jcp 1 FOOD AND DRUG ADMINISTRATION CENTER FOR DRUG EVALUATION AND RESEARCH Joint Meeting of the Anesthetic and Life Support Drugs Advisory Committee and the Drug Safety and Risk Management Advisory Committee Meeting Announcement Thursday, October 21, 2010 8:30 a.m. to 4:07 p.m. Hilton, Washington D.C./North Gaithersburg 620 Perry Parkway Gaithersburg, Maryland PRECISE REPORTING, LLC jcp 2 PRESENT: Jeffrey R. Kirsh, M.D., Chair Kalyani Bhatt, Designated Federal Officer, ALSDAC ANESTHETIC AND LIFE SUPPORT DRUGS ADVISORY COMMITTEE MEMBERS (Voting) Jeffrey R. Kirsch, M.D., Chair Professor and Chair, Department of Anesthesiology and Perioperative Medicine, Associate Dean for Clinical and Veterans Affairs, Oregon Health & Science University, Portland, Oregon Randall Flick, M.D., M.P.H., Assistant Professor of Anesthesiology Mayo Clinic, Rochester, Minnesota Osemwota A. Omoigui, M.D., Consumer Representative, Division of Inflammation and Pain, Los Angeles Pain Clinic, Hawthorne, California Daniel Zelterman, Ph.D., Professor and Acting Division Head, Division of Biostatistics, Epidemiology and Public Health, Yale University School of Medicine, New Haven, Connecticut INDUSTRY REPRESENTATIVE (Non-Voting) Mark P. Fletcher, M.D., FAAAAI, Acting Industry Representative, MPF BioPharma Consultants, LLC, Clinical Drug Development Consulting, Charlottesville, Virginia PRECISE REPORTING, LLC jcp 3 PRESENT: (CONTINUED) DRUG SAFETY AND RISK MANAGEMENT ADVISORY COMMITTEE MEMBERS (Voting) Elaine H. Morrato, Dr.P.H., Assistant Professor, Department of Pediatrics, University of Colorado Denver, Denver, Colorado Sidney M. Wolfe, M.D., Consumer Representative, Director, Health Research Group, Public Citizen, Washington, District of Columbia Lewis Nelson, M.D., Director, Fellowship in Medical Toxicology, New York University School of Medicine, New York, New York TEMPORARY VOTING MEMBERS Warren Bickel, Ph.D., M.D, Director, Arkansas Center for Addiction Research, Department of Anesthesia, Brigham & University of Arkansas for Medical Sciences, Little Rock, Arkansas Warren B. Bilker, Ph.D., Professor of Biostatics, University of Pennsylvania, Philadelphia, Pennsylvania Richard Denisco, M.D., Medical Officer, Pain/Addiction Medicine, National Institutes of Health, National Institute of Drug Abuse, Division of Epidemiology, Services, and Prevention, Bethesda, Maryland PRECISE REPORTING, LLC jcp 4 PRESENT: (CONTINUED) TEMPORARY VOTING MEMBERS (CONT.) Robert Kerns, Ph.D., National Program Director for Pain Management, Yale University School of Medicine, VA Connecticut Health Care System, West Haven, Connecticut Susan Krivacic, Patient Representative, Austin, Texas John Mendelson, M.D., Senior Scientist, Addiction and Pharmacology Research Laboratory, California Pacific Medical Center Research Institute, St. Luke’s Hospital, San Francisco, California Edward Michna, M.D., Director, Pain Trial Center, Women’s Hospital, Harvard Medical School, Boston, Massachusetts Cynthia Morris-Kukoski, Pharm.D., Forensic Examiner, Department of Justice/Federal Bureau of Investigation, Laboratory/Chemistry Unit, Washington, District of Columbia Sharon Walsh, Ph.D., Robert Straus Behavioral Research Building, University of Kentucky, Lexington, Kentucky PRECISE REPORTING, LLC jcp 5 PRESENT: (CONTINUED) FDA MEMBERS (Non-Voting) Ellen Fields, M.D., M.P.H, Team Leader, Division of Anesthesia, Medical Officer, Division of Anesthesia and Analgesia Products (DAAP), CDER, FDA Sharon Hertz, M.D., Deputy Director, Division of Anesthesia and Analgesia Products (DAAP), CDER, FDA Larissa Lapteva, M.D., Medical Officer, Division of Anesthesia and Analgesia Products (DAAP), CDER, FDA Bob Rappaport, M.D., Director, Division of Anesthesia and Analgesia Products (DAAP), CDER, FDA Judy Staffa, Ph.D., R.P.H., Acting Director, Division of Epidemiology (DEPI), CDER, FDA Mary Willy, Ph.D. Deputy Director, Division of Risk Management, Office of Surveillance and Epidemiology (OSE), CDER, FDA PRECISE REPORTING, LLC jcp 6 A-G-E-N-D-A Page Call to Order, Jeffrey R. Kirsch, M.D. 9 Introduction of Committee, Jeffrey R. Kirsch, M.D. 10 Conflict of Interest Statement, Kalyani Bhatt 14 Opening Remarks, Bob Rappaport, M.D. 18 FDA Presentations: Nature of the Problem of Prescription Opioid Misuse and Abuse Overview of the Risk of Abuse and Regulatory Discussions to Date to Reduce Abuse of Opioid Analgesics, Larissa Lapteva, M.D. 24 Premarketing Assessment of Abuse-deterrent Formulations, James Tolliver, Ph.D. 39 Abuse of Marketed Opioid Analgesics and Their Contribution to the National Problem of Drug Abuse, Len Paulozzi, M.D., M.P.H. 52 Clarifying Questions, 68 Data Resources and Metrics Used to Assess Prescription Opioid Misuse and Abuse Designing Observational Studies on Drug Abuse, James C. (Jim) Anthony, Ph.D. 75 Substance Abuse and Mental Health Services Administration: Resources and Methods, Albert Woodward, Ph.D., M.B.A. 94 PRECISE REPORTING, LLC jcp 7 A-G-E-N-D-A (CONTINUED) Page FDA Presentations: {continued) Available Data Resources to Assist in Measuring Abuse Behaviors, Patterns, and Outcomes, Catherine Dormitzer, Ph.D. 105 Clarifying Questions, 119 Study Designs to Assess Prescription Drug Abuse Design Considerations in Epidemiological Studies of Abuse-deterrent Opioids, Cynthia Kornegay, Ph.D. 134 Statistical Considerations for Epidemiological Studies of Abuse-Deterrent Formulations, Stephine Keeton, Ph.D. 143 Lunch, 154 Clarifying Questions, 155 Industry Presentations: Purdue Pharma, LP Introduction, Craig Landau, M.D. 177 Overview of Rationale and Study Program Paul Coplan, DsC 190 Overdose Rates in OxyContin Patients and Non-Patients at Kaiser Permanente, Nancy Perrin, Ph.D. 204 Exposures Reported to Poison Centers, Rick Dart, M.D., Ph.D. 214 OxyContin Abuse Among Entrants to Substance Abuse Treatment Programs Theresa Cassidy, M.P.H. 223 PRECISE REPORTING, LLC jcp 8 A-G-E-N-D-A (CONTINUED) Industry Presentation: (continued) Using Surveys to Assess the Impact of a New Formulation of OxyContin, Howard Chilcoat, ScD 228 Law Enforcement Events in the Drug Diversion Program at RADARS System, Rick Dart, M.D., Ph.D. 236 Doctor-Shopping for OxyContin as Measured by Prescription Monitoring Programs, Paul Coplan, DSc 240 Internet Discussion About Reformulated OxyContin Use, Theresa Cassidy, M.P.H. 244 Changes in Abuse Patterns in a Cohort of People Abusing OxyContin in Rural Kentucky, Carl Leukefeld, DSW 250 Summary and Conclusions, Paul Coplan, DSc 255 Clarifying Questions, 258 Adjourn, 301 PRECISE REPORTING, LLC jcp 9 1 P R O C E E D I N G S 2 (8:00 AM) 3 Call to Order 4 DR. KIRSCH: Good morning, everybody. If 5 everyone could please take their seats, we can get 6 started. 7 silence your cell phones, BlackBerrys, and other devices 8 if you have not already done so. 9 going around the table and introducing ourselves. 10 I’d like to remind everyone present to please 11 We’ll get started by Before we do that, I have two other short announcements. 12 First, as we begin our deliberation for the 13 members of the committee, I wanted to remind you to try 14 to be succinct in your comments and try not to be 15 redundant. 16 prerogative to curtail discussion that I think is 17 redundant or not directed at the question at hand. 18 19 I will, at this meeting, take the Chair’s I’d also like to take this opportunity to wish Dr. Willy a happy birthday. 20 (Applause.) 21 DR. KIRSCH: 22 Happy birthday. We’ll start the introductions with our Industry Representative, Dr. Fletcher. PRECISE REPORTING, LLC jcp 10 1 Introductions 2 DR. FLETCHER: Good morning. I’m Dr. Mark 3 Fletcher. 4 Anesthetics Advisory Committee, and I am a non-voting 5 member. 6 I’m the acting representative for the DR. MENDELSON: And I'm John Mendelson. 7 internist and clinical pharmacologist from San 8 Francisco, and this is my first meeting. 9 DR. NELSON: Lewis Nelson. I’m an emergency 10 physician and a medical toxicologist from New York 11 University School of Medicine. 12 13 DR. KRIVACIC: I’m an I'm Susan Krivacic, and I’m a patient representative from Austin, Texas. 14 DR. WOLFE: I'm Sid Wolfe. I’m an internist. 15 I’m with the Public Citizen Health Research Group on the 16 Drug Safety and Risk Management Advisory Committee. 17 DR. BICKEL: Warren Bickel, Center for 18 Addiction Research, University of Arkansas for Medical 19 Sciences. 20 DR. KERNS: Good morning. I’m Bob Kerns. I’m 21 professor of Psychiatry, Neurology, and Psychology at 22 Yale University, and I’m also with the VA as director of PRECISE REPORTING, LLC jcp 11 1 the Pain Management Program for VA and director of a 2 pain research center at VA Connecticut. 3 4 5 DR. BILKER: Warren Bilker. I’m Professor of Biostatics at the University of Pennsylvania. DR. MORRATO: Good morning. Elaine Morrato, 6 and I’m an epidemiologist from the Colorado School of 7 Public Health and Health Systems Management and Policy. 8 9 10 DR. FLICK: Randall Flick, pediatric anesthesia, intensive care, Mayo Clinic. MS. BHATT: Good morning. I’m Kalyani Bhatt. 11 I’m with the Division of the Advisory Committee, 12 Consultants Management. 13 DR. KIRSCH: I'm Jeff Kirsch. I’m the chair 14 of the Department of Anesthesiology and Peri-Operative 15 Medicine at Oregon Health Sciences University, and the 16 associate dean for Clinical and Veterans’ Affairs. 17 18 19 20 21 22 DR. ZELTERMAN: I'm Dan Zelterman, professor of Biostatistics at Yale University. DR. MICHNA: Ed Michna, anesthesia, pain management, Brigham and Women’s Hospital in Boston. DR. WALSH: Good morning. I’m Sharon Walsh. I’m the director of the Center on Drug and Alcohol PRECISE REPORTING, LLC jcp 12 1 Research and a professor in behavioral science and 2 psychiatry at the University of Kentucky in Lexington. 3 DR. OMOIGUI: Good morning. I’m Osemwota 4 Omoigui. 5 medical director, L.A. Pain Clinic, Hawthorne, 6 California, and also the consumer rep. 7 I’m an anesthesiologist and pain specialist, DR. WILLY: I'm Mary Willy. I’m deputy 8 director, Division of Risk Management in the Office of 9 Surveillance and Epidemiology. 10 DR. STAFFA: Good morning. I’m Judy Staffa. 11 I’m the acting director of the Division of Epidemiology 12 in the Office of Surveillance and epidemiology at FDA. 13 14 DR. FIELDS: I’m Ellen Fields, clinical team leader, Division Anesthesia and Analgesia. 15 DR. LAPTEVA: Good morning, I’m Larissa 16 Lapteva. 17 of Anesthesia and Analgesia Products. 18 19 20 21 22 I’m deputy director for Safety in the Division DR. HERTZ: I'm Sharon Hertz, deputy director, Division of Anesthesia and Analgesia. DR. RAPPAPORT: Bob Rappaport, director, Division of Anesthesia and Analgesia. DR. KIRSCH: Thank you. For topics such as PRECISE REPORTING, LLC jcp 13 1 those being discussed at today’s meeting, there are 2 often a variety of opinions, some of which are quite 3 strongly held. 4 a fair and open forum for discussion of these issues, 5 and that individuals can express their views without 6 interruption. 7 will be allowed to speak into the record only if 8 recognized by the Chair. 9 productive meeting. 10 Our goal is that today’s meeting will be Thus, as a gentle reminder, individuals We look forward to a In the spirit of the Federal Advisory 11 Committee Act and the Government and the Sunshine Act, 12 we ask that the Advisory Committee members take care 13 that their conversations about the topic at hand take 14 place in the open forum of the meeting. 15 that members of the media are anxious to speak with the 16 FDA about the proceedings. 17 from discussing the details of this meeting with the 18 media until its conclusion. 19 We are aware However, FDA will refrain For the convenience of the media 20 representatives, I would like to identify the FDA press 21 contact, Shelly Burgess. 22 stand. If you are present, please PRECISE REPORTING, LLC jcp 14 1 PARTICIPANT: Shelly’s not here. 2 DR. KIRSCH: 3 Also, the committee is reminded to please Thank you. 4 refrain from discussing the meeting topic during breaks 5 or lunch. 6 7 8 9 Thank you. I’ll pass it to Kalyani, who will read the Conflict of Interest statement. Conflict of Interest Statement MS. BHATT: The Food and Drug Administration 10 is convening today’s Joint Meeting of the Anesthetic and 11 Life Support Drugs and Drug Safety and Risk Management 12 Advisory Committee under the authority of the Federal 13 Advisory Committee Act of 1972. 14 temporary voting members of the committee or special 15 government employees or regular federal employees from 16 other agencies and are subject to Federal Conflict of 17 Interest Laws and regulations. 18 All members and The following information on the status of 19 this committee’s compliance with federal ethics and 20 conflict of interest laws covered by, but not limited 21 to, those found at 18 USC Section 208 and Section 712 of 22 the Federal Food, Drug, and Cosmetic Act is being PRECISE REPORTING, LLC jcp 15 1 provided to participants in today’s meeting and to the 2 public. 3 FDA has determined that members and temporary 4 voting members the committee are in compliance with 5 Federal Ethics and Conflict of Interest laws. 6 USC Section 208, Congress has authorized FDA to grant 7 waivers to special government employees and regular 8 federal employees who have potential financial conflicts 9 when it is determined that the agency’s need for a Under 18 10 particular individual’s services outweighs his or her 11 potential financial conflict of interest. 12 Under Section 712 of the Food, Drug, and 13 Cosmetic Act, Congress has authorized FDA to grant 14 waivers to special government employees and regular 15 federal employees with potential financial conflicts 16 when necessary to afford the committee essential 17 expertise. 18 Related to the discussion of today’s meeting, 19 members and temporary voting members of the committees 20 have been screened for potential financial conflicts of 21 their own, as well as those imputed to them, including 22 those of their spouses or minor children and for PRECISE REPORTING, LLC jcp 16 1 purposes of 18 USC Section 208, their employers. 2 interests may include investments, consulting, expert 3 witness testimony, contracts, grants, CRADAs, teaching, 4 speaking, writing, patents and royalties, and primary 5 employment. 6 These Today’s agenda involves discussion of the 7 design of post-marketing studies of OxyContin, 8 controlled release tablets manufactured by Purdue Pharma 9 and Embeda, extended release capsules manufactured by 10 Alpharma Pharmaceuticals and King Pharmaceuticals, 11 approved for the management of moderate to severe pain 12 when a continuous, around the clock opiate analgesic is 13 needed for an extended period of time. 14 marketing studies are intended to epidemiological or 15 observational studies that will assess a known, serious 16 risk of these products, and whether product-specific 17 properties which are intended to discourage misuse and 18 abuse actually result in a decrease in the risk of 19 misuse and abuse and their consequences. 20 The post- This is a particular matter’s meeting during 21 which specific matters related to Purdue’s OxyContin 22 controlled release tablets and Alpha and Kings Embeda PRECISE REPORTING, LLC jcp 17 1 extended-release capsules will be discussed. 