland.  European  Journal  of   Cancer.  2006;  42(3):  388-­‐395.   157.  Ioka  A,  Tsukuma  H,  Ajiki  W,  Oshima  A.  Influence  of   hospital  procedure  volume  on  ovarian  cancer  survival  in  Japan,   a  country  with  low  incidence  of  ovarian  cancer.  Cancer  Science.   2004;  95(3):  233-­‐237.   158.  Mandato  VD,  Abrate  M,  De  Iaco  P,  et  al.  Clinical   governance  network  for  clinical  audit  to  improve  quality  in   epithelial  ovarian  cancer  management.  J  Ovarian  Res.  2013;  6:   19.   159.  Crawford  R,  Greenberg  D.  Improvements  in  survival  of   gynaecological  cancer  in  the  Anglia  region  of  England:  Are   these  an  effect  of  centralisation  of  care  and  use  of   multidisciplinary  management?  BJOG:  An  International  Journal   of  Obstetrics  and  Gynaecology.  2012;  119(2):  160-­‐165.   160.  Brookfield  KF,  Cheung  MC,  Yang  R,  Byrne  MM,  Koniaris   LG.  Will  patients  benefit  from  regionalization  of  gynecologic   cancer  care?  Plos  One.  2009;  4(1):  1.   161.  Munstedt  K,  Von  Georgi  R.,  Misselwitz  B,  Zygmunt  M,   Stillger  R,  Kunzel  W.  Centralizing  surgery  for  gynecologic   oncology  -­‐  A  strategy  assuring  better  quality  treatment?   Gynecologic  Oncology.  2003;  89(1):  4-­‐8.   162.  Berger-­‐Chen  S,  Herzog  TJ,  Lewin  SN,  et  al.  Access  to   conservative  surgical  therapy  for  adolescents  with  benign   ovarian  masses.  Obstetrics  &  Gynecology.  2012;  119(2  Pt  1):   270-­‐275.   164.  Phippen  NT,  Barnett  JC,  Lowery  WJ,  Miller  CR,  Leath  III   CA.  Surgical  outcomes  and  national  comprehensive  cancer   network  compliance  in  advanced  ovarian  cancer  surgery  in  a   low  volume  military  treatment  facility.  Gynecologic   Oncology.2013;  131(1):  158-­‐162.   165.  Soegaard  Andersen  E,  Knudsen  A,  Svarrer  T,  et  al.  The   results  of  treatment  of  epithelial  ovarian  cancer  after   centralisation  of  primary  surgery.  Results  from  North  Jutland,   Denmark.  Gynecologic  Oncology.  2005;  99(3):  552-­‐556.   166.  Riaz  SP,  Lüchtenborg  M,  Jack  RH,  et  al.  Variation  in   surgical  resection  for  lung  cancer  in  relation  to  survival:   Population-­‐based  study  in  England  2004-­‐2006.  Eur  J  Cancer.   2012;  48:  54-­‐60.   167.  National  Comprehensive  Cancer  Network.  NCCN  Clinical   Practice  Guidelines  in  Oncology  (NCCN  Guidelines®):  Non   Small  Cell  Lung  Cancer.  Version  6.2015.  Retrieved  online   www.nccn.org  on  May  11,  2015.  Uploaded  April  28,  2015.   168.  National  Comprehensive  Cancer  Network.  NCCN  Clinical   Practice  Guidelines  in  Oncology  (NCCN  Guidelines®):   Esophageal  and  Esophagogastric  Junction  Cancers.  Version   3.2015.  Retrieved  online  www.nccn.org  on  May  11,  2015.   Uploaded  March  23,  2015.   169.  Vansteenkiste  J,  De  Ruysscher  D,  Eberhardt  WEE,  Lim  E,   Senan  S,  Felip  E,  Peters  S.  Early  and  locally  advanced  non-­‐ small-­‐cell  lung  cancer  (NSCLC):  ESMO  Clinical  Practice   Guidelines  for  diagnosis,  treatment  and  follow-­‐up.  Annals  of   Oncology  2013;24(Suppl  6):vi89-­‐vi98.   170.  Stahl  M,  Mariette  C,  Haustermans  K,  Cervantes  A,  Arnold   D.  Oesophageal  cancer:  ESMO  Clinical  Practice  Guidelines  for   diagnosis,  treatment  and  follow-­‐up.  Annals  of  Oncology   2013;24(Suppl  6):vi51-­‐vi56.   171.  National  Comprehensive  Cancer  Network.  NCCN  Clinical   Practice  Guidelines  in  Oncology  (NCCN  Guidelines®):   Hepatobiliary  Cancers.  Version  2.2015.  Retrieved  online   www.nccn.org  on  May  11,  2015.  Uploaded  February  6,  2015.   172.  National  Comprehensive  Cancer  Network.  NCCN  Clinical   Practice  Guidelines  in  Oncology  (NCCN  Guidelines®):   Pancreatic  Adenocarcinoma.  Version  2.2015.  Retrieved  online   www.nccn.org  on  May  11,  2015.  Uploaded  March  6,  2015.   163.  Du  Bois  A,  Rochon  J,  Lamparter  C,  Pfisterer  J,  AGO   Organkommission  OVAR  PF.  Pattern  of  care  and  impact  of   participation  in  clinical  studies  on  the  outcome  in  ovarian   cancer.  International  Journal  of  Gynecological  Cancer.  2005;   15(2):  183-­‐191.   173.  Khuri  SF,  Daley  J,  Henderson  W,  Hur  K,  Demakis  J,  Aust  JB,   et  al.  The  Department  of  Veterans  Affairs'  NSQIP:  the  first   national,  validated,  outcome-­‐based,  risk-­‐adjusted,  and  peer-­‐ controlled  program  for  the  measurement  and  enhancement  of   the  quality  of  surgical  care.  National  VA  Surgical  Quality   Improvement  Program.  Ann  Surg.  1998;228(4):491-­‐507                         P a g e    135     Appendix  1:  Summary  of  Literature  Review   Databases  Searched:   -­‐ Medline,  Embase,  HealthStar,  Canadian  Health  Research  Collection   Inclusion  and  Exclusion  Criteria:   Inclusions   1. 2. 3. 4. 5. 6. Age:  Greater  than  18  years   Study  population   a. Article  published  after  year  2000  for  ovarian,  lung,  pancreatic,  esophageal,  and  hepatobiliary   since  1990   Outcomes  of  regionalization  practices   Volume-­‐outcome  relationships  (at  either  a  surgeon-­‐  or  a  hospital-­‐level)  with  any  outcome       High-­‐volume  vs.  Low-­‐volume  treatment  centres     Measurable  outcomes:   a. Patient  5-­‐yr  survival  rate   b. In-­‐hospital  mortality   c. Total  length  of  hospital  stay     d. Readmission  rates     e. Adverse  events     f. Quality  of  life     g. Cost  analyses   Exclusions   1. 2. 3. 4. 5.     Titles  and  abstract  in  foreign  languages   Unable  to  locate  the  journal  article     Single  Institution  or  single  surgeon  study   Data  collected  before  1985   Study  published  before  1990   P a g e    136     Literature  Summary  Tables     1) Esophageal  Cancer     Table  1a:  Surgeon  Specific  Factors  and  Patient  Outcomes,  Esophageal  Cancer   Ref  #   Type  of  Resection   Specialty   69   Esophagectomy   NR   NR   74   Esophagectomy   41     Cardioesophageal   NR   NR   NR   NR   75   Esophagectomy   63   Esophagectomy,  Cardia   Cancer  Resection   129   Esophagectomy   607   130   Esophagectomy   195   45   Esophagectomy     131   Esophagectomy   NR       NR   NR     Total  No.  of   surgeons   over  study   period   N  (%)   HV:  >6   MV:  2-­‐6   LV:  <2   HV:  NR   LV:  NR   HV:  >12   MV:  4-­‐11   LV:  <4   HV:  >6   LV:  <5   HV:  >5   LV:  <5     Total  No.  of   patients   over  study   period   N  (%)           1125         275   HV:>6   MV:2-­‐6   LV:  <2   HV:  >6   LV:  <6   LV:  quartiles  1-­‐2   MV:  quartile  3     HV:  quartile  4             NR   1335   Cases/surgeon   Surgeon  1:  38   Surgeon2:  36   Surgeon  3:  16   5   252   Volume   Categories  (per     year)   Mortality   (%)   [30-­‐day  unless   otherwise   specified]   9.2S   13.1   18.8   LowerS   Higher   11.8     6.6  S   15.1     0S   22.0   2%   7%   2.6   2.1%   7.1%   16.9   4.2   NR   S1:  5.2   S2:2.7%   S3:  0%   S4:  5.4%   Morbidity/          Post-­‐ operative   Complications   Length  of   Stay   (days)   Overall   Survival     NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   -­‐#  patients  with  at   least  1  complication:   42%  vs.  49%   -­‐#  patients  with  at   least  2   complications:  18%   vs.  24%   -­‐Risk  of  Anastomotic   leakage:  ~8  fold   more  for  LV  vs.  HVS     NR   NR   NR   LongerS   Shorter   ~18%  (ns   difference)   (12-­‐15  mos)   NR   NR   NR             NR   NR    S1:  23.7%   S2:22.2%   S3:  50%   S4:18.9%   NR   LV:  reference     MV:0.82  (  0.70   to  0.97)     HV:0.85  (0.68   to  1.06)   Mean  survival   (months):   S1:40   S2:31   P a g e    137     Surgeon  4:  74   Surgeon  5:  88   S5:  4.5%   S5:20.4%   76   Esophagectomy,   Gastric  Cardia   NR   LV:<10/yr   HV:≥10/yr   NR   232   HV:  0   LV:  2%     NR   132   Esophagectomy   NR   LV:  2   MV:2-­‐6   HV:>6   NR   615   NR   133   Esophagectomy     Cardio   thoracic     General   HV:  >100   LV:  ≤100   NR   9034   6.1%   9.0%     LV  only:   CardioThoracic:  6.8   (4.3-­‐9.4)   General:  9.8  (9.0-­‐ 10.6)     HV  only:   Cardio  Thoracic:  5.9   (4.9-­‐7.0)   General:  6.7  (5.5-­‐7.9)     HV:  reference   MV:  0.80  (95%  CI:   0.45–1.42)     LV:  0.99  (95%  CI:   0.49  –1.98)   NR   HV:  18  (9– 58)   LV:  18  (7– 102)   NR   NR   S3:22   S4:30   S5:31   NR   NR   NR     Table  1b:  Hospital-­‐Specific  Factors  and  Patient  Outcomes,  Esophageal  Cancer   Study   Type  of  Resection   Total  No.  of   Hospitals     No.  of   Patients   93   Esophagectomy   104   4939   74   Esophageal  Resection     41   Esophagectomy   NR   NR     112   Esophagectomy   47     613   113   Esophagectomy     134   Esophagectomy   40   603   NR   NR       1125     Volume   Categories  (per   year)   Mortality   [in-­‐hospital]  (%)   Morbidity/Post-­‐ operative   Complications   Length  of   Stay  (days)   Overall  Survival   (5-­‐year)   >20   10-­‐20   <10   High   Low   >20   <20   OR:  0.49S   OR:  1.01   reference   Lower   Higher   10.2%   9.8%   NR   NR   NR   NR   NR   NR   NR   NR     NR   NR     >8.8   <8.8   >9.0   <4.0   >10   <3   10.9%   15.6%   6.5%S   15.0%   LowerS   Higher   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   ~18%   (13  months  for  both   groups)*+   NR   NR   NR   NR   NR   NR   P a g e    138     61   Esophagectomy   NR   NR   6   138     >6   <6   >16   <3   NR   NR   3.7%S   11.8%   39%S   48%   NR   NR   135   Esophagectomy   62   Esophagectomy   204   162     >8.5   <8.5   2.5%S   15.4%   31   618   79   