2 Committee members and temporary voting 3 members, not conflict of interest waivers were issued in 4 connection with this meeting. 5 we encourage all standing committee members and 6 temporary voting members to disclose any public 7 statements that they have made concerning the product at 8 issue. 9 To ensure transparency, With respect to FDA’s invited industry 10 representative, we would like to disclose that Dr. 11 Mark Fletcher is participating in this meeting as a non- 12 voting industry representative, acting on behalf of 13 regulated industry. 14 is to represent industry in general and not any 15 particular company. 16 pharmaceutical industry consultant. 17 Dr. Fletcher’s role at this meeting Dr. Fletcher is an independent We’d like to remind members and temporary 18 voting members that if the discussions involve any other 19 products, firms, or issues not already on the agenda for 20 which an FDA participant has a personal or imputed 21 financial interest, the participants need to exclude 22 themselves from such involvement, and their exclusion PRECISE REPORTING, LLC jcp 18 1 will be noted for the record. 2 participants to advise the committees of any financial 3 relationships that they may have with any firms at 4 issue. 5 Thank you. 6 DR. KIRSCH: 7 I’d like to recognize Dr. Rappaport. 8 9 FDA encourages all Opening Remarks DR. RAPPAPORT: Good morning. Dr. Kirsch, 10 members of the anesthesia and life support drugs, and 11 the Drug Safety and Risk Management Advisory Committees, 12 invited guests, thank you for your participation in this 13 important meeting. 14 Many of you have attended some or perhaps all 15 of the numerous advisory committee meetings that we have 16 convened in the past few years to discuss applications 17 for opioid drug products that have been formulated to 18 provide some degree of abuse-deterrents. 19 one of the biggest hurdles we have to face as we review 20 these applications and decide on appropriate labeling 21 for their abuse deterrent properties is how to measure 22 their actual impact on abuse in the community. PRECISE REPORTING, LLC As you know, jcp 19 1 For nearly a decade, we have been quite clear 2 that we would not allow language stating that a product 3 is abuse-deterrent into the labels without documentation 4 that the availability of the purportedly abused 5 deterrent product had actually reduced abuse at the 6 community level. 7 manufacturers to make unsubstantiated, promotional 8 claims that could lead to the same types of 9 misconceptions which resulted when OxyContin was first To do otherwise would permit the 10 marketed. Those misconceptions that OxyContin was less 11 likely to be abused and was less addictive than other 12 potent opioid drug products were based on just a few 13 simple words in the approved label and they played a key 14 role in the public health crisis of abuse, overdose, and 15 addiction that has plagued our communities every since. 16 Now that we have approved opioid drug products 17 with features that are intended to limit their abuse, it 18 has become even more critical for us to define a clear 19 direction for the manufacturers regarding the regulatory 20 requirements for establishing that a product has 21 demonstrated actual abuse-deterrents. 22 must provide a scientifically and clinically sound PRECISE REPORTING, LLC To do this, we jcp 20 1 program of studies that will stand as the evidentiary 2 support for a regulatory determination. 3 But, as has been widely acknowledged in 4 numerous public meetings and in the pertinent medical 5 literature, the available databases used to track abuse 6 were not designed for this challenge, but rather to 7 detect signals of abuse. 8 which databases and which study designs would provide 9 the best quality data for longitudinal tracking of abuse And no clear paradigm for 10 in the community has been established, though many 11 different academic and other centers have been working 12 hard to address the challenge. 13 Today, we’ve brought together leading experts 14 from a number of disciplines that are key to the 15 development of a study or set of studies that we hope 16 will provide us with a foundation upon which we can 17 provide the guidance that is needed to support continued 18 development of these important drug products and that 19 will allow us to fulfill our regulatory mandate of 20 making decisions that are based on sound science and 21 that are not arbitrary and capricious. 22 You will be hearing from a number of FDA PRECISE REPORTING, LLC jcp 21 1 scientists, as well as experts from other government 2 agencies and academia. 3 from King Pharmaceuticals and Purdue Pharma will present 4 their proposals for studies to track abuse levels after 5 the introduction of their own novel products into the 6 market. 7 In addition, representatives While we are not asking to judge those 8 proposals today, as they have provided them in early 9 draft form and without the benefit of agency feedback, 10 we are asking you to consider the elements of these 11 proposals as part of your discussions and in providing 12 your recommendations to us today. 13 The development of a standard by which the 14 agency can judge whether a new product has actually 15 impacted abuse in the community is a significant 16 challenge. 17 that you recommend for us in this endeavor will require 18 testing and validation, and, over time, this methodology 19 will likely change and grow. 20 beginning of this effort, and we may not have a true 21 gold standard for many years. 22 As such, we recognize that these studies We are just at the But we do have to begin someplace if we’re PRECISE REPORTING, LLC jcp 22 1 going to encourage and advance the development of abuse- 2 deterrent opioids. 3 wrote back in 1764, “the perfect is the enemy of the 4 good.” 5 path forward for today and perhaps for tomorrow. So, please remember that as Voltaire And let’s work together to come up with the best 6 Again, thank you for taking your time from 7 your busy schedules to participate in this important 8 process. 9 I do have one additional matter to bring to 10 your attention today before we begin the presentations. 11 Yesterday, Dr. Wolfe sent us an e-mail requesting that 12 we provide you with a copy of a warning letter that FDA 13 issued to King Pharmaceuticals regarding a video news 14 release for Embeda that we considered false or 15 misleading. 16 find it with your package. 17 We have provided that document and you will Warning letters are a type of enforcement 18 action that are used by FDA to correct promotional 19 activities that are not balanced in presenting the 20 benefits and risks of the drug or refer to unapproved 21 off-label uses. 22 in the background package for this meeting, and We did not include the warning letter PRECISE REPORTING, LLC jcp 23 1 generally do not include them in the background package 2 for advisory committee meetings since enforcement 3 actions are not the type of scientific issues for which 4 we seek your advice. 5 Also, today’s meeting is not specific to 6 Embeda. Rather, we are seeking your scientific and 7 clinical input on how best to measure the impact of 8 abuse-deterrent opioid drug products on actual abuse in 9 the community. Your recommendations will help us to 10 develop policies that will affect all companies 11 marketing or developing abuse-deterrent opioid products. 12 We have made the warning letter available to 13 you since it is available to the public on our Web Site. 14 However, the issues described in the warning letter are 15 not directly relevant to your discussions today, and 16 this meeting is not the proper forum to debate the 17 issues underlying the issuance of the letter and any 18 corrective actions taken by King Pharmaceuticals. 19 Thank you. 20 DR. KIRSCH: 21 We’ll now start the presentations by the 22 agency. Thank you. The first presenter is Dr. Lapteva. PRECISE REPORTING, LLC jcp 24 1 Nature of the Problem of Prescription 2 Opioid Misuse and Abuse 3 Overview of the Risk of Abuse and Regulatory Discussions 4 to Date to Reduce Abuse of Opioid Analgesics 5 DR. LAPTEVA: Good morning, Mr. Chairman, 6 Advisory Committee Panel Members, all invited guests, 7 and meeting participants. 8 and I work in the Division of Anesthesia and Analgesia 9 Products in the Office of New Drugs in CDER. 10 My name is Larissa Lapteva, In my presentation today, I will give you a 11 brief overview of the risk of abuse and measures 12 employed to mitigate this risk. 13 the current recommendations for development of abuse- 14 deterrent formulations, then describe formulations with 15 abuse-deterrent properties developed to date, and then 16 discuss some labeling considerations for labeling claims 17 for abuse deterrents. 18 Then, briefly summarize Abuse of prescription drug products, 19 particularly opioid analgesics, has steeply increased 20 over the past decade, and is now recognized by many as a 21 national public health crisis. 22 study report of Treatment Episode Datasets, also known According to the recent PRECISE REPORTING, LLC jcp 25 1 as TEDS, recently released by the Substance Abuse and 2 Mental Health Services Association, the proportion of 3 all substance abuse treatment admissions related to 4 serious medical outcomes associated with abuse increased 5 more than fourfold between the years of 1998 and 2008. 6 This slide presents data from the Treatment 7 Episode Dataset on the treatment admissions involving 8 opioid analgesics between the years of 1992 and 2008. 9 As you can see, the number of admissions for opioid 10 analgesics has increased from less than 30,000 in the 11 year 1992 to more than 185,000 treatment admissions in 12 the year of 2008. 13 medical outcomes is likely multifactorial, but may in 14 part be explained by how these prescribed products could 15 be obtained by people who abuse and misuse opioid 16 analgesics. 17 Such increase in the abuse-related A large proportion of the prescription opioid 18 analgesics that are misused and abused are reportedly 19 obtained by friends and relatives from patients with 20 prescriptions. 21 representation of how respondents to the National Survey 22 on Drug Use and Health, also known as NSDUH, reported On this slide, you see a pictorial PRECISE REPORTING, LLC jcp 26 1 the source of pain-reliever that was taken non- 2 medically. 3 As you can see, about 18 percent of survey 4 respondents obtained it from one doctor, whereas about 5 70 percent obtained their pain reliever from a friend or 6 relative, either for free or because they bought it. 7 And, among the relatives, about 80 percent obtained 8 their pain relievers from one doctor. 9 the drugs from internet and a low proportion obtained Very few obtained 10 the pain reliever from more than one doctor. 11 variety of sources from which an opioid analgesic could 12 be obtained, measuring non-medical use may be 13 particularly challenging. 14 novel systems providing information on the patterns of 15 abuse, as well as development of formulations that deter 16 drug-seeking behaviors become an important part of abuse 17 prevention. 18 Given this Therefore, development of Because measurement of abuse is a developing 19 field, it is important to operate with the terms and 20 definitions that are similarly understood by all 21 parties. 22 On this slide, you see the definitions of PRECISE REPORTING, LLC jcp 27 1 abuse and misuse proposed by the FDA Opioid REMS Working 2 Group at the Opioids REMS Class Advisory Committee 3 Meeting that was held in July of 2010. 4 definitions of abuse and misuse will be employed by the 5 subsequent FDA presenters at this meeting. These 6 Abuse is the non-medical use of a drug 7 repeatedly or even sporadically for the positive 8 psychoactive effects it produces. 9 hand, is the use of a drug outside labeling directions Misuse, on the other 10 or in a way other than prescribed or directed by a 11 health care practitioner. 12 Let me now briefly talk about the scope of 13 measures that could be employed to mitigate the risk of 14 abuse. 15 When a product with abuse potential is 16 approved, the first measure to reduce the risks 17 associated with the product is its appropriate labeling. 18 Labeling can serve is a useful, educational tool for 19 both physicians and patients. 20 release, high-strength opioids contain Boxed Warnings 21 that are used to convey the serious risks of this 22 product and direct the prescribing practices. All approved controlled PRECISE REPORTING, LLC jcp 28 1 Medication guide is the currently used form of 2 patient-directed labeling to aide safe use of extended 3 and controlled release opioids for patients for whom 4 these products have been prescribed. 5 Strategies and public campaigns, in 6 collaboration with other agencies and stakeholder 7 organizations is another way to mitigate and prevent 8 abuse. 9 launched a joint prescription drug abuse prevention 10 educational effort with the goal of preventing and 11 reducing the abuse of narcotic opioid pain-relievers. 12 13 14 For example, in January of 2003, FDA and SAMHSA While abuse is multifactorial problem, it has to be dealt with on different levels. In May of 2010, the Office of National Drug 15 Control Policy released the national strategy with a 16 five-year goal to reduce non-medical use of prescribed 17 drugs and its consequences through a balanced policy of 18 abuse prevention, treatment, enforcement, and 19 international cooperation. 20 efforts to reduce drug trafficking, as well as 21 prevention and treatment of drug abuse. 22 The strategy also includes With the authorities given to the Food and PRECISE REPORTING, LLC jcp 29 1 Drug Administration by the Food and Drug 2 Administration’s Amendments Act after its passage in 3 2007, the FDA and the manufacturers of the opioid drug 4 products started developing risk evaluation and 5 mitigation strategies for classes of opioid analgesics 6 and fentanyl-containing products. 7 would put additional safeguards to the health care 8 system to aide appropriate drug prescribing, dispensing, 9 storage, and safe use. 10 These strategies While risk mitigation by information and 11 strategies, could achieve a certain degree of 12 diminishing abuse, the risk mitigation at the stage of 13 design of the pharmaceutical products appears a very 14 promising venue. Over the past decade, designing pain- 15 relief products with the new physiochemical features to 16 deter abuse has been an ongoing effort of drug 17 manufacturers and academia, highly encouraged by 18 regulatory agencies. 19 Specifically, FDA has been communicating to 20 the manufacturers about abuse deterrent formulations 21 through development of guidance documents, written 22 advice to manufacturers on individual drug development PRECISE REPORTING, LLC jcp 30 1 programs, presentations at academic settings, and 2 discussions at advisory committee meetings. 3 Now, let me give you a brief overview of the 4 current FDA recommendations for the development of abuse 5 deterrent products. 6 Before a product is tested in preclinical 7 studies, or, as we call it, at the pre-investigational 8 new drug application stage, the agency encourages 9 manufacturers to include in the design of the 10 formulation features aiming to deter abuse. 11 features may include formulations with physiochemical 12 barriers to tampering or combination products with an 13 antagonist intended to reduce euphoria when the 14 antagonist is released during in appropriate use, or 15 formulations that include non-analgesic ingredients that 16 cause unpleasant side effects when the product is used 17 inappropriately. 18 Such At the pre-marketing stage, manufacturers 19 would need to demonstrate that the new features of the 20 formulation, in fact, translate into decrease in the 21 abuse potential, observed from different kinds of data: 22 in vitro data of the product’s resistance to tampering, PRECISE REPORTING, LLC jcp 31 1 as well as pharmacokinetic and bioavailability studies, 2 and clinical studies, evaluating the likeability and 3 euphorigenic effects of both manipulated and intact 4 abuse-deterrent product. 5 And finally, at the post-marketing stage, 6 manufacturers would need to demonstrate a meaningful 7 decrease in abuse-related outcomes, including addiction, 8 overdose, and death, as observed from the post-marketing 9 epidemiological studies. 10 Moving on from theory to practice, this slide 11 summarizes the regulatory experience with abuse- 12 deterrent formulations developed to date. 13 combination products formulated with an opioid 14 antagonist naloxone, Talwin, and Suboxone do have some 15 post-marketing data, which I will describe in the next 16 slide. 17 Two approved Of the other recently discussed or approved 18 abuse-deterrent products, OxyContin, reformulated from 19 the original OxyContin with a change in physicochemical 20 properties, and Embeda, oral capsule with the pellets of 21 morphine sulfate and sequestered opioid antagonist 22 naltrexon are both approved products, and their post- PRECISE REPORTING, LLC jcp 32 1 marketing epidemiological programs are the subject of 2 this meeting’s discussion. 3 Remoxy, controlled-released oxycodone, 4 formulated with additional properties to resist 5 manipulation, and Acurox, a combination product of the 6 immediate release oxycodone and niacin tablet, were both 7 designed to be abuse-deterrent, and have not yet met the 8 regulatory criteria for approval. 9 products with abuse-deterrent features are currently in 10 11 Several other development. The story of Talwin is likely well-known to 12 this advisory committee, and probably does not need to 13 be repeated, yet, we come back to it as to an example 14 when some apparent decrease in the abuse of this product 15 was seen upon introduction of risk mitigation 16 strategies. 17 Talwin, also known pentazocine, was approved 18 in 1967 for the relief of moderate to severe pain. 19 first reports of dependence appeared in 1968, and in the 20 late 1970s, increasing frequency of cases of abuse, 21 diversion, overdose, and death were reported. 22 effort to mitigate abuse, Talwin was scheduled under the PRECISE REPORTING, LLC The In an jcp 33 1 Controlled Substance Act in 1979, reformulated with 2 Naloxone, and the original Talwin removed from the 3 market. 4 With these measures, it appeared that the 5 abuse outcomes reported with Talwin declined during the 6 two years after withdrawal of the original formulation 7 from the market, as you can see from the graph on the 8 right side of the slide. 9 factors likely contributed to decrease in abuse in 10 Talwin, it was also possible that the change in the 11 availability of heroin, which occurred at about the same 12 time, played a role in decrease of abuse with Talwin. 13 However, while all of these This exemplifies a setting when the change in 14 abuse trends could have been influenced by multiple 15 factors and not necessarily only by the measures 16 introduced to mitigate abuse. 17 Suboxone is another formulation product 18 formulated with buprenorphrine and naloxone, which was 19 approved in October of 2002 for the treatment of opioid 20 dependence. 21 done to access whether addition of naloxone decreased 22 abuse, the post-marketing reports of intravenous and Although there have been no formal studies PRECISE REPORTING, LLC jcp 34 1 intranasal abuse continued to support existence of abuse 2 with Suboxone, despite the inclusion of the opioid 3 antagonist in the formulation. 4 Again, on this slide and on the following 5 slide, I will briefly discuss the regulatory history of 6 OxyContin and Embeda, and then we’ll move on to the 7 topic of labeling claims. 8 OxyContin is well-known and does not need to be repeated 9 to this committee. But, again, the story of However, several aspects of it 10 pertaining to the development of the reformulated 11 OxyContin are worth to mention. 12 When the growing problem with abuse and misuse 13 of OxyContin was recognized around 2001, it was as early 14 as April of 2001, when the FDA and Purdue Pharma started 15 discussing development of a reformulated product with 16 properties that would improve resistance to product’s 17 manipulation. 18 put the reformulated version through the development 19 program, and in November of 2007, the new drug 20 application for the reformulated OxyContin was 21 submitted. 22 It took the company almost six years to Following the agency’s review and the two PRECISE REPORTING, LLC jcp 35 1 advisory committee meetings, the reformulated OxyContin 2 was approved in April of 2010 with two post-marketing 3 requirements: 4 the post-marketing epidemiological studies. 5 REMS is not the point of discussion of this advisory 6 committee, the epidemiologic studies will be presented 7 to the panel and discussed at this meeting. 8 9 Risk Evaluation Mitigation Strategy and While the Embeda is a formulation similar to another extended-release morphine sulfate product named Kadian, 10 which was approved in 1996. 11 includes the opioid antagonist naltrexone to decrease 12 the euphorigenic effects with inappropriate use. 13 sponsor of Embeda approached the agency in 14 March 2005, before they submitted their investigational 15 new drug application, and, at the time, already planned 16 their post-marketing epidemiological program. 17 Unlike Kadian, Embeda The The original New Drug Application for Embeda 18 was submitted in February of 2008, and then the product 19 was approved in August of 2009 with the post-marketing 20 requirement of REMS. 21 epidemiological program continued in post-marketing, and 22 are the subject of this advisory committee meeting. Discussions about the PRECISE REPORTING, LLC jcp 36 1 Switching gears now to the important 2 regulatory aspect of labeling claims. Before discussing 3 the labeling claims for abuse deterrence, let me explain 4 the difference between the indications and the claims. 5 Indications are the approved uses and populations for a 6 drug or biological product, and they are described in 7 the indication part of the labeling. 8 product could be indicated for treatment of moderate to 9 severe pain in opioid-tolerant patients. For example, a Claims, on the 10 other hand, may be based on any labeled information, not 11 just an indication. 12 They may be explicit or implicit. For example, if one takes a possible claim of 13 abused deterrence, then a statement in the label that 14 the product is abuse-deterrent will be an explicit 15 claim, whereas showing a table, demonstrating more 16 qualities to resist tampering with the product, would be 17 considered an implicit claim for abuse deterrence. 18 Nevertheless, implicit or explicit, claims could be 19 included in the labels when they're accurate and 20 complete reflections of the product’s properties. 21 So, labeling claims for an abuse-deterrent 22 product would require demonstration that a product’s PRECISE REPORTING, LLC jcp 37 1 abuse-deterrent properties studied in the pre-marketing 2 program actually resulted in a reduction in abuse and 3 its outcomes: 4 confirmed in post-marketing epidemiological studies. 5 They would be dependent, as any claims would be 6 dependent, on the adequacy of the data. 7 promotion based on such claims would be limited to 8 presentations of the pertinent data. 9 death, overdose, and addiction, as Any possible And, in conclusion, let me highlight some of 10 the challenges with evaluating the impact of abuse- 11 deterrent formulations. 12 First of all, pre-marketing studies for abuse 13 liability have their limitations, and you will hear a 14 more extensive presentation on this topic from the 15 Controlled Substance Staff later on this morning. 16 difficult to measure abuse since abuse is not a clinical 17 phenomenon or a drug-related adverse reaction, but 18 rather a consequence of non-medical use. 19 collection or measures used in population-based 20 epidemiological studies may not apply to measuring 21 abuse. 22 limitations, of which you will hear in the subsequent It is Standard data Current surveillance systems have their PRECISE REPORTING, LLC jcp 38 1 presentations today, and we are in need of novel 2 surveillance systems. 3 Defining the population of abusers can be 4 difficult, because usually, abusers cannot be adequately 5 identified until a serious outcome occurs or a person 6 self-identifies as a survey responder. 7 And finally, even when decrease in abuse to 8 one product is demonstrated, the overall impact on the 9 abuse problem may not be observed until more abuse- 10 11 deterrent formulations are on the market. It is not easy to measure abuse, and it is not 12 easy to develop abuse-deterrent formulations. 13 government agency, individual drug manufacturer, 14 isolated non-profit or professional organization could 15 defeat this big societal problem. 16 by-step approach by multiple stakeholders will be needed 17 to achieve the desired results. 18 in modern technology, designing and development of 19 abuse-deterrent formulations became possible. 20 No single Collaborative, step- Owing to the advances It is now the turn of clinical and 21 biostatistical sciences to assess whether bringing these 22 formulations to the market will actually result in the PRECISE REPORTING, LLC jcp 39 1 decrease in abuse in the nation. Through conducting 2 this advisory committee meeting, FDA is looking to an 3 open and engaging discussion to help us find the path 4 forward. 5 Thank you for your attention. 6 DR. KIRSCH: 7 Dr. Tolliver? 8 9 Thank you. PreMarketing Assessment of Abuse-deterrent Formulations DR. TOLLIVER: Good morning. My name is 10 James Tolliver. 11 Controlled Substance Staff within the Center of the Drug 12 Evaluation and Research at the Food and Drug 13 Administration. 14 on the pre-marketing assessment of abuse-deterrent 15 formulations from an FDA perspective. 16 I am a pharmacologist for the My presentation this morning will focus By way of introduction, I’d like to provide 17 two definitions relevant to this presentation. 18 first definition is that for abuse-deterrents. 19 According to the FDA-CDER draft document entitled 20 “Guidance for Industry Assessment of Abuse Potential 21 Drugs,” abuse deterrence is defined as the introduction 22 of some limits or impediments to abuse in a drug PRECISE REPORTING, LLC The jcp 40 1 formulation as opposed to the outright elimination of 2 abuse. 3 The second definition is that of abuse, which 4 is defined as the non-medical use of a drug repeatedly 5 or sporadically for the positive psychoactive effects it 6 produces. 7 The pre-market assessment of formulations 8 purported to be abuse-deterrent involves a three tier 9 approach. The first tier is that of in vitro 10 manipulation and extraction studies. 11 by clinical pharmacokinetic studies on the intact and 12 manipulated formulation. 13 This is followed The last tier involves human abuse liability 14 studies. It must be stressed that the three-tier, pre- 15 market assessment is to be conducted on the to-be- 16 marketed product formulation. 17 usually have controlled release mechanism that allow for 18 the release of an opioid over an extended period of 19 time. 20 amounts of opioid that exceed that of immediate release 21 product formulations, making them potentially attractive 22 targets for manipulation with the intent to abuse. Such product formulations These product formulations also generally contain PRECISE REPORTING, LLC jcp 41 1 There are a number of different types of 2 purported abuse-deterrent formulations for opioid 3 analgesics. 4 Three types are presented in this slide. There are formulations that are intended to 5 resist physical and chemical manipulation. 6 example of that is reformulated OxyContin. 7 A good A second type of formulation is that of an 8 opioid agonist in combination with an opioid antagonist. 9 A good example of this formulation is Embeda. With this 10 formulation, the intent is that the opioid antagonist 11 will mitigate positive, subjective effects, and possibly 12 cause adverse effects such as that of opioid withdrawal 13 if the formulated product is used in a manner other than 14 that intended in the label. 15 A third type of formulation is the combination 16 of an opioid agonist with a second component that will 17 produce an aversive effect if the product is not taken 18 as indicated. 19 Here, an example would be Acurox. The purpose of in vitro manipulation and 20 extraction studies is to evaluate the ease with which 21 the abuse-deterrent mechanism of a formulation can be 22 defeated. Physical and chemical manipulation of a PRECISE REPORTING, LLC jcp 42 1 formulation is intended to obtain the opioid in a form 2 more amiable for abuse by desired routes of 3 administration. 4 In the case of opioid agonist, antagonist 5 combination formulations, separation and isolation of 6 the opioid from the opioid antagonist is a 7 consideration. 8 with an aversive agent, a goal would be to neutralize 9 the effects of the aversive agent by separation or other 10 For formulations of an opioid agonist means, while maintaining opioid agonist effects. 11 These studies are designed while keeping in 12 mind the knowledge of the physical and chemical 13 properties of the formulation, including the opioid 14 agonist, opioid antagonist, and other components, such 15 as the aversive reagent, if present. 16 Another consideration in designing these 17 studies is the knowledge of methods available to abusers 18 with different levels of chemical expertise, 19 constituting such groups as recreational abusers, more 20 experienced abusers, and, finally, so-called kitchen 21 chemists. 22 In vitro manipulation and extraction studies PRECISE REPORTING, LLC jcp 43 1 are of three types. 2 the mechanical manipulation of a formulation. 3 there are chemical extraction studies. 4 of study relates to modifying the formulation in 5 whatever way for purposes of abuse by snorting, 6 inhalation, or intravenous use. 7 The first are studies looking at Secondly, The third type In vitro mechanical manipulation studies are 8 intended to evaluate a variety of common household tools 9 for crushing, cutting, grading, and grinding product 10 formulations with comparisons to appropriate extended- 11 release reference products. 12 the manipulation, as well the ease of the manipulation 13 are noted. 14 the formulation, thereby possibly changing the 15 formulation’s controlled release properties. 16 Both the time required for The intent is to reduce the particle size of Dissolution studies of intact versus 17 manipulated product formulations will determine whether 18 the controlled release mechanism is compromised. 