944   3   61   757   928   3   39     Esophagectomy   101  in-­‐total   340   >19   <2   >19   <2   28   1.1   >20   <10   >15   4-­‐15   <3   >5   <5   36   Esophagectomy     140   Esophagectomy   NR   NR   5   196   8.1%S   23.1%   10.6%S   17.0%   2.5%S   9.2%   4.9%S   12.1%   2.7%S   12.7%   16.0%   Early  post-­‐op:   3.0%S   12.2%   3.4%S   17.3%   6%   17%   Less  in  High  Volume   Hospitals   Post-­‐op  mortality   for  patient  with  one   complication:   16.9%  vs  2.5%S  for   those  without   complications   NR   NR   NR   NR   NR   NR   NR   NR   NR   34   Esophagectomy   34   Esophagectomy   136   Esophagectomy   137   Esophagectomy   138   Esophagectomy   139   42   Esophagectomy     1199   118   206   822   43   Esophagectomy     Esophagectomy     1302   44   Esophagectomy     213   45   Esophagectomy   NR   1335               1136     >11   <  5   >30   <5     >10   <10   14.7-­‐107   >35   20-­‐34   13-­‐19   13   >20   <20   LV:  quartiles  1-­‐2   MV:  quartile  3   HV:  quartile  4   Complications:   55%   68%   NR   NR   NR   NR   NR   NR   20%  (>20   days)   28%  (>20   days)   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   13  days   15  days   NR   NR   11S  (9-­‐16)   20  (13-­‐31)   19  (12-­‐33)   14.7  daysS   17.7  days   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   22  days   22  days   NR   NR   NR   NR   27.4%*+   23.8%   33.7%  S*   17.4%   36.8%*   37.6%   38.5%   43.4%   41.7%   22.3%   LV:  reference   MV:  1.03  (  0.87  to   1.22)   HV:  0.93  (  0.77  to   1.13)   4.5%   9.3%     NR   NR     NR   NR                 NR   NR   NR   NR   NR   NR   NR   P a g e    139     119   Esophagectomy   46  (pre   regionalization)   NR   Esophagectomy   5.9%  to  5.8%     NR   NR   NR   15(  post   regionalization)   NR   Pre   regionalization  vs.   post   regionalization   19864   Annual  hospital   volume:   1-­‐10   11-­‐20   21-­‐30   31-­‐40   ≥41   Esophagectomy:   1-­‐10:reference   11-­‐20:  0.82  (0·61,   1·11)   21-­‐30:0.68  (0·50,   0·93)   31-­‐40:0.58    (0·39,   0·85)   41+:0.55  (0·42,   0·72)     Gastrectomy   1-­‐10:  reference   11-­‐20:  0·84  (0·67,   1·05)   21-­‐30:  0·64  (0·41,   0·99)   Q1:  12.82%   Q2:13.41%   Q3:6.35%   Q4:  5.5%   Q5:  2.21%   *Six  month  mortality   Esophagectomy   Very  low:  ref   Low:  0.90  (0.78– 1.03)   Medium:  0.78  (0.62– 0.97)   High:  0.48  (0.38– 0.61)   Gastrectomy   Very  low:  ref   Low:  0.95  (0.84– 1.07)   Medium:  0.  95   (0.83–1.08)   High:  1.10    0.82– 1.49)   NR   NR   NR   NR   NR   2  year  survival:   Esophagectomy   1-­‐10:  reference   11-­‐20:  0·92  (0·78,   1·08)   21-­‐30:  0·84  (0·63,   1·11)   31-­‐40:  0·77  (0·63,   0·94)   41+:  0·79  (0·66,   0·96)   Gastrectomy   1-­‐10:  reference   11-­‐20:  1·04  (0·93,   1·15)   21-­‐30:  1·01  (0·84,   1·22)   NR   NR   NR   Hazard  ratio   LV:  ref   MV:  0.974   HV:0.865   VHV:0.660   NR     141   Esophagectomy   Gastrectomy     142   Esophagectomy   217   1210   Divided  into   quintiles   46   Esophagectomy   Gastrectomy   NR   24,246   Very  low  (1–5)   Low  (6–10)   Medium  (11–20)   High  (≥21)   143   Gastrointestinal  operations   (Esophagectomy,   Gastrectomy)       3870   Cases/year   LV:1-­‐10   MV:11-­‐20   HV:21-­‐30   VHV:≥31       3  year  survival   Esophagectomy   Very  Low:  reference   Low:  1.01(  0.94– 1.10)   Medium:  0.90  (0.81– 0.99)   High:  0.77(  0.70– 0.85)   Gastrectomy   Very  Low:  reference   Low:  0.99  (  0.91– 1.07)   Medium:  0.99  (0.90– 1.08)   High:  0.98  (0.86– 1.12)   NR   P a g e    140     47   Esophagectomy   NR   321   144   Esophagectomy   709   31380   133   Esophagectomy   NR   9034   76   Esophagectomy,     Gastric  cardia  resection   NR   232   145   Esophagectomy   12   903   146   Esophagectomy   NR   11838   147   Esophagectomy,   Pancreaticoduodenectomy,   Gastric  resection   NR   NR         Esophagectomy   between  2000-­‐ 2006   LV:≤10     MV:11-­‐  20   (inclusive)    HV:  ≥20   0-­‐4   5-­‐9   10-­‐19   20-­‐39   40-­‐79   80+   HV:  >100  cases   LV:  ≤  100  cases   HV:≥10/year   LV:  <10/year     HV:  1  of  the   hospitals  under   study     LV:    11  of  the   hospitals  under   study     VLV:≤9   LV:10-­‐19   MV:20-­‐29   HV30-­‐39   VHV:  40+   LV:<5   MV:5-­‐20   HV:>20   LV:  6.4%,  95%CI   1.7–18.6%   MV:  4.3%,  1.6– 10.3%   HV:  2.6%,     1.0–6.4%   NR   NR   0-­‐4:  6%  (5.3–6.8)   5-­‐9:  4.7  (4.2–5.3)   10-­‐19:  4.2  (3.7–4.7)   20-­‐39:  3.2  (2.8–3.7)   40-­‐79:  2.2  (1.8–2.6)   80+:  1.8%  (1.3–2.3)   HV:  6.3%   LV:  9.6%   HV:0%   LV:  3%   NR   NR   3  year  survival  rate:   LV:  45.1%  (23.4– 64.6%)   MV:  58.0%  (46.1– 68.2%)   HV:  64.4%  (CI  53.8– 73.2%)   NR   NR   NR   NR   NR   NR   HV:  5%S   LV:  13%   P<0.001   NR   HV:  19  (9– 57)   LV:  17.5  (7– 102)   HV:  14   LV:21    (median   days)   P<0.001   VLV:11.8%   LV:8.3%   MV:6.0%   HV:9.0%   VHV:4.5%     *most  recent  dataset   in  results   Esophagectomy   LV:  22.2  (12.7–34.5)   MV:  3.3  (1.8–7.0)   HV:  8.3  (6.2–12.6)   Pancreatic-­‐ duodectomy   LV:  6.3  (0.2–30.2   MV:  7.6  (3.7–13.7)   HV:  5.6  (1.6–13.8)   Gastric  Resection   LV:  8.2  (5.0–12.7)   MV:  8.0  (4.6–12.2)   HV:  8.4  (3.7–15.9)     NR   VLV:25.3%   LV:  25.3%   MV:20.9%   HV:25.2%   VHV:22.7%   *%  with   “prolonged   hospital  stay”   NR   NR   NR   NR   NR   P a g e    141     148   Esophagectomy,   Gastric  cardia  resections   74   1429   LV:  <5/year   MV:5-­‐15   HV:>15   Hospital  mortality   LV:10.4%   MV:6.3%   HV:3.5%   NR   149   Esophagectomy   111   2801   #  done  over   period   LV:  <50   MV:50-­‐149   HV:>150   LV:5.6%   MV:2.6%   HV:1.7%   NR   150   Esophagogastric  (OG)   resection   NR   456   Esophagectomy   874   4498   NR   NR   47   Esophagectomy   NR   321   Precentralization:   10.3%   Post  centralization:   3.6%   P=0.006   LV:  2.2  (1.3–3.7)   MV:  1.6  (1.0–2.5)   HV:  Reference   LV:  6.4%,  (95%CI   1.7–18.6%)     MV:  4.3%,  (95%CI   1.6–10.3%)     HV:  2.6%,  95%CI   1.0–6.4%   NR   151   Pre  vs  post   centralization     (  prior  and  post   2006)   LV:1-­‐6   MV:7-­‐32   HV:  33+   LV:  ≤10   MV:11-­‐20   HV:  >20   LV:  21  (16-­‐ 32)   MV:  18  (13-­‐ 26)   HV:  18  (15-­‐ 23)   LV  and  MV:   “more  than   25”   HV:  20   (median)     NR   LV:  23.4%,  (95%CI   12.8–38.4%)     MV:  31.0%,  (95%CI   23.0–40.3%)     HV:  18.7%,  (95%CI   13.1–25.9%)   NR   LV:7%     MV:  19%    HV:22%   (p=0.02)     NR   Median  survival   Precentralization:   1.1  years   Post  centralization:   1.5  years   NR   3  year  survival   LV:  45.1%  (95%CI   23.4–64.6%)     MV:  58.0%  (95%CI   46.1–68.2%)     HV:  64.4%  (95%CI   53.8–73.2%)   *Overall  5-­‐year  Survival.  +Tumour  stage  included  in  analysis       Table  1c:  Hospital  type  and  Patient  Outcomes,  Esophageal  Cancer   Study   152   118       Hospital  Type   Specialist     District   NR   Type  of  Resection   Esophagectomy   Upper  gastrointestinal   (UGI)  resections   Volume   Categories  (per     year)   NR   NR   Pre  -­‐Post   regionalization   Total  No.  of   patients  over   study  period     53   60   606   Mortality   (%)   [30-­‐day  unless   otherwise   specified]   5.6%   12.5%   Pre:  2.5%     Post:  2.4%   Morbidity/Post-­‐ operative   complications   NR   NR   Pre:  40%     Post:  45%   Length  of  stay   (days)   NR   NR   Pre:  Median  16   (2-­‐72)       Post:  Median  13   (3-­‐49)       P=0.024   Overall  Survival   (Years)   1-­‐year:63%   1-­‐year-­‐62%   NR   P a g e    142     153   154   155   High  volume   referral  centre,   but  has  low   volume  of  actual   operations       Low  volume   centre     Esophagectomy   NR   87   1%   53%   NR   1  year:  89%   3  year:  60%   Esophagogastrectomy   NR   128   1.6%   53.9%   12  (median)   Low  volume   centre   Laparoscopic  transhiatal   esophagectomy   NR   16   0   43.75%   Mean  hospital   stay  was  16.7   days  (range,  9  to   30)   32.4%  for   adenocarcinoma   and  47.7%  for   squamous     cell  carcinoma.   NR     2) Pancreatic  Cancer   Table  2a:  Surgeon  Specific  Factors  and  Patient  Outcomes,  Pancreatic  Cancer   Ref  #   Type  of  Resection   Specialty   72   Pancreaticoduodenectomy   15   Pancreatic  Head  Resection   Pancreatic   (training)   Pancreatic   73   Pancreaticoduodenectomy   NR   S=   Total  No.  of   Volume   surgeons  over   Categories                   study  period   (per    year)   N  (%)   HV:  >=12   NR   LV:<12   LV:<1   98   MV:1-­‐3   HV:>3   Total  No.  of   patients  over   study  period   N  (%)   630   1   2   3   4-­‐6   7-­‐9   10-­‐16   >17   698   * Significant  difference     *No  studies  found  comparing  effects  of  surgeon  specialty  on  outcomes         282   350   Mortality   (%)   [30-­‐day  unless   otherwise  specified]   3.9   4.3   14S   10   3S   6.6   5.4   3.4   10.2   8.7   0.0   2.6   Morbidity   (%)   Length  of   stay  (days)   55.3   65.6   42S   41S   25S   15.5   16.9  (mean)   24S   23S   18S   NR   27   19.4   19.7   19.7   19.6   17.6   16.8   Overall  Survival       Pancreatic  Ductal   Adenocarcinoma     LV:12%   MV:11%   HV:  17%     Ampullary   adenocarcinoma:   LV:  27%   MV:36%   HV:44%   NR   P a g e    143     Table  2b:  Hospital-­‐Specific  Factors  and  Patient  Outcomes,  Pancreatic  Cancer   Ref  #   Type  of  Resection   Total  No.  of   Hospitals     No.  of   Patients   Volume   Categories  (per   year)   Mortality   [30-­‐day]  (%)   Morbidity   Length  of   stay  (days)   Overall  Survival   (5-­‐year)   15   Pancreatic  Head  Resection   33   350   LV:  <5   MV:5-­‐10   HV:>10   LV:13%   MV:7%   HV:4%   41%   41%   29%   23   23   18    All   significant   12   Pancreaticoduodenectomy   10   158   24.4%   2.6%  (defined  as   “before  discharge  to   patients  home,  nursing   home,  or  rehab   centre”)   82%   38%   (Intra  and  post  morbidities   combined)   P=<0.001   NR   13   “pancreas  cancer  surgery”   25-­‐38   (Ontario)     24-­‐37   (Quebec)       10.4-­‐>2.2%  in  Ontario   (1994  vs  2004)     7.4,  9.9%   Quebec   (1994  vs  2004)   17.8,  21.7  %  (1994  vs  2004,   Ontario)   25.2  vs  39.3  (1994  vs  2004,   Quebec)   NR   NR   101   Pancreaticoduodenectomy   221   1895   (1994-­‐ 2004  in   Ontario)     1396   (1994-­‐ 2004  in   Quebec)   1576   Between  January   1995  and  April   2000  (pre-­‐   regionalization)       Between  July   2005  and  July   2009  (post   regionalization)   HV:  >=10/yr   LV:  <10/yr   Pancreatic  ductal   adenocarcinoma     LV:11%   MV:14%   HV:  15%     Ampullary   adenocarcinoma:   LV:  27%   MV:36%   HV:44%     NR   Low:  12.4%   Med:  7.8%   High:  5.9%   V.  