19 studies may also examine the impact of first freezing or 20 heating the formulation on the ability to mechanically 21 manipulate that formulation. 22 artificial saliva, can be used to predict the effects of Other Chewing simulators, using PRECISE REPORTING, LLC jcp 44 1 chewing the product formulation on the controlled 2 release of the opioid agonist and other components of 3 the formulation. 4 In vitro studies also evaluate the ease of 5 chemical extraction of opioids from intact and 6 manipulated product formulations. 7 to be made to intact and mechanically-manipulated 8 reference products. 9 Comparisons are also In the case of opioid agonist/antagonist 10 combination products, consideration is given to the 11 chemical separation of the opioid agonist from the 12 opioid antagonist. 13 In chemical extraction studies, a variety of 14 chemicals are tested as solvents. These include water, 15 beverages or simulated beverages, household chemicals, 16 buffers of different ph, and other chemicals 17 constituting different molecular polarities. 18 Extractions are conducted under continuous agitation and 19 at room temperature and elevated temperature. 20 of opioid extracted is determined at selected time 21 points out to 12 or 24 hours or until the opioid is 22 mostly extracted. PRECISE REPORTING, LLC Percent jcp 45 1 Finally, in vitro studies are conducted with 2 the intent of determining the ease with which a product 3 formulation may be modified mechanically and chemically 4 to prepare for abuse by intranasal inhalation and/or 5 intravenous injection. 6 administration, both particle size and the behavior of 7 the manipulated formulation at the lining of the nasal 8 cavity are important considerations. 9 For the purpose of intranasal With respect to possible inhalation, it is 10 important to consider the vaporization and degradation 11 temperatures for the opioid agonist of interest. 12 is abuse by inhalation of an opioid is not feasible when 13 the temperature at which the opioid agonist chemically 14 decomposes is less than the temperature required to 15 vaporize the opioid agonist. 16 may be done to evaluate the possible conversion of the 17 opioid agonist to a form more amiable to inhalation. 18 That In such a case, studies In the case of preparing for intravenous 19 injection, the intent is to obtain a small volume of 20 solution with sufficient opioid agonist concentration 21 such that upon intravenous injection subjective 22 reinforcing effects, such as euphoria, may be achieved. PRECISE REPORTING, LLC jcp 46 1 The injectable solution must be of a sufficiently low 2 viscosity to allow the solution to be taken up into a 3 syringe, in other words, injectability, and subsequently 4 to be injected via needle, that is the injectability of 5 the solution. 6 Clinical pharmacokinetic studies constitute 7 the second tier of pre-market assessment, purported 8 abuse deterrent formulations. 9 purported abuse deterrent formulation both intact and In these studies, the 10 manipulated is compared to one or more reference 11 extended-release products and to a reference 12 immediately-release product. 13 Oral ingestion, in chewing or, most common 14 modes, administration with these types of studies, but 15 other routes are also found from time to time. 16 Additional studies looked at the effects of 17 Concomitant food and ethanol ingestion on the control- 18 release mechanism or purported abuse-deterrent 19 formulations. 20 These various studies tend to be open-labeled, 21 randomized, single-dosed, and crossover in design, using 22 healthy adult volunteers under opioid agonist blockade. PRECISE REPORTING, LLC jcp 47 1 Plasma concentrations of opioid agonists and possibly 2 other metabolites of the agonist are followed as a 3 function of time following dose administration. 4 of an agonist-antagonist product, formulations, plasma 5 levels of the antagonist or opioid agonist are also 6 determined over time. 7 In case A variety of pharmacokinetic parameters are 8 determined. The most important of these includes the 9 peak plasma concentration, designated Cmax. The time to 10 peak plasma concentration, designated Tmax, and the area 11 under the concentration versus time curve for some time 12 from zero to some time point usually just a few hours. 13 This last parameter reflects the amount of 14 drug exposure over a designated time period. 15 compromise of the extended-release mechanism of a 16 purported abuse-deterrent formulation is indicated when 17 the peak plasma concentration of the manipulated 18 formulation is greater than that of the peak plasma 19 concentration achieved with the intact formulation and 20 when the time to peak plasma concentration of the 21 manipulated formulation is less than that of the time to 22 peak plasma concentration of the intact reparation. PRECISE REPORTING, LLC A jcp 48 1 And finally, also, when the area over the 2 concentration curve in a short period of time for the 3 manipulation product formulation is greater than the 4 area under the concentration curve for that same period 5 of time for the intact product. 6 In that event that the results of in vitro 7 manipulation and extraction studies and clinical 8 pharmacokinetic studies indicate that the controlled 9 release mechanism of a purported abuse deterrent 10 formulation can be compromised. 11 move to the third and final tier of the pre-market 12 assessment, namely human abuse liability studies. 13 purpose of these studies is to compare the subjective 14 effects produced between intact and manipulated 15 formulation of the purported abuse deterrent product. 16 Additional comparisons are with subjective effects 17 produced by intact and manipulated extended release 18 reference products, as well as with an immediate release 19 reference product and possibly placebo. 20 It is appropriate to The In addition to the subject effects, another 21 pharmacokinetic dynamic effects measured, 22 pharmacokinetic parameters, as previously described, are PRECISE REPORTING, LLC jcp 49 1 2 also generally determined. Studies are randomized. Placebo-controlled, 3 single-dose, double-blind, crossover in design, and are 4 conducted in a controlled setting. 5 completed using approximately 30 subjects, consisting of 6 opioid-experienced, non-dependent volunteers who can 7 discriminate the subjective reinforcing effects of the 8 opioid in question from that of placebo. 9 Studies are Subjective endpoints measures are obtained 10 using a variety of standardized questionnaires used to 11 assess the reinforcing effects, such as the Visual 12 Analog Scale for drug liking, and also for dysphoric 13 effects, using, for example, the Visual Analog Scale for 14 bad drug effects. 15 measurements, are recorded just before, and as a 16 function of time following treatment. 17 Subjective effects, as well as other Maximum effect is designated as Emax. Time to 18 maximum effect is designated as Tmax, and the area under 19 the time effect curve from zero to some time, t are 20 determined along with other parameters, including 21 pharmacokinetic parameters. 22 individual response data are analyzed. Both mean, as well as PRECISE REPORTING, LLC With respect to jcp 50 1 subjective reinforcing effects, increases in the maximum 2 effect, and area under the effect curve, and a decrease 3 in the time to maximum effect of a manipulated 4 formulation compared to the intact formulation suggests 5 compromise of the controlled-release properties of the 6 formulation. 7 Human abuse liability studies do have 8 limitations. Considering that these studies involve 9 subjective measures, it is not surprising that 10 substantial variability may exist in the subjective 11 endpoints. 12 effects may be difficult to collectively interpret. 13 Results are specific for the route of administration and 14 doses used in the study. 15 differences in subjective effects may not necessarily 16 represent a meaningful difference in a potential for a 17 drug product to be abused. 18 In addition, multiple scales of subjective Statistically, significant With respect to summary and conclusion, 19 currently, pre-market marketing assessment of abuse- 20 deterrent formulations using in vitro manipulation and 21 extraction studies, chemical pharmacokinetic studies and 22 human abuse liability studies provide information that PRECISE REPORTING, LLC jcp 51 1 suggests how and to what extent a product purported to 2 be abuse-deterrent may be manipulated and abused once 3 the product is on the market. 4 importantly, it should be noted that only post-marketing 5 epidemiological studies will reveal the extent to which 6 a product purported to be abuse-deterrent will actually 7 be manipulated and abused after the product has been on 8 the market. 9 However, very In addition, post-marketing epidemiological 10 studies may also establish that appropriateness of the 11 pre-market assessment studies in predicting patterns and 12 extent of abuse of products purported to be abuse- 13 deterrent once the products are placed on the market. 14 Thank you. 15 DR. KIRSCH: 16 While Dr. Paulozzi comes up to the podium, I’d 17 like to take a minute to have Dr. Morris-Kukoski and Dr. 18 Denisco introduce themselves. 19 MS. MORRIS-KUKOSKI: Thank you. Hi, Dr. Cynthia Morris- 20 Kukoski, FBI forensic examiner and toxicology at the FBI 21 Laboratory, clinical pharmacist, toxicologist, United 22 States Navy Reserve. PRECISE REPORTING, LLC jcp 52 1 MR. DENISCO: Richard Denisco, medical officer 2 at the National Institute of Drug Abuse, specialized in 3 pain medicine and addiction medicine and public health 4 and statistics in epidemiology. 5 DR. KIRSCH: 6 Dr. Paulozzi? Thank you. 7 Abuse of Marketed Opioid Analgesics and Their 8 Contribution to the National Problem of Drug Abuse 9 MR. PAULOZZI: Good morning, everyone. My 10 name is Len Paulozzi. 11 the National Center for Injury Prevention and Control at 12 the Centers for Disease Control and Prevention. 13 I’m a medical epidemiologist in I’m going to be talking about the abuse of 14 marketed analgesics and its contribution to the national 15 problem of drug abuse. 16 When FDA asked me to cover this topic, they 17 asked me how much time I wanted. 18 days do you have? 19 I leave anything out, forgive me. 20 I said, well, how many And we settled on 20 minutes. I begin many talks with this slide. So, if It is the 21 rater of unintentional drug overdose death from all 22 drugs in the United States from 1970 through 2007, which PRECISE REPORTING, LLC jcp 53 1 is the latest year of national mortality data being 2 available. 3 overdose death rate through the 1990s and through 2007. 4 The rate of 2007 actually represents about 27,700 5 deaths. 6 are beginning to approach the numbers of death related 7 to motor vehicle crashes in the United States, and that 8 is an unprecedented occurrence. 9 You can see a dramatic increase in the drug The numbers of deaths related to drug overdose Why is the rate going up? We have some data 10 that breaks down the drug types from 1999 through 2007 11 shown here on this slide. 12 dependent on the coroner or medical examiner putting 13 down the name of the drug on the death certificate. 14 Some death certificates still come in as recorded as 15 drug overdose and nothing else or a narcotic overdose or 16 even opioid overdose, and you can’t tell what kind of 17 drug it is exactly. 18 certificates where they do specify the type of drug. 19 And, of course, many or most deaths involve more than 20 one type of drug. 21 22 The occurrence of drugs are This data represents death So, some deaths are represented twice on this slide. But the point is that the opioid analgesic- PRECISE REPORTING, LLC jcp 54 1 related deaths have increased more rapidly than any 2 other of the major types of drugs shown here on this 3 slide. 4 Heroin deaths are basically flat from 1999 5 through 2007. 6 but the biggest increase is in the opioid analgesic 7 category. 8 total number of deaths involving opioid 9 analgesics exceed the total number involving either 10 Cocaine deaths have gone up appreciably, And it was actually a few years ago that the heroin or cocaine in the United States. 11 The number of deaths shown here for opioid 12 analgesics and unintentional overdoses shown in yellow 13 is the same as on the previous slide. 14 to it is the opioid sales in the United States as 15 tracked by the Drug Enforcement Administration ARCOS 16 Program, and computed milligrams in morphine milligram 17 equivalents per person over time. What I’ve added 18 The opioid sales are shown on the right axis, 19 and they have increased dramatically up to the point in 20 2007. 21 person being distributed of opioid analgesics in the 22 United States. Preliminary figures show about 700 milligrams per And, clearly, the increases have PRECISE REPORTING, LLC jcp 55 1 2 occurred in parallel. Turning from mortality data for a moment, this 3 is data from emergency department visits as recorded by 4 the Drug Abuse Warning Network. 5 emergency departments across the United States and the 6 numbers are projected upward to national estimates. 7 in 2008, the numbers of ED visits involving legal drugs, 8 the first bar on the left in yellow, from misuse or 9 abuse of those drugs surpassed 1 million emergency It’s a sample of And 10 department visits. 11 involving legal drugs exceeded the number involving 12 illicit drugs shown in the first bar in green. 13 analgesics and benzodiazepines were the major 14 contributors to the legal drug category, whereas cocaine 15 and heroin were the major contributors to the illicit 16 drugs in emergency department visits. 17 Therefore, the number of visits Opioid Although there are thousands of deaths 18 associated with opioid analgesics in the United States 19 today, they really are the tip of the iceberg or the top 20 of the pyramid, if you prefer. 21 overdose deaths related to opioid analgesics at the top 22 here, and in 2007, there were at least 11,499 documented I put the unintentional PRECISE REPORTING, LLC jcp 56 1 involvement of opioid analgesics on death certificates 2 in the United States. 3 By comparison, there are about 105,000 opioid 4 treatment admissions, where opioid analgesics were the 5 primary drug. 6 the misuse or abuse of opioid analgesics. 7 27 ED visits for every one death. 8 National Survey of Drug Use and Health shows that in 9 2009, there were almost 2 million people in the United 10 States who self-reported abuse or dependence on opioid 11 analgesics in the past year. 12 all is the 5.3 million people who reported non-medical 13 use of opioid analgesics in the past month on the 2009 14 national survey of Drug Use and Health. 15 There were about 306,000 ED visits for That's about And data from the And the largest figure of So, as the deaths have gone up, they really 16 are just representing a small part of the morbidity and 17 mortality associated with this problem, which affects 18 millions of people now in the United States. 19 So, I’m now going to turn to some different 20 sources of data to try to quantify the prevalence of 21 abuse of opioid analgesics using a variety of different 22 data sources. We’re going to look at some circumstances PRECISE REPORTING, LLC jcp 57 1 of pharmaceutical overdose deaths in medical examiner 2 studies. 