High:  2.6%   NR   NR   NR   64   Pancreatectomy,   Pancreaticoduodenectomy   466   3189   Low:  <=5   Med:  6-­‐13   High:14-­‐51   Very  high  :  89-­‐ 104  (only  2   hospitals  in   region)   HV:  >=10   LV:  <10   HV:3.3%  S   LV:7.6%   (UA)   NR   HV:  12  S   LV:14   (median)   NR       P a g e    144     11   Various  pancreatic   resections   1743   39463     HV:  >18   resections/yr   MV:5-­‐18   resections/yr   LV:  <5   resections/yr   103   Pancreatic  Resections   68   842   HV:  >42     MV:22-­‐42   LV:<22   104   Pancreaticoduodenectomy   1772   7229   Very  LV:  0-­‐.99/yr   LV:1-­‐1.99/yr   MV:2-­‐4.99/yr   HV:5+/yr   14   Various  resections   1033   NR   106   Pancreaticoduodenectomy   Centralization   to  5  major   cancer  centres   during  study   period   126   1794   94   Pancreatic  resections   (various)   1000  (approx.)   107   Pancreaticoduodenectomy   Distal  Pancreatectomy   Total  Pancreatectomy     NR       5.9%  overall   LV:  3.3   MV:2.1   (odds  ratio)     Pancreactomy:   LV:9.2%   HV:2.4%   PD:   LV:8.2%   HV:4.8%       3.4%   12.8%   14.4%     All  values  significant   (In  hospital,  30day)   VLV:  (16.1%,12.9%)   LV:  (12.7%,9%)   MV:   (10.1%,  7.3%)   HV   (4.1%,  3.0%)       P<0.001   Went  from  3.5  to  1.8%   during  study  period   NR   NR   NR   NR   32.2   39.9   36.8   NR   NR   VLV:22     LV:21   MV:20   HV:16     P<0.001   NR   NR   NR   NR   VLV:  1-­‐2/yr   LV:  3-­‐5/yr   MV:  6-­‐10/yr   HV:  11-­‐20/yr   VHV:  >20/yr   VLV:  11.4%   LV:  10.2%   MV:  10.7%   HV:  5.9%   VHV:  5%   NR   NR   8370   HV:  avg  >20/yr   LV:  avg  <20/yr   NR   103,222   <3   3-­‐5   6-­‐11   12-­‐23   24-­‐35   HV  :3.1%   LV  :  9.0%     P<0.0001   PD:  4.78   Distal   pancreatectomy:3.84   Total  pancreactomy:   2.60     Negative   association;   plateau  after   10  PDs/  yr     P=0.027   NR   NR   NR   Pancreaticoduodenectomy    >36:  19.76%-­‐48.41%   <36:27.74-­‐51.76%     distal  pancreatectomy   NR   P a g e    145     >36     Odds  ratios  show   higher  likelihood  of   mortality  in  low   volume  hospital   >36:  17.78%-­‐36.3%   Total  pancreatectomy   >36:8.06-­‐35.2%   <36:28.57-­‐57.26     Ranges  are  lowest  and   highest  value  in  category   Ranges  contain  data      for  all   ages,  ranges  for  “<36”   contain  data  for  all  ages  and   surgeon  volumes  up  to  36           3) Liver  Cancer   Table  3a.  Surgeon  Specific  Factors  and  Patient  Outcomes,  Liver  Cancer   Ref  #   Type  of  Resection   48   Liver  lobectomy   Partial   Hepatectomy     Specialty     Volume   Categories  (per     year)   HV:  >96   MV:20-­‐95   LV:  <19   Total  No.  of   surgeons  over   study  period   N  (%)   286  (Mean     Volume  per   surgeon  9.8   operations)   Total  No.  of   patients   over  study   period   N  (%)   2799   Mortality   (%)   [30-­‐day  unless   otherwise   specified]       Morbidity/          Post-­‐ operative   Complications     Length  of   Stay  (days)   Overall  Survival       46.8S   40.8   33.7   Unadjusted  and   adjusted   significant     **No  studies  found  comparing  effects  of  surgeon  specialty  on  outcomes       Table  3b:  Hospital-­‐Specific  Factors  and  Patient  Outcomes,  Liver  Cancer   Study   18       Type  of  Resection   Hepatic  Resection     Total  No.  of   Hospitals     247  (9%  high   volume)   No.  of  Patients   9919  (75%  in   high  volume)   Volume   Categories  (per   year)   >10   1-­‐9   Mortality   [30-­‐day]  (%)   In-­‐hosp   4.6   6.1   (S  on  univariate   analysis,  NS  at   multivariate)   Morbidity/Post-­‐ operative   Complications     Length  of  Stay   (days)   12.1  (1-­‐267)   13.2  (1-­‐188)   Overall  Survival   (5-­‐year)     P a g e    146     17   Segmentectomy,   Lobectomy  and  Extended   lobectomy   855   18  046   VHV:  >70   HV:  36-­‐70   LV:  18-­‐35   VLV:  <18     HV:  >299   MV:  88-­‐298   LV:  <87   0.4%  (OR:.16)S   1.1%  (OR:.52)S    1.3%,  (OR:.70)   1.6%   (OR:  ref)     NR   21.5+16.5   20.5+17.2   21.6+17.2   24+20.5S  longer   NR   48   Liver  lobectomy   Partial  Hepatectomy     90  (Mean   Volume:  31.2)   2799       507   >17per  5  yrs   <16  per  5  yrs     11.3  S+   13.1-­‐14.3   52   606   >15   <15   9.4S+   15.4-­‐22.7   (risk-­‐adjusted)   1.5S   7.9   43.1%  45.1%   34.0%   Unadjusted  signif   greater  in  HV,   adjusted  NS     37   Major  Hepatic  Resections   (Hepatectomy)   138   20   Hepatic  Resections       Hepatic  Resection     16,582       Hepatic  Resection   475   2097     >10   <9   >10   <9   12.7   13.2   (adj  outcomes  signif   better  in  favor  of  LV)     19   38     Median   7  (5-­‐9)S   8  (5-­‐11)   (S  after  adjusting   for  case-­‐mix)     In-­‐hosp:     5.8S   8.9   3.9S   7.6   (OR=.60,  S  after   adjusting  for   case-­‐mix)   + *Overall  5-­‐year  Survival.   Adjusted  for  patient  characteristics       Table  3c:  Hospital  type  and  Patient  Outcomes,  Liver  Cancer   Study   31     Community-­‐based  teaching   hospital  with  a  surgical   residency  training  program                 Hospital  Type   Type  of  Resection   Major  Liver   Resections   (1992-­‐2002)   Volume   Categories   (per    year)   NR   Mortality   (%)   [30-­‐day  unless   otherwise   specified]   Total  No.  of   patients  over   study  period     46  (14  surgeons)   23  by  1  surgeon     Morbidity/Post-­‐ operative   complications   Length  of  stay   (days)   (16/46)  34%   9.7   Overall  Survival   (Years)   36%  (8  of  22   patients)     2  yr:  61%  (20  of  33)   P a g e    147     4) Lung  Cancer   Table  4a:  Surgeon  Specific  Factors  and  Patient  Outcomes,  Lung  Cancer   Ref  #   Type  of  Resection   111   Lobectomy     69   Lung  resections   70   Lobectomy   71   Specialty   Total  No.  of   surgeons   over  study   period   N  (%)   Volume   Categories   (per    year)   Total  No.  of   patients   over  study   period   N  (%)   Thoracic   Cardiothoracic   NR   NR   NR   NR   HV:  >47   LV:  <15   HV:  >17   LV:    <7   HV:  >31   LV:  <22               Lung  resections   NR   LV:≤46   MV:47-­‐131   HV:  132+     377   79   78   Lung  resections     Lung  resections   HV:  >20   HV:>20   HV:>20   HV:  ~58   LV:  ~22     79   Thoracic   Cardiothoracic   General   Thoracic   Cardiothoracic       80   Lobectomy       Pneumonectomy     Thoracic   General       Thoracic   General   NR   NR       NR   NR       Mortality   (%)   [30-­‐day   unless   otherwise   specified]   2.5%   2.7%   5.0S   6.1   0.9S   2.6   Morbidity/          Post-­‐ operative   Complications   Length  of   Stay   (days)   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   LV:  reference   MV:  0.43(0.24– 0.77)  S   HV:  0.26(  0.13– 0.53)S   5.1%   5.2%   6.1%   5.5%   7.7%   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   3.0%   5.3%     11.8%   20.2%   NR   NR       NR   NR   11.8   11.6     11.5   14.9   Overall   Survival     NR   NR   NR   NR   NR   NR       NR     NR   NR   NR   31%   32%     NR   NR      NR   NR     Table  4b:  Hospital-­‐Specific  Factors  and  Patient  Outcomes,  Lung  Cancer   Study   70   Lobectomy     112   113       Total  No.  of   Hospitals     Type  of  Resection   No.  of   Patients     Lung  resection   15   134   47     Lobectomy     37         Volume  Categories   (per  year)   <168   <38   >8.8   <8.8   >37   <19   Mortality   [30-­‐day]  (%)   0.9%S   3.1%   3.5%S   4.9%   3.5   4.3   Morbidity/Post-­‐ operative   Complications   NR   NR   NR   NR   NR   NR     Length  of   Stay  (days)   Overall  Survival   (5-­‐year)   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR     P a g e    148     114     Pneumonectomy     Lung  Resections   34   Lobectomy   115   Lung  Resections   116   Lung  Resections   117   Lung  Resections   36   Pneumonectomy   118   Lung  Resections     119   Pneumonectomy,   Lobectomy,     120   Lung  Resections   121   122           512     NR   NR   79   1806   NR   NR   2   34   107  in-­‐total   NR   NR   206     46  (pre   regionalization)     15(  post   regionalization)   436   NR   Pre-­‐post   regionalization   Pneum:  10.9%  to   5.6%  S     Lobe:  2.2%  to  1.9   %     NR   NR     NR   NR   6.6   5.4   NR   NR   NR   NR   33%   44%   NR   NR   NR   NR   37.5%  S*   43.5%   NR   460   Quintiles   Quintile  1  (1–2)     Quintile  2  (3–6)     Quintile  3  (7–12)     Quintile  4  (13–23   Quintile  5  (≥24)           NR   NR   NR   Lobectomy   926(2007)   855(2008)   19831   Low   (24/Year)   Medium-­‐Low   (25–43/Year)   Medium-­‐High   (44–67/Year)   High   (>68/Year)   Q1:  3.52  (0.92– 13.52)   Q2:  0.85  (0.23– 3.14)   Q3:  0.82  (0.20– 3.30)   Q4:  0.37  (0.10– 1.41)   Q5:  Reference   Low:1.00   Medium-­‐low:   0.68  (0.43–1.08)     Medium-­‐high:   0.82  (0.53–1.28)     High  :0.60  (  0.36– 0.99  )   NR   NR   Lung  Resections   NR   13469   HV:  teaching  facility   (TF);  upper  1/3rd   tertile  by  volume   LV:  non-­‐teaching   facility  (NTF);  lower   2/3  tertile  by  volume   TF:1.1%S   NTF:2.6%S     LVC:  2.7%s   HVC:  1.6%S   NR   Low:  15.9   (15.5–16.3)   Medium   Low:  13.1   (12.7–13.5)   Medium   High:  12.4   (12.0–12.7)   High:  11.5   (11.2–11.8)   NR             822   >41   <8   >19   <2   >35   <5   >66   <9   14-­‐44   <5   >11   <  5   0.3-­‐3.8     8.9   10.6   <4.9%   >6.9%   4.2%S   6.4%   3.5%   6.9%   3.0%S   6.0%   5.2%   7.1%   10.7%   13.8%       NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR       NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR   NR     TF:59.2%   NTF:63.9%S     LVC:  59.3%s   HVC:  63.5%S   P a g e    149     123     Lung  Resections   NR   4878   124   Lung  resections   540   72000   approx.   