3 results of urine drug testing among pain patients in 4 brief, data on patients receiving opioid analgesics 5 tracked in insurance claims data or Prescription Drug 6 Monitoring Program information, and data on the route of 7 administration or exposure of people entering substance 8 abuse treatment because that's the topic of this 9 meeting. 10 These are state-specific studies, look at some First, the overdose deaths and medical 11 examiner data. 12 we conducted on this topic in the State of West 13 Virginia, which, at the time, had the highest drug 14 overdose rate in the United States. 15 West Virginia was affected primarily by legal drugs 16 rather than illicit drugs. 17 pharmaceutical overdose deaths occurred in West 18 Virginia. 19 This was one of the earlier studies that The years 2006, In that year, 295 Within that group, 231 decedents, or 78 20 percent, had a history of substance abuse, whether 21 alcohol or drugs. 22 substance abuse was observed or noted in the medical Other mental illness other than PRECISE REPORTING, LLC jcp 58 1 examiner’s records for 43 percent of the decedents. And 2 63 percent of the decedents had one or more of the 3 prescription drugs involved in their death for which 4 they did not have any prescription recorded in the State 5 Prescription Drug Monitoring Program. 6 percent of the deaths showed some evidence of non- 7 medical route of administration, such as injection or 8 snorting the drugs. 9 five or more prescribers of controlled substances in the Twenty-two Twenty-one percent had a history of 10 past year in the State Prescription Drug Monitoring 11 Program, and about seventeen percent had a history of a 12 previous overdose. 13 All told, a population that had a lot of 14 indicators of history of substance abuse and their past 15 history and then the circumstances surrounding their 16 death. 17 Another study done more recently in Utah, 2008 18 and 2009, involving a 155 deaths. Similar results, 19 history of substance abuse in 60 percent, signs of non- 20 medical use, in this case a broad definition, were found 21 in 51 percent, but that category includes any opioid 22 involved without a prescription, which was involved in PRECISE REPORTING, LLC jcp 59 1 37 percent of the decedents. Again, linking decedents 2 to the State Prescription Drug Monitoring Program. 3 Eighty-two percent had a history of chronic pain. 4 Typically, this was headaches or back pain or other 5 muscular skeletal problems, and 57 percent had mental 6 illness and that had been diagnosed by a provider and 7 was available to the state medical examiner. 8 This is the data from a Web report put out by 9 the Ohio Department of Death recently for data for 2006 10 through 2008. 11 of substance abuse by linking their deaths, again, with 12 their State Prescription Drug Monitoring Program. 13 they looked at unintentional drug overdose deaths in 14 total, over 1,000 deaths in Ohio during these three 15 years. 16 prescriptions from an average of 5 prescribers per year 17 over the three years of data that they looked at. 18 I’m going to show you a lot of information about numbers 19 of prescribers that's oftentimes used as a surrogate for 20 a label of doctor shopping, a use of multiple providers 21 to obtain similar types of drugs. 22 They were able to look at some indicators And And 16 percent of those individuals had filled And As in previous studies, a lot of the people PRECISE REPORTING, LLC jcp 60 1 had no prescriptions in the Prescription Drug Monitoring 2 Program for the drugs in their death; in this case for 3 the opioids. 4 so, they obtained the drug by some route other than 5 through prescription, through drug diversion, 6 presumably. 7 because it was the leading drug among the deaths 8 involved, and they saw that 71 percent, most of the 9 people, had no prescription in the Prescription Drug Twenty-five percent had no prescription, And they looked in particular at methadone 10 Monitoring Program for methadone among the methadone- 11 related deaths. 12 substance abuse treatment programs and rather than just 13 for treatment of pain. 14 Methadone, of course, is also used in And this is sort of a compilation of a number 15 of studies focused on methadone. It was the leading 16 drug among opioid analgesic deaths in the United States 17 for most of this decade. 18 years that state studies have shown that oxycodone has 19 come back up to the top. 20 the other drugs, a small proportion of the decedents had 21 a prescription. 22 prescription. It’s only in the last few But, in general, as seen for That's the last column, percent with Going back to early studies through the PRECISE REPORTING, LLC jcp 61 1 1990s, it’s basically a third or so of the people had a 2 prescription for methadone when examined by the medical 3 examiner. 4 So, you can see a variety of different 5 problems around the country in different states, taken a 6 look at by medical examiners, sometimes using different 7 definitions. 8 information, but there’s a large prevalence of abuse 9 associated with opioid analgesic deaths. 10 It’s hard to standardize some of the Turning to another topic, patients with 11 chronic nonmalignant pain, a lot of literature, again, 12 on this topic. 13 to you this one systematic review of the literature. 14 It’s prevalence of abuse-related behaviors in patients 15 with chronic non-cancer pain and chronic opioid 16 analgesic treatment. 17 reviewed a number of these studies and pulled out a 18 number of different prevalances from them. 19 clinician-determined development of addiction was 20 recorded in a number of different ways and different 21 studies, among 24 studies that looked at this prevalence 22 of addiction among these patients under treatment. Many studies. I chose just to present And Fishbain in this 2008 paper PRECISE REPORTING, LLC The jcp 62 1 Three percent were recognized as having or recorded as 2 having addiction. 3 percent of patients in 17 studies with aberrant drug- 4 related behaviors, things such as reporting a loss of 5 your prescription, early refills, calls to the office, 6 antagonistic behavior, a variety of different measures 7 in different studies, 11 percent was the prevalence. 8 9 However, when they looked at the And finally, in 5 studies, aberrant behaviors determined by urine drug testing showed the highest 10 prevalence of all, about 20 percent. 11 in this case meant that the person did not have the 12 prescribed opioid in their urine when tested or they had 13 opioids that were not prescribed to them in their urine. 14 Either one qualified as aberrant behaviors. 15 generally recognized that observation and even 16 questionnaires administered to patients are not very 17 sensitive to the overall prevalence of this problem. 18 Aberrant behaviors And this And, more recently, people are looking at 19 large datasets, such as insurance claims and 20 Prescription Drug Monitoring Programs to try to get a 21 handle on this using surrogate markers for a misuse and 22 abuse of drugs. PRECISE REPORTING, LLC jcp 63 1 This a study in Maine based on insurance 2 claims data published by White last year. 3 at behaviors during just a three-month period of time, 4 and these are all privately-insured patients who were 5 all opioid users. 6 row, opioid abuse, diagnosis, and claims data, 3.5 7 percent. 8 clinicians and recorded as a diagnosis. 9 the 3.3 percent I showed on a previous slide for 10 11 They looked If you look at the bottom, the lowest That's actually, therefore, observed by It’s similar to recognized addictions by clinicians. They also looked at combinations of 12 prescription claims and identified people who had used 13 two or more pharmacies for opioids; about 20 percent 14 during 3 months. 15 prescribers. 16 opioid prescription. 17 indicative of misuse or abuse necessarily, but they did 18 find the significant associations between these 19 behaviors, these uses of multiple pharmacies, 20 physicians, and early refills with the presence of an 21 opioid abuse diagnosis in claims data. 22 marker, although a non-specific one, for opioid abuse. Twenty-six percent used two or more Sixteen had one plus early refill or an Certainly not pathognomonic or all PRECISE REPORTING, LLC So, it is a jcp 64 1 The recent study from California Prescription 2 Drug Monitoring Program using 2007 data, they looked at 3 the prevalence of the same drug obtained from two or 4 more prescribers and dispensed at two or more pharmacies 5 within 30 days, a fairly tightly-circumscribed 6 definition in 2007, they looked at different classes of 7 drugs, and they found that for opioid analgesics 8 prescriptions, 12.8 percent of all the prescriptions 9 were involved in this type of a situation, use of two or 10 11 more prescribers, two or more dispensers, and so on. Benzodiazepine is 4.2 percent, smaller 12 percentages for stimulants and so on. Eight point four 13 percent overall for any Schedule II through IV 14 controlled prescription drug. 15 and two pharmacies is not necessarily indicative of 16 abuse, but that may happen through people legitimately 17 losing their prescriptions or choosing to go to 18 different pharmacies. 19 happen by such innocent occurrences might be represented 20 by the 1 percent stimulant or an anorectic, and, 21 therefore, the difference between that and the 12.8 22 percent of all opioid prescriptions meeting this Again, use of two doctors But the percentages that might PRECISE REPORTING, LLC jcp 65 1 definition, I think, is remarkable. 2 Another study from Massachusetts in 2006 used 3 slightly different definitions. 4 consistent definitions of doctor shopping in studies to 5 date. 6 of Schedule II drugs, which is primarily opioid 7 analgesics in the State Prescription Drug Monitoring 8 Program. 9 had used 3 or more prescribers during one year, and 10 11 There’s really no They looked at use of three or more prescribers Found that it was about 8 percent of patients about 2.5 percent had used 3 or more pharmacies. When they combined prescribers and pharmacies, 12 you can see the numbers in the last column. 13 the data only in terms of the percent of prescriptions. 14 So, 7.7 percent of Schedule II prescriptions were 15 included in this definition, 3 or more prescribers and 16 pharmacies. 17 larger numbers of prescribers and pharmacies. 18 They gave Correspondingly, smaller percentages with Finally, some data collected by surveying 19 State Prescription Drug Monitoring Programs. 20 comes from the PMP Center of Excellence at Brandeis 21 University, and it is a survey of Bureau of Justice 22 Assistance Funded Harold Rogers Grantees among the State PRECISE REPORTING, LLC This data jcp 66 1 Prescription Drug Monitoring Programs. 2 the numbers of either individuals or doses that met 3 their definition of doctor shopping, which was five plus 4 prescribers and five plus pharmacies in six months. 5 prevalence among individuals was .4 percent, using data 6 from 7 PDMPs, and for doses, it was 1 percent of all 7 these Schedule II through IV prescriptions. 8 9 10 They looked at The So, as you can see, the as definitions get tighter, as the numbers of prescribers and pharmacies goes up, the prevalence goes down. 11 It makes sense. Lastly, route of exposure, I included just two 12 slides here to give what there is available about route 13 of exposure, given the topic of this meeting. 14 information from Butler from a surveillance system known 15 as NAVIPPRO, data from 2007 and 2008. 16 adult drug-users entering substance abuse treatment 17 using opioid analgesics, and it asked them about their 18 routes of administration, and they could record more 19 than one route of administration per type of opioid 20 analgesic. 21 22 This is This is from So, for oxycodone, 76 percent of the users reported that they used it orally some of the time, and PRECISE REPORTING, LLC jcp 67 1 that would include chewing and sublingual exposures. 2 Forty-five percent inhaled, thirteen percent injected, 3 smaller percentages smoked and used other routes of 4 exposure. 5 inhaled, 56 percent injected. 6 route of exposure is reported here, obviously, because 7 these numbers add up to more than 100 percent. 8 9 For morphine, 40 percent oral, 29 percent Again, more than one There is comparable data available from the treatment data exposure dataset of the Substance Abuse 10 and Mental Health Services Administration, also known as 11 TEDS. 12 The difference here is that TEDS looks the at most 13 commonly used or records the most commonly used route of 14 exposure, and this is just oxycodone because that's all 15 that TEDS had information on. 16 are basically consistent with the NAVIPPRO statistics, 17 which I’ve repeated in the first column here. 18 Basically, the most common route is oral, including 19 chewing, 30 percent inhaled, 13 percent injected, and 20 smaller percentages for smoking and other routes of 21 exposure. 22 generating this information for me. This is data from 2008 shown in the last column. But I think the numbers And my thanks to Deborah Trunzo for Deborah’s with PRECISE REPORTING, LLC jcp 68 1 SAMHSA Group. 2 And that's all I have. 3 DR. KIRSCH: Thank you. Thank you. We now have a few 4 minutes to ask the speakers clarifying questions. 5 way that I like to run this part of it is if you raise 6 your hand, we’ll mark you down and call on you by 7 individual. 8 Dr. Bickler? 9 10 11 The I’m sorry, Dr. Bickel? Clarifying Questions DR. BICKEL: though overall. I do like Dr. Bickler's name It’s very nice. 12 (Laughter.) 13 I have sort of a comment and an inquiry for 14 Dr. Tolliver, and I agree with your assessment, 15 generally speaking, right, that abuse liability 16 approaches are using methodological procedures that are 17 in excess of 30-years-old and have certain limitations 18 about them. 19 in the study of behavioral economics of the consumption 20 of additive commodities that show increased sensitivity, 21 a greater selectivity, and have been demonstrated to 22 being increasingly predictive of subsequent behavior. however, there have been dramatic advances PRECISE REPORTING, LLC I jcp 69 1 was wondering if you have considered or the FDA has 2 considered updating their methods to the more novel 3 approaches that have been demonstrating those effects. 4 DR. TOLLIVER: I think for the most part, the 5 abuse, the human abuse liability studies that I 6 mentioned are the ones that we’ve been using. 7 know if that answers the question or not. 8 9 DR. BICKEL: I don't So, I guess I would just suggest that you contemplate or look into the options of 10 updating some of these methodological procedures using 11 more recent events as an understanding how reinforcing 12 substances are influenced choice across a broad set of 13 conditions as indicated by behavioral economics. 14 DR. TOLLIVER: All right. 15 DR. KIRSCH: Dr. Morrato? 16 DR. MORRATO: Thank you. I had a clarifying 17 question with regard to definition of claims, and since 18 we’ll be talking about what’s the evidence needed to put 19 something into label or to make a claim. 