71   Lung  resections   79   4841   LV:1-­‐135   MV:136-­‐467   HV:468+     125   Lung  resections   330   14456   Developed  linear  and   non-­‐linear  models   126   Lung  resections   NR     LV:  <20   HV:≥20   127   Thoracotomy?   19   2994   128   Lung  Resections   729   11668   LV:  1—43  cases/year   MV:  44—54  cases/   Year   HV:  55  or  more   cases/year   LV:  less  than  90/year   HV:  90+  a  year             average  of   cases/year     VHV:  >20     HV:  11—20   MV:  5—10   LV:0—4     Cases  during  period:   ≤24   25-­‐49   50-­‐99   100+   NR  (  all  data  was   reported  as  “not   significant”)   NR   NR   NR   ≤24:  1.4300   (0.9557–  2.1397)   25-­‐49:  1.1062   (0.7461–  1.6399)   50-­‐99:  1.0809   (0.7121–1.6406)   100+:  reference   LV:  reference   MV:  0.72  (  0.40– 1.30)   HV:  0.68(  0.37– 1.25)   NR   NR   NR   NR   NR   NR   NR   increasing  the  surgical   volume   from  10  to  30   procedures  decreased   the  adjusted  odds  of   inpatient  death  by  24%   (OR=   0.76,  95%  C.I.  0.65– 0.90)  when  measured   using  prior  12  month   volume  and  by  21%  (OR   =0.79,  95%  C.I.  0.67– 0.94)  when  measured   using  average  annual   volume.   NR   NR   NR   NR   NR   NR   NR   LV:  0.4   MV:0.37   HV:0.38   NR   NR   NR   HV:  0.76  (0.56  to   1.03))   LV:  reference   LV:  7.6%   MV:6.6%   HV:  6.7%   LV:  4.8%   HV:3.2%   P a g e    150     5) Ovarian  Cancer   Table  5a:  Surgeon  Specific  Factors  and  Patient  Outcomes,  Ovarian  Cancer   Study   Type  of  Resection   Specialty   55   Oophorectomy   Staging/  Cytoreduction   NR   22   Cytoreduction   (debulking  )     24   Ovarian  cancer   surgeries   Gyn/oncologists   Gyn/obs   General  surgeon   Other   58   Ovarian  cancer   surgeries   Gynecologic     Oncologists,   Gynecologists,  or   General  surgeons   24   Ovarian  cancer  surgery   Gyn/oncologists   Gyn/obs   General  surgeon   other   22    Cytoreduction   (debulking  )     Specialized   Semi  specialized   General   Volume   Categories   (per    year)   Total  No.   of   surgeons   over  study   period   N  (%)   Total  No.  of   patients   over  study   period   N  (%)   LV:  <10     HV:  >10     LV:  <6     MV:  7-­‐12   HV:  >12   LV:  1-­‐2     MV:  3-­‐9   HV:  10+     352     1894   NR   1077   NR   #  cases  per  8-­‐ year  study   period   LV:  1-­‐3   MV:  4-­‐19   HV:  20-­‐61   n/a   n/a   Mortality   (%)   [30-­‐day  unless   otherwise   specified]   Morbidity   Length  of   stay  (days)   Overall   Survival  (%)   (5-­‐year   unless   otherwise   specified)     NR   LV:  4.05%     HV:  1.69%S+     NR   NR   Mean  LV:8.85   HV:8.12     NR   NR   OR:  0.7  (0.5-­‐1)   [HV  vs.  LV]+   3585   LV:  3.36   MV:  2.89   HV  :  2.89   NR   NR   1365   2952   60-­‐day:   LV:  9.2     MV:  8.0   HV:  6.5     NR   LV:  6.1     MV:  4.1     HV:  3.3+   5  year   mortality:   LV:  48   MV:  36   HV:  48   2  year   mortality:   LV:  43.2     MV:  42.9     HV:  39.5     NR   3585   2.53     2.04   8.67   5.28     nr   nr   NR   1077   NR   NR     NR   5  year:   35+   51+   25  (ref)   41   NS  difference     Table  5b:  Hospital-­‐Specific  Factors  and  Patient  Outcomes,  Ovarian  Cancer   Study   55       Type  of  Resection   Oophorectomy   Staging/   Cytoreduction   Total  No.  of   Hospitals     43   No.  of  Patients   1894   Mortality   [30-­‐day]  (%)   Volume  Categories  (per  year)   LV:  <20     HV:  >20     2.83   2.21     Morbidity   NR   Length  of   stay  (days)   Overall  Survival   (5-­‐year)   Mean     8.97   8.01+     NR   P a g e    151     24     Ovarian  cancer   surgery   NR   3815   58   Primary  ovarian   cancer  surgery   423   2952   (65  and  over)   22   Ovarian  cancer   surgery   18  (random   sample)   1077   59   Oophorectomy   1166   36624   158   Overall  treatment   40   2163   159   Gynecological   9   cancer     (all  relevant  organs)     3406   57   Stage  IIIC  and  IV   ovarian  cancer   1430   45929   Annual  average   LV:<9   MV:9-­‐20   HV:21-­‐35   VHV:>35   160   Ovarian  cancer   surgery   249   15131  (ovarian   cancer  patients)   161     Ovarian  cancer   surgery   74   953   Hospitals  grouped  by  volume  over   entire  period   LV:    lower  tertile     MV:  middle  tertile   HV:  highest  tertile     Hospitals  classified  according  to   hessian  hospital  legislation     Primary  care  (28)   Secondary  care  (28)   Tertiary  care(9)   Central  referral  centres(9)         LV:1-­‐15     MV:16-­‐99   HV:100+   #  cases  per  8-­‐year  study  period   LV:  1-­‐3   MV:  4-­‐19   HV:  20-­‐61   LV:  <6     MV:  7-­‐12   HV:  >12   Average  annual  volume  over  study   period:   3  volume  terciles   LV:  <10     MV:  11-­‐20     HV:  >  21       Observing  changes  in  mortality  after   implementation  of  centralization   guidelines         3.7   2.6   0.9   60-­‐day:   8.6   9.6   6.6     NR   NR   NR   NR   5.5   5.7   3.5+   NR       20.4   23.4S   24.6S+   NR   NR   NS  difference     20.1   18.2S   14.7S   NR   NR   Relative  hazard  ratio     (2006)     1996  diagnosis:  1.00     2003  diagnosis:  0.69   (0.56-­‐0.86)  95%  CI   *significant  reduction  in   hazard  ratio  post  2000   Hazard  ratio  (very  high   volume  referent),  95%   C.I:   LV:1.16  (1.09-­‐1.24)  MV:   (1.01-­‐1.15)   HV:  (0.98-­‐1.09)     LV:7.8%   MV:5.5%   HV:5.0%     NR   NR   NR   NR   24.3%   26.1%   28.5%   28.9%   NR   NR   NR   NR   Primary  care:   0%   Secondary   care:14.9%   Tertiary   care:11.8%     Referral   centres:  17.1%   NR   NR   LV:  1.8   MV:  1.6+   HV:  1.5+     HV  lower  risk  of  death   than  LVS   45   45   37   2-­‐year  Mortality:   45.2   41.1   40.4+   LV  lower  risk  of   4-­‐year  survival+   P a g e    152     Table  5c:  Hospital  type  and  Patient  Outcomes,  Ovarian  Cancer   Total  No.  of   patients  over   study  period     2024   Mortality   (%)   [30-­‐day  unless   otherwise  specified]     w/gynecologist   other  cancer   specialist   remaining   3815   LV:<=6  new   cases/yr   MV:  7-­‐12  new   cases  a  year   HV:  >12  new  cases   a  year     District  Hospitals     Central  hospitals   University  referral   centres   Cytoreduction   “primary   surgery”   Volume   Categories  (per     year)   Hospital  Type   Type  of   Resection   56       Referral  centres     compared  to   regional  hospitals   Stage  IIIC  and   IV  ovarian   cancer   NR   24   w/gynecologist   other  cancer   specialist   remaining   Ovarian  cancer   surgery   22   General   Semi  Specialized   Specialized     Screening  and   debulking  of   ovarian  cancer     156   District  Hospitals     Central  hospitals   University  referral   centres   Surgical   treatment  for   ovarian  cancer   164   Low  volume   165   Large  high  volume   centre  (post   regionalization   policy)   Study       Length  of   stay  (days)   Morbidity   Overall  Survival  (Years)     NR   Better  survival  for  center   patients  was  most   noticeable  in  the   first  years  after  the   diagnosis,  and  overall   survival  in  centres   and  other  hospitals   appeared  to  be  equal  after   approximately   5  years”   5  year  survival   w/gynecologist:38%   other  cancer   specialist:45%   remaining:  46%   NR   NR   1077   w/gynecologist:   2.86%     other  cancer   specialist:2.97%     remaining:3.22%           307   NR   NR   NR   All  data  taken  from   one  low  volume   hospital   48   NR   %  with  no   macroscopic   tumour  post  op:   District:  19%   Central:10%   University:  34%   27%   All  patients  survived                       (mean  37.1  months)   Observations  from   the  Gynecologic   Oncologic  centre  in   Aalborg  –  after   regionalization   policy   implemented     107   3.7%   6.5%   8.2  (optimal   surgery)     11.3  (sub   optimal   surgery)   NR   4  year  survival  :  49.2%   Median  46  months   P a g e    153     Appendix  2:  ICD-­‐10  diagnosis  and  intervention  codes     Table:  ICD-­‐10  Diagnosis  Codes   CANCER  TYPE   Esophageal  Cancer   Pancreatic  Cancer   Liver  Cancer   Lung  Cancer   Ovarian  Cancer   ICD-­‐10  DIAGNOSIS  CODES   C150,  C151,  C152,  C153,  C154,  C155,  C158,  C159,  D377   C250,  C251,  C252,  C253,  C254,  C257,  C258,  C259     Primary:  C220,  C221,  C222,  C223,  C224,  C227,  C229,  D376                   Secondary:  C787   C3400,  C3401,  C3409,  C3410,  C3411,  C3419,  C342,  C3430,  C3431,  C3439,  C3480,  C3489,  C3490,  C3491,  C3499,  C390,  C398,  C399,  D381                   C560,  C561,  C569,  C5700,  C5701,  C5709,  C571,  C572,  C573,  C574,  C578,  D391     Table:  ICD-­‐10  Intervention  Codes     Esophageal   Cancer  Surgeries   PROCEDURE  NAME   Minimally  invasive  esophagectomy   Open  esophagectomy   Pancreatic   Cancer  Surgeries   Distal  pancreatectomy   Minimally  invasive  distal  pancreatectomy   Whipple  procedure   Liver  Cancer   Surgeries   Lung  Cancer   Surgeries   Minimally  invasive  liver  resection   Open  liver  resection   Vats  sublobar  resection   Open  sublobar  resection   Vats  lobectomy   Open  lobectomy   Open  pneumonectomy   Ovarian  Cancer   Surgeries       Vats  pneumonectomy   Minimally  invasive  oophorectomy   Open  oophorectomy   INTERVENTION  CODES   1NA89DB,  1NA89FA,  1NA91DB,  1NA91FA,  1NA88DCXXG,  1NA88FCXXG,   1NA87FB,  1NA87FC,  1NA87DC,  1NA87DD,  1NA87EY,  1NA87EZ   1NA87QG,  1NA87QH,  1NA88LBXXG,  1NA88QFXXG,  1NA89LB,   1NA89QF,  1NA90LBXXG,  1NA90LBXXG,  1NA90QFXXG,  1NA91LB,   1NA91QF,  1NA92LBXXF,  1NA92LBXXG,  1NA92QFXXG,  1NA87LD,   1NA87LE,  1NA87QC,  1NA87QD   1OJ87LA,  1OJ87VK,  1OJ87VC   1OJ87DA   1OK87LA,  1OK87VZ,  1OK87WA,  1OK87XN,  1OK91LA,  1OK91XN,   1OK89LA,  1OJ89LA,  1OJ89VZ   1OA87DA   1OA87LA,  1OA87LAAZ   1GR87DA,  1GR87PN,  1GT87DA   1GR87NW,  1GR87QB,  1GT87NW,  1GT87QB   1GR89DA   1GR89NW,  1GR89QB,  1GR91NW,  1GR91NWXXA,  1GR91NWXXG,   1GR91NWXXN,  1GR91QB,  1GR91QBXXA,  1GR91QBXXF,  1GR91QBXXG,   1GR91QBXXN,  1GR91QBXXQ   1GT89NW,  1GT89QB,  1GT91NW,  1GT91NWXXF,  1GT91NWXXG,   1GT91NWXXN,  1GT91NWXXQ,  1GT91QB,  1GT91QBXXF,  1GT91QBXXG,   1GT91QBXXN,  1GT91QBXXQ   1GT89DA   1RB87DA,  1RB89DA,  1RD89DA   1RB87LA,  1RB89LA,  1RD89LA   P a g e    154     Trans-­‐vaginal  oophorectomy   Minimally  invasive  fallopian  tube  resections   Open  fallopian  tube  resection   Vaginal  fallopian  tube  resection   Vaginal  hysterectomy  simple   Vaginal  hysterectomy  radical   Minimally  invasive  hysterectomy  simple   Minimally  invasive  hysterectomy  radical   Open  hysterectomy  simple   Open  hysterectomy  radical   Omentectomy             1RD89RA,  1RB87RA,  1RB89RA   1RF87DA,  1RF89DA   1RF87LA,  1RF89LA   1RF87RA,  1RF89RA   1RM87BAGX,  1RM89CA,  1RM87CAGX   1RM91CA   1RM89AA,  1RM87DAGX,  1RM89DA,  1RM87DAAG   1RM91AA,  1RM91DA   1RM89LA,  1RM87LAGX   1RM91LA   1OT87LA,  1OT87DA   P a g e    155     Appendix  3:  Data  Selection  and  Methodology  for  CIHI  Data  Analysis   The   pan-­‐Canadian   data   (excluding   Quebec)   was   extracted   from   the   Canadian   Institute   of   Health   Information   (CIHI)   Discharge   Abstract   Database   (DAD),   which   contains   hospital   records   for   in-­‐ patient  admissions.  Except  for  liver  cancer,  only  cases  with  primary  cancers  and  associated  surgical   procedure(s)   over   the   9-­‐year   period   (2004-­‐2012)   were   included   in   this   analysis   (See   Appendix   2   ICD-­‐10   diagnosis   codes).   Resections   for   primary   liver   cancer   constitute   one-­‐third   of   all   liver   procedures,   while   the   remaining   procedures   are   performed   for   metastasized   colorectal   tumours.   The   expert   surgeons   identified   high   risk   procedures   within   each   cancer   type.   A   list   of   ICD-­‐10   procedure  codes  used  for  this  analysis  is  presented  in  Appendix  2.     The   outcomes   of   interest   were   in-­‐hospital   mortality   at   discharge,   and   length   of   hospital   stay   for   those  discharged  alive,  up  to  a  maximum  of  30  days.  The  exclusion  of  patients  who  experienced  a   length   of   stay   beyond   thirty   days   was   selected   because   those   patients   would   typically   be   more   complicated   case,   and   are   not   representative   of   the   usual   patient   population.   The   crude   mortality   rates,  averaged  over  the  9  year  period,  are  presented  as  the  simplistic  representation  of  mortality   rates   over   time   in   Canada.   The   low   annual   case   volume   for   smaller   provinces   makes   direct   year-­‐to-­‐ year   comparisons   between   provinces   challenging.   For   instance,   the   crude   mortality   rate   for   a   province  with  annual  case  load  of  20  surgeries  can  increase  from  0%  to  5%  with  one  instance  of  in-­‐ hospital   mortality.   Therefore,   9-­‐year   age-­‐adjusted   mortality   rates   (using   the   Canadian   population   over   the   9-­‐year   period   as   a   standard)   are   presented   for   each   province   to   facilitate   more   fair   comparisons.         • Multivariate   logistic   regression   analysis   was   performed   to   quantify   the   impact   of   patient-­‐ related   factors   (age,   sex,   and   co-­‐morbidities),   procedural   complexities,   year,   and   system-­‐ specific   factors   (hospital   volume,   surgeon   volume,   surgeon   specialty)   on   outcomes   of   interest.  Hospital  volume  was  used  as  a  primary  surrogate  measure  to  assess  the  extent  of   consolidation   of   services   since   surgeon   specialty   is   not   reliably   coded   in   administrative   databases  and  accurate  surgeon  volume  counts  are  obscured  by  instances  of  reused  surgeon   identification  codes  in  multiple  centres.     • The   population   size,   and   hence   the   number   of   surgeries,   differs   vastly   amongst   the   provinces   and   territories   of   Canada.   Given   the   small   sample   size   for   some   provinces,   statistical   significance   of   outcome   disparities   between   provinces   could   not   be   determined   reliably  in  advanced  analysis.  Similarly,  differences  may  exist  between  hospitals  in  both  the   quality   and   organization   of   care   regardless   of   volume.   Multi-­‐level   analysis   was   used   to   account   for   the   differences   between   provinces   (i.e.   health   care   structure,   policies,   population  health  status)  and  hospitals  on  outcomes  of  care.  Advanced  survival  analysis  was   performed   to   assess   the   impact   of   factors   of   interest   on   hospital   length   of   stay.     Where   pertinent,   provincial   outcomes   are   reported   compared   to   their   peers   although   this   was   only   possible   with   Ontario.   As   Ontario   is   the   most   populous   province   with   a   sufficient   sample   size,  its  results  were  compared  with  the  rest  of  Canada  in  multivariate  analysis  in  order  to   benchmark   outcomes   to   a   larger   population   by   evaluating   the   deviations   of   outcomes   in   the   province  relative  to  the  larger  national  mean.   P a g e    156     • • •     The  analysis  of  travel  times  and  evolution  of  hospital  catchment  areas  was  completed  using   Geographic  Information  System  (GIS)  methods  utilizing  ArcGIS  software  and  Census  Canada   data.   GIS   is   a   uniquely   integrative   mix   of   hardware,   software,   and   practices   that   enable   spatial  epidemiology  –  the  geo-­‐visualization  of  the  distribution  and  determinants  of  health   phenomena.   GIS   can   assist   in   identifying   the   distribution   of   health   events   and   health   services,  and  allows  the  application  of  a  spatial  lens  to  the  complex  relationships  between   resources,   procedure   volumes,   and   outcomes.   This   technique   allows   us   to   descriptively   map   and   analyze   the   distribution   and   trends   of   cancer   surgery   while   stratified   for   institutional   volume.   Travel   time   was   calculated   using   the   first   three   digits   of   de-­‐identified   patient   postal   codes   relative   to   the   nearest   surgical   institution,   adjusted   for   road   network   data   such   as   speed  limits,  bodies  of  water  and  traffic  lights.  This  work  has  resulted  in  figures  that  visually   describe  differences  in  health  care  delivery  at  2004-­‐2007and  2011-­‐2012  cohorts.   Age-­‐standardized   resection   rates   were   calculated   using   1991   Canadian   population   as   a   standard.   The   rates   were   presented   as   aggregated   sum   of   number   of   surgeries   in   the   first   and   last   three   years,   divided   by   the   adjusted   provincial   populations   in   the   middle   years   (2005   and   2011,   respectively)   for   each   time   period.   The   denominator,   adjusted   provincial   populations   in   2005   and   2011,   were   multiplied   by   3   in   order   to   obtain   rates   per   person-­‐ years.  For  instance,  the  aggregated  3-­‐year  resection  rate  for  first  3  years  was  calculated  as:     (Number  of  Surgeries  from  2004-­‐2006)  x  100,000   Aggregated  3-­‐year  Resection  Rate=   (Age-­‐adjusted  Provincial  Population  in  2005)  x  3                                                                                                                                     Potential   Years   of   Life   Lost   calculated   using   method   described   by   Urbach   et   al.33   Potential   number  of  hospital  days  were  calculated  using  unadjusted  median  for  the  highest  hospital   volume  category.   