20 I was wondering if there’s any precedent or 21 examples that we could work from that talks about claims 22 that might be time-dependent. So, an easy one would be PRECISE REPORTING, LLC jcp 70 1 this is the drug that's number one prescribed, and that 2 changes over time. 3 FDA in which a claim gets into the label, it needs to be 4 monitored over time because it’s a time-dependent kind 5 of claim, and, therefore, what's the process of taking a 6 claim out once it’s in and how long that takes, et 7 cetera? 8 9 So, are there examples within the DR. RAPPAPORT: I can't think of an example, but, certainly, we modify the label all the time as new 10 information becomes available. 11 difficult once something gets in there to take it out. 12 But not terribly difficult. 13 I will say it’s more We still can do it. If the information in there is found to be 14 incorrect and raises safety concerns, we can get it out 15 of there. 16 do that when it’s essential to do so. 17 changes to every label, to every sentence in every label 18 because of minor changes, but if there are significant 19 changes, we do work with the companies to make those. 20 Does that address your question? 21 DR. MORRATO: 22 But to modify the label as time goes on, we I think so. We can’t make So, the norm is you get data and you put a statement into the label as PRECISE REPORTING, LLC jcp 71 1 opposed to a type of claim that you know could be 2 changing and evolving over time, such as is it abuse- 3 deterrent? 4 ends up in a label. 5 It has this benefit, it has this side effect, but still 6 evolving -- 7 That's not the normal type of claim that It’s more like an absolute fact. DR. RAPPAPORT: Yes, we only allow into the 8 label whatever is supported by data. So, if there’s no 9 data at this time to support that something is abuse- 10 deterrent in the community, it’s not going in the label. 11 If that changes in time, we’d be only too happy to get 12 it into the label because I think that would be 13 beneficial to the community and to prescribers. 14 DR. MORRATO: Right. I was just talking about 15 the case in which it got into the label, but then, over 16 time, you don’t see it, and -- 17 18 DR. RAPPAPORT: We had to change it. take it out. 19 DR. MORRATO: Yes. 20 DR. RAPPAPORT: 21 Sharon, did you want to add something? 22 DR. KIRSCH: Yes. Dr. Nelson? PRECISE REPORTING, LLC We could jcp 72 1 2 DR. NELSON: Thanks. Just another question for Dr. Tolliver. 3 DR. KIRSCH: Okay. 4 DR. NELSON: Just my assumption is that the 5 three-tiered approach you outlined is a regulatory 6 requirement. 7 that most companies do before they market a drug? 8 9 Is that right? DR. TOLLIVER: Or is that just something I don't know that it’s a “regulatory requirement.” I don’t think that we have 10 specific guidance in place right now. 11 of data that I provided to you today or the kind of 12 studies are what are provided to the Food and Drug 13 Administration for us to look at with respect to 14 evaluating the abuse -- 15 DR. NELSON: 16 DR. HERTZ: 17 DR. NELSON: 18 DR. HERTZ: However, the kind Okay, can I -These are -- excuse me. Sorry. These are recommendations that 19 have evolved as we started to interact with companies 20 who have been seeking different approaches to developing 21 these products. 22 discussions at advisory committees, as well. It’s been informed in part by PRECISE REPORTING, LLC So, it’s jcp 73 1 not a regulatory requirement, but it is the 2 recommendation so we can begin to understand sort of in 3 a step-wise manner what different formulations’ 4 characteristics are. 5 DR. NELSON: That makes sense. Just the 6 reason I actually asked that was because you had said 7 that if the in vitro or the clinical pharmacokinetic 8 studies show that a tablet can be compromised, then 9 there are the likeability studies, which implies that if 10 they don’t show in either of those first two phases that 11 there's a potential problem. 12 required or required or anything like that. 13 DR. HERTZ: Likeability studies aren't We know that for the Schedule II 14 Opioids, there's a certain amount of likeability 15 associated with the drug substance. 16 So, we generally don’t require that type of 17 study. 18 is, for instance, to avoid manipulation that can defeat 19 extended release characteristics and the formulation was 20 actually able to resist attempts to dose-dump, it still 21 has the abuse liability of the Schedule II Opioid. 22 If the goal of the development of a formulation So, we don’t think it’s not attractive or PRECISE REPORTING, LLC jcp 74 1 likeable, but we know that it resists dose-dumping, 2 which can contribute to some of the morbidity and 3 mortality. 4 so, they have to be able to deliver the opioid. 5 matter what, an overdose is going to be possible, it’s 6 going to be likable to some extent because it’s got to 7 deliver the opioid in order to function as an analgesic. 8 And so, it depends on what the actual intent is and what 9 the results are. The trouble is these are opioid analgesics, The reality is no product that we’ve 10 seen so far is completely capable of resisting 11 manipulation. 12 So, no So, when it is manipulated, then the question 13 is how much do we need to know about what that does to 14 the likeability and that's when we start asking for 15 likeability studies. 16 DR. KIRSCH: We have three other people on the 17 list to ask questions, but we need to go on to the next 18 speaker, and I will start in order at our next question 19 period with Dr. Wolfe, Mendleson, and Omoigui. 20 21 22 So, we’re going to go on to the next speaker, who is Dr. Anthony. Data Resources and Metrics Used to Assess Prescription PRECISE REPORTING, LLC jcp 75 1 Opioid Misuse and Abuse 2 Designing Observational Studies on Drug Abuse 3 DR. ANTHONY: Good morning. I’m aware I stand 4 between you and your break. So, I will move along. You 5 can see I’m Professor of Epidemiology at several 6 universities, as listed here, and I thank the FDA for 7 inviting me to give this talk. 8 advisory committee meetings for 35 years or more, and 9 when they asked me to talk about this topic, I wondered I’ve been coming to 10 whether there was anything I could say that the advisory 11 committee wouldn’t already have heard and why they 12 needed me to say anything else. 13 And as we looked into the issue a little bit 14 more, it turns out that my research group has been 15 working on some population rate perspectives on 16 evaluation of drug experience and that these were novel 17 with respect to some ideas that have, perhaps, not 18 previously been seen here. 19 So, what I’m going to do today is work around 20 the topic I was requested to cover, mainly focusing on 21 some conceptual issues and introducing some new ideas 22 for evaluation in the post-marketing context. PRECISE REPORTING, LLC Most of jcp 76 1 the drugs I’ve studied are not pharmaceuticals, and I’ll 2 be giving you some examples outside the domain of 3 pharmaceuticals, but I think you’ll be able to see how 4 the concepts and the research approaches can carry over 5 to the evaluation of products that at least where the 6 attempt is to improve patient safety. 7 I’ve given you an outline so you could see 8 where I’m headed. 9 outline in the interest of time, but it’s there just if 10 11 I’m not going to read through this you’d like a roadmap of where I’m going. The points of departure, I’m mindful that you 12 all are trying to focus on and clarify concepts and 13 approaches for risk management plans, and some of the 14 products that are to be evaluated have at least in 15 theory some safety advantages over already-marketed 16 products, but I also am mindful that there is a good bit 17 of knowledge and history and expertise in the room, and 18 there’s no need for me to go over issues that have to do 19 with the basics of epidemiology and design of 20 observational studies in epidemiology as one might do 21 with a less-educated audience. 22 I’m going to focus on a Cross-Sectional PRECISE REPORTING, LLC jcp 77 1 Approach that we are using mainly because that Cross- 2 Sectional Approach finesses some problems that have to 3 do with differential mortality. 4 that occurs quickly between the onset of drug use and a 5 follow-up at three, six, or one-year intervals, which is 6 the type of study that I’ve generally been doing over 7 the past three years. 8 people who are engaged in what would conform to the 9 current FDA definition of drug abuse often are not to be Mortality, for example, We find very often that the 10 found when we go to look for them in our follow-up 11 studies, and that's caused us, in part, to work on 12 Cross-Sectional Approaches. 13 that. 14 So, I’m going to focus on This will come as somewhat of a surprise to 15 those of you who have studied epidemiology because one 16 of the principles of basic epidemiology is that the 17 prospective and longitudinal design inherently is 18 superior to a cross-sectional design. 19 not always the case, and we’ll see that in this context 20 it may not be the case. 21 22 That actually is I was asked to talk a little bit about societal perspectives on drug abuse, and I have to say I PRECISE REPORTING, LLC jcp 78 1 teach my trainees not to use the concept of drug abuse 2 because it’s stigma-laden. 3 pay attention to the communication value of the words 4 that we use, and it doesn’t turn out to be very useful 5 to use the concept of drug abuse. 6 as a piece of baggage. 7 In public health, we have to So, we think of that If you go back to Latin, the term impediment 8 is impedimentum. It’s baggage that the Roman Army had 9 to carry along that kind of impeded their fast progress, 10 and abuse is a piece of baggage. 11 and pull out selected facets of abuse that can be 12 studied, but as a scientific concept and as an object of 13 study for an observational study, it probably is not 14 very helpful to us. 15 it. 16 We can reach inside Congress, however, is very fond of Same is true for abuse liability, but, here, 17 the problem is mainly that it conveys the idea that 18 abuse liability or dependence liability is a property of 19 the drug, and there are more productive ways of thinking 20 about becoming dependent or becoming a drug abuser if 21 you want to think about it that way, and I’ve sketched 22 one here on this slide. PRECISE REPORTING, LLC jcp 79 1 So, in coverage of societal perspectives on 2 drug abuse, I think we have to realize that this is a 3 pejorative stigma-laden term, and it may be wise for us 4 not to think about abuse-deterrent products because it’s 5 unlikely that any product is going to be deterrent of 6 abuse across all the range of facets of abuse that we 7 might be studying. 8 9 In my own work, I focus on drug dependence, and a point of departures, this early study by L. 10 Lasagna in which he tried to find out what the response 11 of healthy volunteers would be to placebo and a profile 12 of other drugs, one of which was heroin. 13 What you can see here is that among 20 exposed 14 to heroin, 4 said they would have liked to repeat it and 15 70 of the 20 would not like to repeat it at all. 16 you can see that in contrast to the popular conception 17 about heroin, this isn’t a drug that if you take it 18 once, you’ll become hooked, and certainly that is true 19 of the marketed opioid analgesics that are in discussion 20 here. 21 22 So, Now, using that as a point of departure, in our studies on drug dependence, which is not exactly the PRECISE REPORTING, LLC jcp 80 1 same as wanting to repeat a drug experience, we focus on 2 a definition that has to do with three facets: 3 disturbance of the mental life. 4 comes into play. 5 asking people about their mental life, and these are 6 obsession-like ruminations about the drug experience and 7 cravings. 8 9 One is a This is where craving The only way we’d know about it is by Another domain of disturbance and dependence running together with those in the mental life is 10 compulsion-like behavior. 11 manifest, you could actually see it. 12 necessarily have to ask about it, but it’s like a 13 compulsion in psychiatry, and it may in some cases be a 14 compulsion. 15 behavior. 16 So, that here, something You wouldn’t Its repetitions are rounds of drug-involved And then, finally, the third facet of this 17 syndrome is neuro-adaptation, as typically manifest in 18 tolerance, which might be subjectively felt or 19 demonstrated in the lab or a withdraw syndrome. 20 you think about this syndrome definition, what we’ve 21 tried to do is ask how often people who use different 22 types of compounds develop this drug-dependent syndrome. PRECISE REPORTING, LLC So, if jcp 81 1 This is a summary of work that we’ve done. 2 We’re updating these values with more recent data, but 3 we’re not getting much change, and if we work our way 4 from the top to the bottom, if we look among people who 5 smoke tobacco cigarettes, even one cigarette, and ask 6 what proportion of them become dependent or develop a 7 dependent syndrome, as we’ve just defined it, it’s about 8 one in three. 9 one in four or five, and notice the similarity to the 10 If we do the same for heroin, it’s about Lasagna experimental evidence. 11 Going around the circle, you can see crack- 12 cocaine is followed by a crack-dependent syndrome 13 slightly more often than cocaine-hydrochloride powders, 14 followed by a cocaine-dependent syndrome. 15 work our way around to the opioid analgesic drugs to 16 about 1 in 11. 17 And we can I should note here that the context is 18 extra-medical use. This is not a medically-prescribed 19 user. 20 were to include in the denominators people who are 21 getting these medicines from the doctor in a prescribed 22 context, such as a pain management clinic. I think the values would be much smaller if we PRECISE REPORTING, LLC jcp 82 1 These are people who will acknowledge to us 2 that they’ve used it outside the boundaries of what's 3 been prescribed either for feelings that it produces or 4 they’ve taken it for reasons the doctor didn’t prescribe 5 it. 6 foot surgery, and they woke up in the morning and felt a 7 hangover after heavy drinking and took the medicine to 8 help relieve that hangover. 9 medical use, and we’re including those people in these So, they might have gotten a pain medicine after That would count as extra- 10 denominators. 11 the medicine as the prescriber intended. 12 We’re not including people who are taking So, these are the relative proportions that we 13 get when we go out into the population and accumulate 14 over the experiences of drug-users drug-by-drug, looking 15 at each form of the dependent syndrome. 16 In terms of the design, you’re going to hear 17 more about this because the National Surveys on Drug Use 18 and Health, which will be discussed next, is often a 19 source of data for our studies. 20 pre-designated U.S. population studies. 21 working in 22 different countries. 22 estimates for other countries before too long. Typically, these are We’re now So, we should have PRECISE REPORTING, LLC We have jcp 83 1 multi-stage area probability sampling of dwelling units, 2 and then probability sampling of the respondents. 3 recruit with IRB-approved protocols. 4 standardized assessments that are either anonymous or 5 confidential. 6 computer-assisted self interviews or personal 7 interviews. 8 prewritten items, and routing patterns, branching 9 patterns through the assessments so that we can follow- There are Nowadays, they tend to be The assessments include standardized, 10 up and give details about experiences as they're 11 expressed. 