P a g e    157     Appendix  4:  In-­‐hospital  Mortality  Analysis     Multivariate,  Multilevel  Regression  Models:   1) Esophageal  Cancer   Dependent  Variable=  Mortality  [Odds  Ration  greater  than  1  means  higher  risk  of  in-­‐hospital  mortality]     Age  (1  unit=10  year  increment)   Charlson  Comorbidity  Score   Fiscal  Year  (1  unit=1  year)   Ontario  vs.  Rest  of  Canada   Hospital  Volume  (1  unit=10  year   increment)   Surgeon  Volume  (1  unit=10  year   increment)   General  Surgeons   (Thoracic  Surgeon=Reference)     Odds  Ratio   p-­‐value   [95%  Confidence  Interval]   2.06   1.07   0.96   0.71   0.60   <0.001   0.026   0.290   0.128   0.030   [1.71-­‐2.48]   [1.01-­‐1.13]   [0.90-­‐1.03]   [0.45-­‐1.11]   [0.37-­‐.95]   1.78   0.094   [0.91-­‐3.5]   1.15   0.696   [0.57-­‐2.3]   Significant  Interactions:     Surgeon  Specialty*  Hospital  Volume   • General  Surgeon       Surgeon  Specialty*  Hospital  Volume   • General  Surgeon         Hospital  Clustering  Effect             2.43   Odds  Ratio   0.011   p-­‐value   [95%  Confidence  Interval]   [1.23-­‐4.83]   0.24   0.019   [0.08-­‐0.79]   Estimate   0.42     p-­‐value   0.0242   [95%  Confidence  Interval]   [0.20-­‐0  .87]   P a g e    158     2) Pancreatic  Cancer   Dependent  Variable=  Mortality  [Odds  Ration  greater  than  1  means  higher  risk  of  in-­‐hospital  mortality]     Age  (1  unit=10  year  increment)   Charlson  Comorbidity  Score   Sex  (ref=Females)   Main  Resection  (ref=Distal   Pancreatectomy)   • Whipple  Procedure     Fiscal  Year  (1  unit=1  year)   Bowel  Resection     Ontario  vs.  Rest  of  Canada   Hospital  Volume  (1  unit=10  year   increment)   Surgeon  Volume  (1  unit=10  year   increment)     2.51   1.88   1.28       1.61   Odds  Ratio   <0.001   0.004   0.154       0.03   p-­‐value   [95%  Confidence  Interval]   [1.85-­‐3.41]   [1.22-­‐2.88]   [0.91-­‐1.80]       [1.04-­‐2.49]   0.96   3.12   1.03   0.82   0.290   <0.001   0.865   0.018   [0.9-­‐1.03]   [1.92-­‐5.05]   [0.69-­‐1.56]   [0.7-­‐0.96]   1.03   0.86   [0.71-­‐1.49]   Significant  Interactions:     Age*Comorbidity  Score           Hospital  Clustering  Effect                   Odds  Ratio   0.92   Estimate   0.30     p-­‐value   [95%  Confidence  Interval]   [0.87-­‐0.98]   p-­‐value   [95%  Confidence  Interval]   [0.13-­‐0.73]   0.011   0.067   P a g e    159     3) Liver  Cancer   Dependent  Variable=  Mortality  [Odds  Ration  greater  than  1  means  higher  risk  of  in-­‐hospital  mortality]     Age  (1  unit=10  year  increment)   Charlson  Comorbidity  Score   Main  Resection  (Minimally   Invasive=ref)   • Open  Liver  Resection   Fiscal  Year  (1  unit=1  year)   Number  of  Liver  Resections/visit     Ontario  vs.  Rest  of  Canada   Hospital  Volume  (1  unit=10  year   increment)   Surgeon  Volume  (1  unit=10  year   increment)     1.65   1.76       2.58   Odds  Ratio   <0.001   <0.001       0.01   p-­‐value   [95%  Confidence  Interval]   [1.46-­‐1.88]   [1.32-­‐2.36]       [1.25-­‐5.31]   0.91   2.59   0.89   0.99   0.025   0.008   0.468   0.61   [0.83-­‐0.99]   [1.27-­‐5.27]   [0.66-­‐1.21]   [0.95-­‐1.02]   0.71   0.009   [0.55-­‐0.92]   Significant  Interactions:     Surgeon  Volume*Fiscal  Year         Odds  Ratio   1.04   p-­‐value   [95%  Confidence  Interval]   [1-­‐1.08]   p-­‐value   [95%  Confidence  Interval]   [0.004-­‐3.42]   0.03       Hospital  Clustering  Effect                   Estimate   0.12   0.38   P a g e    160       4) Lung  Cancer   Dependent  Variable=  Mortality  [Odds  Ration  greater  than  1  means  higher  risk  of  in-­‐hospital  mortality]     Age  (1  unit=10  year  increment)   Charlson  Comorbidity  Score   Sex   Main  Resection  (VATS  Sublobar=ref)   • Open  Sublobar   • VATS  Lobectomy   • Open  Lobectomy   • Pneumonectomy   Fiscal  Year  (1  unit=1  year)   Number  of  Lung  Resections/visit     Ontario  vs.  Rest  of  Canada   Hospital  Volume  (1  unit=10  year   increment)   Surgeon  Volume  (1  unit=10  year   increment)   General  Surgeons   (Thoracic  Surgeon=Reference)       Hospital  Clustering  Effect                     1.82   1.13   1.79     2.05   0.91   1.61   5.62   Odds  Ratio   <0.001   <0.001   <0.001     <0.001   0.704   0.003   <0.001   p-­‐value   [95%  Confidence  Interval]   [1.66-­‐2]   [1.10-­‐1.16]   [1.52-­‐2.11]     [1.5-­‐2.82]   [0.59-­‐1.42]   [1.17-­‐2.23]   [4.02-­‐7.86]   0.97   1.59   0.87   0.97   0.044   0.001   0.222   .046   [0.93-­‐1]   [1.22-­‐2.09]   [0.70-­‐1.08]   [0.95-­‐1]   0.99   0.57   [0.95-­‐1]   1.15   0.248   [0.91-­‐1.45]   Estimate   0.24   p-­‐value   0.004   [95%  Confidence  Interval]   [0.14-­‐0.43]   P a g e    161         5) Ovarian  Cancer   Dependent  Variable=  Mortality  [Odds  Ration  greater  than  1  means  higher  risk  of  in-­‐hospital  mortality]     Age  (1  unit=10  year  increment)   Charlson  Comorbidity  Score   Main  Resection  (Minimally   Invasive=ref)   • Open  Procedures   Fiscal  Year  (1  unit=1  year)   Number  of  Gyne  Resections/visit     Ontario  vs.  Rest  of  Canada   Hospital  Volume  (1  unit=10  year   increment)   Surgeon  Volume  (1  unit=10  year   increment)   Surgeon  Specialty  (Obs/gyne=ref)   • GyneOnc   • General     3.22   2.62       1.33   0.95   0.58   1.27   0.93   Odds  Ratio   <0.001   <0.001       0.776   0.09   <0.001   0.19   0.012   p-­‐value   [95%  Confidence  Interval]   [2.4-­‐4.3]   [1.71-­‐4]       [0.18-­‐9.77]   [0.88-­‐1.01]   [0.46-­‐0.74]   [0.89-­‐1.8]   [0.89-­‐0.98]   1.14   0.148   [0.95-­‐1.35]     1.55   3.38     0.058   <0.001     [0.99-­‐2.44]   [1.96-­‐5.83]   Significant  Interactions:     Age*Comorbidity  Score           Hospital  Clustering  Effect               Odds  Ratio   0.90   p-­‐value   <0.001   Estimate   -­‐-­‐-­‐   [95%  Confidence  Interval]   [0.85-­‐0.96]   p-­‐value   -­‐-­‐-­‐   [95%  Confidence  Interval]   -­‐-­‐-­‐   P a g e    162     Appendix  5:  Survival  Analysis   1) Esophageal  Cancer   Dependent  Variable=  30-­‐day  Length  of  Stay  [Exp(b)  less  than  1  means  higher  chance  of  early  discharge]     Age  (1  unit=10  year  increment)   Sex   Charlson  Comorbidity  Score   Fiscal  Year  (1  unit=1  year)   Main  Resection  (ref=Minimally   Invasive)   • Open  Esophagectomy   Ontario  vs.  Rest  of  Canada   Hospital  Volume  (1  unit=10  year   increment)   Surgeon  Volume  (1  unit=10  year   increment)   General  Surgeons   (Thoracic  Surgeon=Reference)     0.93   1.25   0.97   1.02     0.67   Exp  (b)   0.011   <0.001   0.004   0.158     0.004   p-­‐value   [95%  Confidence  Interval]   [0.88-­‐0.98]   [1.10-­‐1.41]   [0.95-­‐0.99]   [0.99-­‐1.05]     [0.51-­‐0.88]   1.01   1.15   0.91   0.072   [0.82-­‐1.25]   [0.98-­‐1.36]   2.06   <0.001   [1.38-­‐3.10]   1.14   0.133   [0.96-­‐1.34]   Significant  Interactions:     Surgeon  Volume*  Hospital  Volume           Odds  Ratio   0.80   p-­‐value   0.009   95%  CI   [0.68-­‐0.95]   P a g e    163     2) Pancreatic  Cancer   Dependent  Variable=30-­‐day  Length  of  Stay  [Exp(b)  less  than  1  means  higher  chance  of  early  discharge]     Age  (1  unit=10  year  increment)   Charlson  Comorbidity  Score   Sex  (ref=Females)   Main  Resection  (ref=Distal   Pancreatectomy)   • Whipple  Procedure     Fiscal  Year  (1  unit=1  year)   Bowel  Resection     Cholecystectomy   Number  of  Pancreatic  resections/year   Ontario  vs.  Rest  of  Canada   Hospital  Volume  (1  unit=10  year   increment)   Surgeon  Volume  (1  unit=10  year   increment)   0.90   0.98   0.98       0.31   Exp(b)   <0.001   0.003   0.481       <0.001   p-­‐value   [95%  Confidence  Interval]   [0.87-­‐0.93]   [0.97-­‐0.99]   [0.91-­‐1.05]       [0.25-­‐0.39]   1.05   0.48   0.89   0.48   0.71   1.12   <0.001   <0.001   0.028   0.006   <0.001   <0.001   [1.02-­‐1.08]   [0.41-­‐0.57]   [0.80-­‐0.99]   [0.28-­‐0.8]   [0.61-­‐0.83]   [1.06-­‐1.17]   1.34   0.011   [1.06-­‐1.68]   Significant  Interactions:     Surgeon  Volume*  Hospital  Volume                       Odds  Ratio   0.93   p-­‐value   0.001   [95%  Confidence  Interval]   [0.90-­‐0.98]   P a g e    164     3) Liver  Cancer   Dependent  Variable=  30-­‐Day  Length  of  Stay  [Odds  Ration  greater  than  1  means  higher  likelihood  of  shorter  length  of  stay     Age  (1  unit=10  year  increment)   Sex   Charlson  Comorbidity  Score   Main  Resection  (Minimally   Invasive=ref)   • Open  Liver  Resection   Fiscal  Year  (1  unit=1  year)   Secondary  Cancer  (ref=Primary)     Ontario  vs.  Rest  of  Canada   Hospital  Volume  (1  unit=10  year   increment)   Surgeon  Volume  (1  unit=10  year   increment)                               Exp(b)   p-­‐value   [95%  Confidence  Interval]   [0.88-­‐0.92]   [0.94-­‐1.04]   [0.92-­‐0.95]       [0.33-­‐0.55]   0.91   0.98   0.93       0.43         <0.001   0.631             <0.001       <0.001   1.03   1.33       0.005   <0.001   0.88   1.01   0.026   0.511   [1.01-­‐1.05]   [1.24-­‐1.42]     [0.78-­‐0.99]   [0.99-­‐1.01]   1.05   0.010   [1.01-­‐1.09]   P a g e    165     4) Lung  Cancer   Dependent  Variable=  30-­‐day  Length  of  Stay  [Exp(b)  less  than  1  means  higher  chance  of  early  discharge]     Age  (1  unit=10  year  increment)   Charlson  Comorbidity  Score   Sex   Main  Resection  (VATS  Sublobar=ref)   • Open  Sublobar   • VATS  Lobectomy   • Open  Lobectomy   • Pneumonectomy   Fiscal  Year  (1  unit=1  year)   Number  of  Lung  Resections/visit     Ontario  vs.  