12 We For the drug-dependent syndrome, we have what 13 we call testlets, each facet of the syndrome is 14 evaluated with multiple items. 15 cumulative incidence proportions for each drug group, 16 and we pay attention to variants in the constraints. 17 We then estimate Now, one of the questions that was raised in 18 pre-discussions of this talk is whether there process 19 phenotypes on the way to the full dependent syndrome 20 that might be investigated in the marketing of new 21 products, and the answer is yes, there are, and I’m 22 going to focus for the next few minutes on that topic. PRECISE REPORTING, LLC jcp 84 1 They typically are going to require fairly large samples 2 to identify them. 3 The idea of the process phenotype can be seen 4 in this graphic. It’s a stage transition model. At 5 number one, you see the onset of drug use. 6 two, a building up of count of drug experiences for 7 those who use the drug more than once. 8 sometimes after repeated drug experience, you’ll get the 9 formation of the clinical features that I described At number Number three, 10 earlier, the craving, the ruminations, the compulsive- 11 like behavior, and so on. 12 those features can coalesce into a syndrome which we 13 would call the drug-dependent syndrome. 14 number five, there could be secondary complications of 15 that syndrome. 16 And then, number four, that And then, at Now, of course, there is potential cessation 17 of use at each step, and one of the questions is how 18 many people use the drug once and then will continue to 19 use the drug and how many people are not likely to use 20 it again? 21 rate perspective on drug abuse as opposed to the 22 individual risk perspective. This brings us to the concept of a population The heritage of this PRECISE REPORTING, LLC jcp 85 1 concept goes back to an early geneticist, Wilhelm 2 Johannsen, who coined the terms “genotype” and 3 “phenotype,” and most of the time nowadays, because of 4 prominence of molecular genetics and biology, we think 5 about the individual type of genotypes and phenotypes. 6 Johannsen thought more broadly, and he thought about a 7 population perspective on phenotypes and thought of them 8 as population characteristics. 9 A related idea was introduced more recently by 10 Epidemiologist Rose, who drew distinctions between 11 causes of incense and causes of cases. 12 In the interest of time, I’m not going to be 13 able to say much about this, except I’m going to 14 illustrate with an example that has to do with drug 15 dependence, and you have the references here if you’re 16 interested in that. 17 If you want to think about this idea of the 18 process phenotype, you can think about a population that 19 studied from birth to death. 20 take a marketed product one time, never repeat it again. 21 Others will repeat it. 22 quickly after the first try, which could be a Some of them try a drug or Sometimes this will happen PRECISE REPORTING, LLC jcp 86 1 manifestation of liability to dependence, or it could 2 happen after a lengthy lag interval. 3 If we observe these people longitudinally to 4 death, we can know the count and lag times of these drug 5 experiences, but in a cross-sectional survey, we cannot 6 know. 7 someone was recently using the drug, but we wouldn’t 8 know among those who were not using it whether they 9 would ever use it again. 10 We take a slice in time and we’d see whether Well, if we set up the problem appropriately, 11 we actually can estimate those who would never use it 12 again not at the individual level, but at the subgroup 13 level, and if we think about ethanol as a drug 14 understudy and if we were to take into account potential 15 protection, and here, I’m thinking of the protection 16 that you might like to have in a so-called abuse- 17 deterrent formulation, what we would expect to see with 18 respect to ethanol is that in population subgroups that 19 have an excess prevalence of a null variant of liver 20 metabolizing enzyme alleles, we would have less 21 likelihood to become dependent upon that drug, and this 22 should be manifest in a process phenotype that shows up PRECISE REPORTING, LLC jcp 87 1 very soon after the onset of the drug, and this is where 2 I think these ideas might be portable to the context of 3 post-marketing surveillance of new products where the 4 goal is to try to confer some protection by virtue of 5 the product characteristic. 6 hypothesizing has to do with genetic variation and 7 responses to ethanol. Here, the protection we’re 8 So, I can't point to any Asian-American in a 9 cross-sectional sample and declare whether this person 10 might be in the future a persistent drinker or never 11 again drink, but if I look at the Asian-Americans who 12 very recently have started to drink and I ask whether in 13 the most recent interval of time, say 30 days, whether 14 they have had even one drink and then what is the rate 15 of drinking in those 30 days, I then can estimate 16 whether Asian-Americans are over or under-represented in 17 a group of people who are not likely to drink, again, 18 versus those who are likely to drink again. 19 estimate the rate of drinking conditional on the 20 membership in these classes. 21 22 I can also And what you can see here in data that are just submitted, the PP is the persistence parameter of PRECISE REPORTING, LLC jcp 88 1 this regression model, and we get an inverse 2 association, non-Hispanic, Asian-Americans compared to 3 non-Hispanic whites are less likely to persist in their 4 use of ethanol, and we hypothesize this is related to 5 the pharmacokinetic substrate. 6 later. 7 I’ll come back to that The RR parameter is the rate ratio, so, 8 conditional upon membership in the persistent drinking 9 or using class, we have a negative sign; the rate for 10 the Asian-Americans is lower than the rate for whites. 11 And so, here we’re seeing before anyone has developed an 12 alcohol-dependent syndrome or actually a few of these 13 individuals will have developed it, but even if we set 14 aside the alcohol-dependent individuals, we can see a 15 subgroup in which there is an apparent protection 16 against the risk of becoming drug-dependent. 17 In theory then, in terms of post-marketing 18 surveillance, this type of approach could be used within 19 12 to 24 months of release of the drug in order to see 20 if, in fact, we would see manifestations of reduced risk 21 of these process phenotypes. 22 We don't find these relationships for Asian- PRECISE REPORTING, LLC jcp 89 1 Americans for tobacco, cocaine, or cannabis. 2 studying them for opioid analgesic compounds, but I’m 3 not ready to report that yet. 4 We’re now Now, I will show you some process phenotypes 5 that have to do with the opioid analgesic drugs, and 6 this may overlap a little bit with a talk that’ll be 7 given later on, but these process phenotypes are the 8 actual clinical features of drug dependence, and they’re 9 listed here. 10 So, without asking who has developed the drug- 11 dependent syndrome, we can ask about the accumulative 12 incidents soon after onset of use of each of these 13 clinical features of drug dependents as steps on the way 14 to the full phenotype. 15 Here, the subgroups under study are kids, 16 adolescent onset drug-users, kids who start using these 17 drugs; in this case, it’s opioid analgesics, before age 18 18, and the contrast group is those who start as young 19 adults or a little later. 20 And what you can see, if we look across the profile of 21 these process phenotypes, at the individual clinical 22 features of dependence, among people with or without Most of them are 18 to 25. PRECISE REPORTING, LLC jcp 90 1 respect to whether they’re become dependent, we see five 2 aspects where the adolescent onset kids seemed to be at 3 greater risk. 4 drugs, spending a lot of time getting over the effects 5 of the drug, needing more drugs to take to get the same 6 effect, having emotional problems connected with their 7 drug use, reducing other activities, non-drug 8 activities, in order to use drugs, and then having 9 withdrawal experiences. 10 One is getting over the effects of the So, this would be another example of a 11 phenomenon of drug dependence related to drug 12 dependence, but not the full dependent syndrome that 13 could be studied in the post-marketing context within 12 14 to 24 months after onset of use. 15 estimating these parameters for a lag time of 180 days 16 elapsed since first use of the drug. 17 data in a fairly timely fashion via cross-sectional 18 surveys on these topics. 19 We’re actually So, we could get And then, I'm here just going to contrast 20 male/female differences, and here, we’re looking at the 21 newly incident opioid analgesic users using outside the 22 context of medical practice. They’re about half and PRECISE REPORTING, LLC jcp 91 1 half male and female, and what you can see here on the 2 left of the gray bars for males and females combined, 3 the white bar, you can’t see very well, but it’s the one 4 in the middle is for males and the yellow bar is for 5 females. 6 are more likely to develop these clinical features. 7 There’s one exception of the set. 8 see how this type of approach, evaluated within 12 to 24 9 months after onset of the use of the drug can be used to 10 Here, you can see that, in general, females And again, you can study steps on the way to becoming drug-dependent. 11 Now, in all of these studies, there’s a need 12 to pay attention to potentially confounding variables. 13 If you consider the Asian-American example, we’re 14 hypothesizing a protection based on a null variance of 15 the alleles, but Asian-Americans are different from 16 other parts of the population in other ways. 17 often more family cohesion, more family attention, 18 parental monitoring, and the like, and those would have 19 to be taken into account before we could attribute the 20 protection to one factor or another. 21 22 There’s I do want to note though that in this context, there is a potential for over-control of what is PRECISE REPORTING, LLC jcp 92 1 suspected as a confounding variable. 2 says well, why don’t you control for school dropout or 3 income levels, well, those may be responses to the drug 4 use. 5 So, if someone I made some notes for future directions, but I 6 want to turn the time over to you for questions, and if 7 anyone wants to talk about these issues, I’d be happy to 8 talk about them. 9 emphasize is that there are other domains of The one point that I was encouraged to 10 pharmaceutical products where, in theory, manufacturers 11 are trying to improve the safety profiles, and we’re not 12 just talking about FDA, but also about other devices, 13 other consumer commodities. 14 useful to look outside of the narrow boundary of the 15 drugs that we pay attention to in this advisory 16 committee in order to borrow some experience. 17 And it may be that it’s Now, the Energy Department, for example, is 18 asking for some evidence of the firm’s financial 19 capacities to deal with the damage if something happens 20 in offshore oil drilling. 21 insurance plans in order to protect the society and the 22 firms against a potential hazard or catastrophe with a We could think of collective PRECISE REPORTING, LLC jcp 93 1 product that, from a theoretical and pre-marketing point 2 of view, looks like it has advantages. 3 perspective is brought to bear in the advisory 4 committee’s work on these problems. 5 to block out the gate and not let any new products out 6 because we’re worried about a repetition of past 7 experiences, but the American public needs products and 8 needs innovations that are trying to improve safety of 9 these products, and we may need to look outside narrow I hope that that The tendency may be 10 boundaries of abuse context in order to borrow from 11 other domains of FDA or federal government regulation. 12 Thank you very much. 13 DR. KIRSCH: Thank you. Before you leave the 14 podium, can I assume that you’ll be here throughout the 15 day? 16 DR. ANTHONY: 17 DR. KIRSCH: 18 Yes. So that at the question period, you’ll be able to answer questions? 19 DR. ANTHONY: Yes. 20 DR. KIRSCH: Okay. 21 to take a break. 22 duration. So, with that, we’re going The break is going to be 15 minutes in Committee members, please remember that there PRECISE REPORTING, LLC jcp 94 1 should be no discussion of the meeting topic during the 2 break amongst yourselves or with any other member of the 3 audience. And we will resume at 10:32. 4 (Break.) 5 DR. KIRSCH: While you are taking your seats, 6 I will ask our next speaker, Dr. Woodward, to approach 7 the podium, please. 8 (Pause.) 9 DR. KIRSCH: Okay. Dr. Woodward? 10 Substance Abuse and Mental Health Services 11 Administration: Resources and Methods 12 DR. WOODWARD: First of all, thank you for 13 inviting me to talk briefly about the datasets that are 14 of interest to FDA and the field. 15 CBHSQ is an acronym you may not be familiar 16 with. 17 and Quality. 18 Office of Applied Studies. 19 July, and I’ll explain a little bit more about that in a 20 second. 21 22 It’s the Center for Behavioral Health Statistics You may be more familiar with the old The name was changed in I have no conflicts of interest being a good government bureaucrat, keeping my head as low as I can. PRECISE REPORTING, LLC jcp 95 1 Let’s see. The Center has more 2 responsibilities than the old office did, but I’m not 3 going to be talking about the new responsibilities; I’m 4 going to be talking about the three main datasets that 5 the CBHSQ, or, if we had IT would be CBHSQIT, has, and 6 they were consistent with the Office of Applied Studies. 7 There are three main datasets, the National Survey of 8 Drug Use and Health, the Treatment Episode Dataset, and, 9 finally, the Drug Abuse Warning Network that I’ll talk 10 about. I’m going to give a very high-level, 10,000 foot 11 view of the datasets, given the time that I’ve got. 12 The first is the National Survey on Drug Use 13 and Health, as you can see from the slide; these are the 14 main features of the dataset. It’s an interview of 15 about 68,000 people per year. It takes an hour, it’s 16 very detailed, it’s computer-assisted so that we’re able 17 to provide the questionnaire in a number of different 18 languages. 19 So, there’s a component of the population we don’t 20 capture in the NSDUH that's of interest to this field so 21 that we’re able to present prevalence/incidence data for 22 the nation and in each of the states. It’s only for people over the age of 12. PRECISE REPORTING, LLC In each state, we jcp 96 1 can provide direct estimates. 2 estimated. The others are indirectly 3 There are two main changes of historical note. 4 In 1999, the computer-assisted approach was used so that 5 produced a break in the trend. 6 payment of $30 per individual, and both times, the 7 prevalence went up, the response rate improved. 8 9 2002, we introduced a The overall response rate for this large a survey is just about 70 percent, which we’d like to 10 improve, but given how difficult it is to collect survey 11 information these days, it’s pretty sound. 