Rest  of  Canada   Hospital  Volume  (1  unit=10  year   increment)   Surgeon  Volume  (1  unit=10  year   increment)   General  Surgeons   (Thoracic  Surgeon=Reference)                                                               0.89   0.97   0.94     0.45   0.75   0.32   0.27   Exp(b)   <0.001   <0.001   <0.001     <0.001   <0.001   <0.001   <0.001   p-­‐value   [95%  Confidence  Interval]   [0.87-­‐0.90]   [0.96-­‐0.98]   [0.92-­‐0.97]     [0.35-­‐0.58]   [0.66-­‐0.85]   [0.26-­‐0.42]   [0.22-­‐0.33]   1.05   0.85   0.84   1.01   0.003   0.002   0.039   0.012   [1.01-­‐1.07]   [0.77-­‐0.94]   [0.71-­‐0.99]   [1.00-­‐1.02]   1.04   0.017   [1.01-­‐1.08]   1.14   0.065   [0.99-­‐1.31]   P a g e    166     5) Ovarian  Cancer   Dependent  Variable=  30-­‐day  Length  of  Stay  [Exp(b)  less  than  1  means  higher  chance  of  early  discharge]     Age  (1  unit=10  year  increment)   Charlson  Comorbidity  Score   Main  Resection  (Minimally   Invasive=ref)   • Open  Procedures   • Radical  Hysterectomy   • Omentectomy   Fiscal  Year  (1  unit=1  year)   Ontario  vs.  Rest  of  Canada   Hospital  Volume  (1  unit=10  year   increment)   Surgeon  Volume  (1  unit=10  year   increment)   Surgeon  Specialty  (Obs/gyne=ref)   • GyneOnc   • General   Bowel  Resection   Number  of  Gyne  Resections/visit         Exp(b)   p-­‐value   [95%  Confidence  Interval]   [0.84-­‐0.90]   [0.92-­‐0.94]       [0.21-­‐0.32]   [0.07-­‐0.16]   [0.03-­‐0.09]   0.87   0.93       0.26   0.10   0.06   <0.001   <0.001       <0.001   <0.001   <0.001   1.1     0.88   1.03   <0.001   0.167   0.018   [1.04-­‐1.07]     [0.74-­‐1.05]   [1.01-­‐1.06]   0.90   <0.001   [0.87-­‐0.93]     0.90   0.51     0.315   <0.001     [0.73-­‐1.10]   [0.44-­‐.57]                                     0.45   0.95   <0.001   <0.001   [0.44-­‐0.48]   [0.92-­‐0.98]   P a g e    167     Appendix  6:  Citizen  Panel  Methodology       The   citizen   panels   were   convened   in   Autumn   2014   in   three   cities:   Hamilton   (Ontario),   Edmonton   (Alberta),  and  Charlottetown  (Prince  Edward  Island).  These  locations  were  purposefully  selected  to   reflect   different   degrees   of   regionalization   for   complex   cancer   surgeries   and   access   to   surgical   centres   of   excellence.   In   particular,   there   have   been   active   efforts   in   Ontario   to   regionalize   some   complex   cancer   surgeries,   the   creation   of   surgical   centres   of   excellence   and   the   development   of   guidelines   about   minimum-­‐volume   standards   for   surgical   sites   and   surgeons   by   Cancer   Care   Ontario.  In  Alberta,  there  has  been  a  passive  form  of  regionalization  for  some  surgical  procedures   over  the  years.  The  transition  took  place  in  large  part  because  surgeons  preferred  to  provide  care  in   a   team   setting   rather   than   on   their   own.   In   contrast,   very   few   complex   cancer   surgeries   have   been  performed  in  Prince  Edward  Island  (PEI)  over  the  last  9  years,  and  patients  must  travel   to  a  neighbouring-­‐province  to  access  complex  cancer  surgeries  (most  often  Nova  Scotia).   Each  panel  consisted  of  a  one-­‐day  structured  deliberation  facilitated  by  a  member  of  the  research   team   and   included   deliberations   about   the   problem,   three   options   to   address   the   problem,   and   implementation  considerations.  The  brief,  which  was  circulated  one  week  prior  to  each  panel,  was   taken   as   read   for   each   panel,   allowing   the   deliberations   to   build   on   the   pre-­‐circulated   material   to   generate   rich   insights   from   panel   members.   The   citizen   panels   did   not   aim   for   consensus.   However,   it   aimed   to   highlight   areas   of   common   ground   and   differences   of   opinion   among   panel   members   and  (where  possible)  identify  the  values  underlying  different  positions.     The  study  protocol  was  approved  by  the  Hamilton  Integrated  Research  Ethics  Board  at  McMaster   University   and   all   panel   members   provided   voluntary,   informed   consent.   For   those   who   are   interested   in   obtaining   more   information   about   the   citizen   brief,   it   is   freely   available   on   the   McMaster  Health  Forum  website  (www.mcmasterhealthforum.org).   Preparing  the  citizen  brief   We   prepared   the   citizen   brief   through   four   main   steps.   First,   we   convened   a   steering   committee   comprised   of   representatives   from   the   McMaster   Health   Forum,   as   well   as   clinicians   and   researchers  from  McMaster  University  who  were  involved  in  a  broader  program  of  research  about   complex  cancer  surgeries.  The  role  of  the  steering  committee  was  to  provide  guidance  and  expert   advice  across  all  stages  of  the  process.     Second,   in   collaboration   with   the   steering   committee,   we   developed   terms   of   reference   for   the   citizen  brief,  which  provided  a  preliminary  outline  of  the  framing  of  the  problem,  three  options  for   addressing   it,   and   implementation   considerations   (i.e.,   barriers   and   facilitators   for   moving   forward).  We  then  conducted  ten  key  informant  interviews  with  policymakers,  managers  (e.g.,  from   health   regions   and   healthcare   institutions),   stakeholders   (e.g.,   from   provider   associations   or   cancer   societies)   and   researchers   who   were   actively   engaged   in   the   issue   of   complex   cancer   surgeries.   The   terms   of   reference   were   iteratively   revised   based   on   feedback   from   the   key   informants   and   the   steering  committee  and  then  used  to  structure  the  writing  of  the  citizen  brief.   Third,   we   identified,   selected,   appraised   and   synthesized   relevant   research   evidence   about   the   problem,   three   options   (among   many)   to   address   it,   and   implementation   considerations.   Whenever   possible,  we  summarized  research  evidence  drawn  from  systematic  reviews  and  occasionally  from       P a g e    168     single   studies   when   reviews   were   not   identified.   We   identified   published   literature   that   provided   an   understanding   of   the   problem   by   searching   PubMed   to   identify   published   literature.   In   addition,   we   searched   for   grey   literature   about   the   problem   by   reviewing   the   websites   of   a   number   of   Canadian  and  international  organizations.  To  identify  research  evidence  about  the  three  options  in   the   brief,   we   searched   Health   Systems   Evidence   (www.healthsystemsevidence.org),   which   is   a   continuously  updated  database  that  now  contains  more  than  4,000  systematic  reviews  of  delivery,   financial   and   governance   arrangements   within   health   systems.   The   reviews   were   identified   by   searching   the   database   for   ‘cancer’   and   ‘surgery’   in   the   title   and   abstract,   and   by   searching   topic   categories   addressing   features   of   each   of   the   options   (e.g.,   ‘quality   monitoring   and   improvement   systems’   and   ‘regionalization’).   The   searches   were   reviewed   for   relevance   by   the   lead   author   of   the   brief  (FPG).     Fourth,  we  synthesized  key  findings  in  the  form  of  a  citizen  brief.  Specifically,  we  drafted  the  brief  in   such   a   way   as   to   present   concisely   and   in   accessible   language   the   research   evidence.   The   final   version   of   the   brief   consisted   of   a   description   of:   1)   the   context   for   the   panel;   2)   the   problem;   3)   three  options  (among  many)  to  address  it;  4)  implementation  considerations;  and  5)  the  questions   to  be  discussed  during  the  citizen  panels.  The  brief  also  contained  a  separate  insert  for  each  panel,   which  described  particular  features  of  the  local  health  system.  We  then  undertook  a  merit  review   process  for  the  brief,  which  was  reviewed  by  one  public/patient  representative,  one  policymaker,   one   stakeholder,   and   one   researcher   in   order   to   ensure   its   system   relevance,   its   scientific   rigour   and  its  accessibility.   Convening  the  citizen  panels   We   worked   collaboratively   with   the   steering   committee   to   plan   and   convene   the   three   citizen   panels.   Two   strategies   were   implemented   to   recruit   potential   candidates:   1)   targeted   invitations   sent   to   the   AskingCanadiansTM   panel   (a   market-­‐research   panel   of   more   than   600,000   Canadians   who   have   opted-­‐in   to   participate   in   online   surveys);   and   2)   targeted   online   advertisements   via   cancer   organizations   and   societies.   We   selected   10-­‐14   panel   members   for   each   panel   based   on   explicit  criteria  (e.g.,  experiences  as  patients  or  informal/family  caregivers,  experiences  with  cancer   and  surgeries,  types  of  cancer,  gender,  socioeconomic  and  ethno-­‐cultural  background).  We  describe   the  profile  of  panel  members  for  each  panel  in  Appendix  9.  The  study  protocol  was  approved  by  the   Hamilton   Integrated   Research   Ethics   Board   at   McMaster   University   and   all   panel   members   provided  voluntary,  informed  consent.   Each  panel  consisted  of  a  one-­‐day  structured  deliberation  facilitated  by  a  member  of  the  research   team   (JA   or   FPG)   and   included   deliberations   about   the   problem,   three   options   to   address   the   problem,   and   implementation   considerations.   The   brief,   which   was   circulated   one   week   prior   to   each   panel,   was   taken   as   read   for   each   panel,   allowing   the   deliberations   to   build   on   the   pre-­‐ circulated  material  to  generate  rich  insights  from  panel  members.  The  citizen  panels  did  not  aim  for   consensus.   However,   it   aimed   to   highlight   areas   of   common   ground   and   differences   of   opinion   among   panel   members   and   (where   possible)   identify   the   values   underlying   different   positions.     P a g e    169      Appendix  7:  Features  of  the  problem  of  delivering  complex  cancer  surgeries  as  described  in  the  citizen  brief   Issue   Cancer  represents  a   significant  burden  on   individuals,  the  health  system   and  society   Factors  contributing  to  the  issue   • With  population  growth  and  aging,  the  number  of  new  cases  of  cancer  in  Canada  is  expected  to  increase.   • Since  surgery  is  the  primary  treatment  option  for  certain  high-­‐risk  cancers,  the  costs  associated  with  cancer   surgeries  are  expected  to  rise  over  time.   Patients  in  need  of  complex   cancer  surgeries  and  their   families  face  a  difficult   journey   • Patients  and  families  have  to  make  complex  and  potentially  life-­‐changing  decisions  (e.g.,  undergoing  surgery  or   not,  undergoing  surgery  at  a  local  low-­‐volume  hospital  or  travelling  to  a  high-­‐volume  hospital  with  the  hope  of   better  outcomes),  but  information  to  guide  such  decisions  is  often  not  readily  available.   • The  long-­‐term  outlook  for  those  diagnosed  with  any  of  these  five  types  of  cancer  is  generally  quite  intimidating,   since  a  significant  number  of  patients  will  die  despite  curative-­‐intent  surgery.   • Delivery  arrangements   o There  are  disparities  in  access  to  complex  cancer  surgeries  across  Canada  and  even  in  settings  that  have   access  there  are  disparities  in  the  availability  of  expertise  to  conduct  these  surgeries.  In  some  provinces,  the   surgeries  are  being  delivered  in  any  hospital  setting,  without  restriction.  There  have  been  efforts  to   regionalize  (or  centralize)  some  of  these  complex  cancer  surgeries  into  high-­‐volume  centres  (i.e.,  centres   providing  surgical  care  to  many  patients)  in  a  few  provinces.   o There  is  a  lack  of  support  for  the  informal  and  family  caregivers  of  the  patients  undergoing  complex  cancer   surgery.   • Financial  arrangements   o The  predominant  funding  model  for  Canadian  hospitals  (i.e.,  global  budgets)  provides  little  incentive  for   hospitals  to  focus  on  increasing  surgery  volumes,  improving  quality  of  care,  or  coordinating  care  across   facilities  and  sectors.     • Governance  arrangements   o There  is  minimal  regulation  as  to  which  procedures  surgeons  can  perform  within  their  specialty  area,  or   how  frequently  they  need  to  perform  these  procedures  to  ensure  their  surgical  skills  remain  up  to  date.   o There  is  minimal  regulation  as  to  which  procedures  can  be  performed  in  hospitals  or  how  frequently  they   need  to  perform  them  to  ensure  that  quality  remains  high.  In  addition,  most  hospitals  are  regulated  by   legislation  that  establishes  an  appeal  process  for  doctors  who  feel  aggrieved  by  decisions  made  by  hospital   boards,  which  can  make  it  difficult  to  change  where  different  types  of  surgical  procedures  can  be  performed.   o There  is  a  lack  of  coordinated  effort  among  all  stakeholders  to  improve  complex  cancer  surgeries  across  the   country  (e.g.,  no  agreed  set  of  quality  indicators  at  the  pan-­‐Canadian  level).   The  health  system  is  not   currently  designed  to  provide   optimal  care  for  such  patients       P a g e    170     Appendix  8:  Three  options  for  improving  the  delivery  of  complex  cancer  surgeries  in  Canada Option   1  -­‐  Encourage  the  local  adoption  of  quality-­‐ improvement  initiatives  to  improve  the   delivery  of  complex  cancer  surgeries  where   they  are  now  being  provided   Option  focus  and  elements   The  first  option  aims  to  encourage  healthcare  providers  (e.g.,  surgeons,  nurses  and  others)  and   managers  to  adopt  quality-­‐improvement  initiatives  in  local  hospitals  in  order  to  improve  the   delivery  of  high-­‐risk  and  resource-­‐intensive  cancer  surgeries.  This  option  assumes  that   healthcare  providers  and  hospital  managers  will  adopt  quality-­‐improvement  initiatives  without   the  need  for  supports,  incentives  or  directives  to  change  behaviour,  and  without  the  need  for   regulatory  changes  about  where  (and  by  whom)  these  cancer  surgeries  can  be  provided.  In  sum,   this  option  proposes  a  locally  driven  approach  to  improve  the  delivery  of  cancer  surgeries.     • There  are  a  large  number  of  quality-­‐improvement  strategies  that  can  be  driven  locally  by   healthcare  providers  and  hospital  managers,  for  example:   o Promoting  audit  and  feedback;   o Promoting  the  use  of  clinical  decision  support  systems;     o Promoting  continuing  medical  education;  and   o Implementing  enhanced  recovery  programs.   • The   second  option  aims  to  implement  province-­‐wide*  quality-­‐improvement  initiatives  to   2  -­‐  Implement  province-­‐wide  quality-­‐ improve  the  delivery  of  complex  cancer  surgeries.  Like  the  first  option,  this  option  is  not   improvement  initiatives  to  improve  the   intended  to  change  where  and  by  whom  these  cancer  surgeries  are  being  provided.  However,  in   delivery  of  complex  cancer  surgeries  where   contrast  to  the  first  option,  this  option  proposes  a  top-­‐down  approach  to  quality  improvement   they  are  now  being  provided   and  assumes  that  healthcare  providers  and  hospital  managers  can  achieve  significant   improvements,  but  that  they  need  appropriate  support,  incentives  and  directives  to  do  this.   • These  province-­‐wide  quality-­‐improvement  initiatives  could  take  different  forms,  for  example:   o Developing  provincial  guidelines  and  standards  for  these  cancer  surgeries;   o Implementing  pay-­‐for-­‐performance  for  hospitals;     o Developing  or  expanding  supports  for  patients  and  families;  and   o Establishing  requirements  for  reporting  to  the  public  about  quality  indicators  and  other   performance  measures.     3  –  Regionalize  complex  cancer  surgeries  into   • The  third  option  aims  to  regionalize  complex  cancer  surgeries  into  designated  surgical  centres  of   excellence.  This  option  includes  efforts  to  change  the  structure  of  the  health  system  and  to  set   designated  surgical  centres  of  excellence   province-­‐wide  standards  to  support  the  regionalization  of  complex  cancer  surgeries.  This  option   assumes  that  changes  to  who  performs  the  surgeries  and  where  they  are  performed  will  be   needed  to  improve  the  delivery  of  care.  This  option  proposes  a  top-­‐down,  province-­‐wide   approach  to  design  and  implement  changes  to  who  does  what  and  where  across  the  province.  As   with  option  2,  this  option  can  include  developing  or  expanding  supports  for  patients  and   families.   *  For  small  provinces,  province-­‐wide  should  be  taken  to  mean  across  the  small  province  or  across  both  the  small  province  and  a  neighbouring,  larger   province  to  which  referrals  are  frequently  made.       • P a g e    171     Appendix  9:  Profile  of  panel  members   Hamilton  panel  (Ontario)   Edmonton  panel  (Alberta)   Charlottetown  panel  (PEI)   How  many  panel  members?   14   Hamilton-­‐Niagara-­‐Haldimand-­‐Brant   Local  Health  Integration  Network   13   Within  two  hours  driving   distance  from  Edmonton   11   PEI  (10)  and  rural  Nova  Scotia   (1)   Participant  Age   25-­‐44   45-­‐64   65+   21%   36%   43%   7%   54%   39%   0%   70%   30%   Participant   Gender   Men   Women   50%   50%   38%   62%   64%   36%   What  was  the   educational   level  of  panel   members   No  schooling   Elementary  school   High  school   Community  college   Technical  school   Bachelor’s  degree   Post-­‐graduate  training   0%   0%   14%   14%   14%   50%   7%   0%   8%   17%   33%   0%   42%   0%   0%   0%   20%   20%   30%   30%   0%   What  was  the   work  status  of   panel   members?   Self-­‐employed   Working  full-­‐time   Working  part-­‐time   Unemployed   Retired   Student   Homemaker   Disabled   0%   29%   7%   0%   36%   7%   0%   21%   10%   20%   20%   0%   40%   0%   0%   10%   10%   20%   10%   0%   50%   0%   0%   10%   What  was  the   income  level  of   panel   members?   Less  than  $20,000   $20,000  and  $40,000   $40,000  and  $60,000   $60,000  and  $80,000   More  than  $80,000   Preferred  not  to  answer   21%   29%   21%   0%   14%   15%   17%   25%   33%   0%   8%   17%   10%   30%   20%   20%   10%   10%   Where  were  they  from?       1.1" N. CANADIAN PARTNERSHIP AGAINST CANCER .- PARTENARIAT CANADIEN CONTRE LE CANCER