12 The household survey started out with an 13 emphasis on illicit drug use. So, what I wanted to do 14 is to just briefly review the kinds of information 15 that's collected. 16 use and non-medical use of prescription drugs, there’s 17 recency of use, frequency of use as much as daily so 18 that we can report on, say, somebody who uses every day 19 in a given year. 20 drugs are captured. 21 drug use is captured in the diagnostic, statistical, 22 manual criteria. So, including both with illicit drug The initiation data for particular Dependence and abuse resulting from Also, we collect whether or not PRECISE REPORTING, LLC jcp 97 1 treatments received is the result of a particular 2 problem. 3 As far as prescription drugs, the strategy of 4 the NSDUH is to try to report on the four major 5 categories of drugs: 6 reliever, stimulants, sedatives, and tranquilizers. 7 There is information collected on specific 8 pharmaceuticals, including brand names and generic 9 drugs. therapeutic classed as a pain- What the respondent gets is a set of 10 photographs, what we call pill cards, showing each of 11 the drugs by their categories. 12 have. 13 other drugs. 14 They report what they If they don’t see anything, they can type in The reporting of the drugs is largely 15 aggregated at the therapeutic class level. 16 the NSDUH is trying to improve the definition of drugs 17 of what is considered drug use so that we won't really 18 be able to collect information on dose, but we will be 19 able to get a better sense of the type of use, whether 20 it’s over medication or misuse. 21 really can’t distinguish very well in that area. 22 Right now, Right now, the NSDUH The second major dataset that I want to talk PRECISE REPORTING, LLC jcp 98 1 about is the Treatment Episode Dataset. 2 larger data collection system called the Drug and 3 Alcohol Services Information System, DASIS, which has 4 some other components with it. 5 administrative dataset that states report to our office. 6 We crosswalk the data to make sure there’s consistency 7 in what we collect from the data. 8 system, very few states collect individual patient 9 identifiers. 10 It’s part of This is an It’s an episode data So, we can’t really track patients very well. 11 Most of it is treatment admission data. 12 some discharge data. 13 reporting discharge information. We do collect A lot of the states are now 14 It’s facilities that the states keep track of, 15 and these are largely public-funded, specialty treatment 16 facilities or clinics, if you will. 17 something like 13,000 of them throughout the nation. 18 course, with the economic downturn, there are fewer. 19 We estimate that we’re picking up about 80 percent of 20 the facilities. 21 facilities, as much information on them. 22 slightly under 2 million admissions every year. And there are Of We don’t have the private-funded PRECISE REPORTING, LLC And there are jcp 99 1 The information that's collected, as you can 2 see, our demographic variables, the discharge dataset 3 includes some socioeconomic information, employment 4 history, and insurance coverage. 5 between admissions and discharge records. 6 There is a link So, the information in the state report is the 7 three main substances, at admission, route of 8 administration, how the drug’s taken, frequency of use, 9 age at first use. Treatment variables largely focus on 10 the type of treatment, where the individual’s going. 11 There is some limited information on treatment outcomes, 12 length of stay. 13 useful, if you will, for the field because there isn’t 14 really information collected on brand names or 15 formulations. 16 we can’t really disaggregate further. 17 So, this particular database isn't as It’s general. There are groupings that For example, opiates are aggregated into a 18 class, and we can’t really drill down further, if you 19 will. 20 analgesics. 21 field. 22 In a minority of states, 16 report on opioid So, it’s probably a limited use to this The last dataset I want to touch on briefly is PRECISE REPORTING, LLC jcp 100 1 the Drug Abuse Warning Network, which has been around, 2 like the NSDUH, for quite awhile. 3 was to design a sentinel public health surveillance 4 system. 5 health problems that are captured in emergency 6 departments throughout the nation. Initially, the intent It’s now sort of focused on nationally public 7 The information is collected by trained 8 recorders who go to about 250 hospitals throughout the 9 nation. Some of the hospitals have their own recorders 10 who are also trained to report the data that we collect 11 directly from the hospital emergency record. 12 we focus on, it’s a fairly short data collection form 13 with about 20 data elements. 14 has drugs reported either as a direct cause or 15 contributing factor, either determined from toxicology 16 records or what’s in the notes in the reports from the 17 medical record. 18 And what It is where an ED visit There are about 4 million drug-related visits 19 that we capture each year. That's out of about 110 20 million ED visits throughout the nation. 21 4 percent. 22 effort to try to get to that 4 percent in terms of So, it’s about So, as you can imagine, it takes a lot of PRECISE REPORTING, LLC jcp 101 1 screening. 2 What we do is to select a representative group 3 of EDs throughout the hospital. 4 inclusion is short-term, general, non-federal hospitals 5 with 24-hour emergency departments. 6 12 metro areas, and we have a remainder sample of 7 hospital EDs so that we can, along with the metro areas, 8 provide nationally representative information for the 9 country. 10 The criteria for We over-select for Now, the estimates include the usual 11 statistical adjustments for sample design. 12 strata size is the most important. 13 hospitals where we don’t have data reporting. 14 adjusted for if the hospitals don’t report for a full 15 year, we make adjustments for that. 16 introduced one in three sub-sampling; that is every 17 third ED record is reviewed or screened. 18 simply so that we can save costs without sacrificing 19 efficiency. 20 do that. 21 22 The biggest Strata size is They’re We also have And that's There isn’t much increase in bias when we This gives a sense of the information that we collect, how it breaks out for use in analysis. PRECISE REPORTING, LLC In the jcp 102 1 left, you can see pharmaceuticals under “medical use.” 2 This represents the adverse effects that we can pick up. 3 Non-medical use can be broken into pharmaceuticals, 4 illicit drugs, and alcohol. 5 clarify that if the only drug onboard, if you will, that 6 is associated with the ED visit for somebody under the 7 age of 21, we don’t have any age restrictions as the 8 NSDUH does. 9 over 21, we only capture alcohol if there are other We capture that, but if the individual is 10 drugs present. 11 basically an illicit drug. 12 For alcohol, I need to In alcohol for somebody under 21 is Non-medical use is defined, as you can see, 13 exceeded prescribed or recommended doses. 14 using a particular drug when it was prescribed for 15 somebody else. 16 administration of a drug to somebody intentionally, as 17 well as any substance abuse that we pick up from the 18 medical record. 19 non-medical use, and we include suicide thought and 20 plans. 21 22 Somebody An intentional poisoning, intentional We exclude suicide attempts and the Finally, for the DAWN, the value of the DAWN to the field is that it provides detailed brand level PRECISE REPORTING, LLC jcp 103 1 specific drug information at the ED visit, which is 2 important because any ED visit where there are drugs as 3 a part of the visit indicate a fairly serious 4 consequence. 5 We aren't able to collect dose and source of 6 drug. As you can imagine, it’s very difficult to 7 collect that information if the patient really isn’t 8 fully conscious. Oftentimes, patients, they aren't able 9 to provide that. If they bring in a bottle, they may 10 know how much they’ve taken, but, oftentimes, they 11 don’t. 12 The ED record often doesn’t have enough 13 specificity for the recorder to enter the detail level 14 of information that we would like to collect. 15 often, there’s some ambiguity. 16 the databases -- most of the drugs reported are fairly 17 well documented. 18 adjust for street level names, and we use that to try to 19 get to as great a specificity and the types of drugs 20 that are included in the ED visit. 21 22 So, By and large, I think We used the Multum Lexicon which we If you want to find out more, you can go to the Web Site under the old OAS. The Web Site, we’re PRECISE REPORTING, LLC jcp 104 1 trying to improve it. 2 information and various publications. 3 But it will provide a wealth of We have public use files for the NSDUH and for 4 the TEDS under SAMHDA. That's the Substance Abuse 5 Mental Health Data Archive. 6 University of Michigan public data file archives, and 7 it’s a very powerful source of information. 8 build your own tables, download your own files, and do a 9 certain amount of statistical analysis, variance It’s part of ICPS, our You can 10 calculations, regressions with the SAMHDA. 11 to put the more current DAWN information on there, and 12 we hope to have that public use files, data tables ready 13 in the next few months. 14 We’re trying Finally, I just wanted to say that we are 15 working with FDA through an interagency agreement to 16 provide data requests. 17 if we’re subcontractor to FDA to try to answer specific 18 requests. 19 try to extend our limited staff resources, and we have 20 been working with FDA to make them understand to make 21 sure that we comply with the federal privacy and 22 confidentiality laws. Sometimes, our office feels as So, we’ve set up an interagency agreement to Even though we don’t collect any PRECISE REPORTING, LLC jcp 105 1 information that's directly identifiable, that is where 2 there’s personal health information collected, we are 3 aware that through triangulation and other data sources 4 it may be possible to identify individuals, which is 5 against every federal law dealing in that area. 6 we’re very careful about that. 7 8 9 10 And that's it. So, It’s a very high-level level, and, so, once again, thank you for allowing me to talk. DR. KIRSCH: Thank you. Our next speaker is Dr. Dormitzer. 11 Available Data Resources to Assist in Measuring Abuse 12 Behaviors, Patterns, and Outcomes 13 DR. DORTMITZER: Hi, good morning. My name is 14 Cathy Dormitzer. I’m an epidemiologist in the Division 15 of Epidemiology in the Office of Surveillance and 16 Epidemiology. 17 I’ll present standard data sources, both public and 18 proprietary for numerator and denominator data. 19 give a brief description of the Prescription Drug 20 Monitoring Programs, and then finish with a summary of 21 the challenges with the current data sources. 22 But, first, what are numerators and I will start with a brief background. PRECISE REPORTING, LLC I will jcp 106 1 denominators, and why do we need them? 2 measure outcomes. 3 numbers of events of interest, and denominators provide 4 some context of the burdens of these outcomes. 5 Well, numerators So, in absolute numbers, what are the These are a few data sources that contain 6 information on -- there are very few data sources that 7 contain information on both numerator and denominator 8 information and are actually able to link them. 9 use both numerator and denominator sources of data. 10 So, we So, the data sources presented are ones that 11 measure events related to drug misuse and abuse. 12 are reports on the non-medical use of drugs, events such 13 as emergency room visits, and outcomes such as drug 14 dependence or drug-related deaths. 15 measures of misuse and abuse were mainly used for 16 illegal drugs, but now we are using these same data 17 sources for prescription drugs that are approved and 18 regulated by the FDA. 19 They In the past, these So, standard data sources, most of them are 20 nationally representative, usually multi-stage 21 probability sample. 22 been used for many years, and they’ve also been There are data sources that have PRECISE REPORTING, LLC jcp 107 1 2 presented both by Dr. Woodward and by Dr. Paulozzi. It’s not exhaustive. It does include the 3 sources that FDA has used in the past though, and we use 4 them to examine drug abuse outcomes. 5 So, as you can see from the list, half of them 6 are funded by SAMHSA and were previously presented by 7 Dr. Woodward. 8 the number of events related to misuse, abuse, and 9 related outcomes, and all of these are publicly-funded 10 They are sources of numerator data. So, data sources. 11 So, DAWN provides national estimates of drug- 12 related emergency room visits, and they also provide it 13 as numbers per 100,000. 14 the SAMSHA-defined constructs of non-medical use of 15 pharmaceuticals that Dr. Woodward related to, and they 16 are the cases that are classified as over-medication and 17 other, as well as malicious poisoning, but malicious 18 poisoning is usually very low. 19 abuse-related, and, so, that's called NMUP. 20 in addition to NMUP, there’s also ALLMA, which are all 21 the NMUP cases plus ED visits where alcohol or illegal 22 drugs were present in the patient. And particularly important are But they are considered And then, And FDA also PRECISE REPORTING, LLC jcp 108 1 examines DAWN medical examiner data. 2 are not nationally representative, but they do provide 3 data on a consistent panel of medical examiners on a 4 number of drug-related deaths. 5 Now, these data And the strengths of DAWN are that it is 6 nationally representative. 7 specific to substance, formulation, and sometimes even 8 brand. 9 route of administration, but it’s very limited. 10 And we do have data that's And it’s very limited, but there is also data on Now, the limitations of these data is that 11 there is lag time because national estimates are 12 generally not available until 9 or 10 months after the 13 end of the calendar year from which the data is 14 produced, and ME data did not provide information on 15 drug formulation, and it’s also not nationally 16 representative. 17 And the last important limitation of DAWN data 18 is that it provides data on misuse and abuse that 19 resulted in a medical outcome, either an ED visit or a 20 death, not on the behaviors. 21 22 The National Survey on Drug Use and Health does collect data on drug abuse behaviors, and the PRECISE REPORTING, LLC jcp 109 1 questions are taking the drug not prescribed for you or 2 just for the feeling it caused? 3 questions on prescription drugs, but it’s pain- 4 relievers, tranquilizers, stimulants, and sedatives. 5 Now, it does ask these And notice the strengths are that it collects 6 data on behaviors. So, we’re interested in that. And 7 it collects it directly from the respondents, and it’s 8 behaviors that may not have resulted in a medical event. 9 The limitations are that it’s got the same nine-month 10 lag time and that it focuses on drug classes. 11 pain-relievers, not specific opiates, although there are 12 questions that were added for OxyContin in 2002. 13 So, it’s The last SAMHSA dataset I’ll be discussing is 14 TEDS, and it collects data on the number and 15 characteristics of a person’s admittance to substance 16 abuse treatment program. 17 the top three substances of abuse at the time of 18 admission. 19 administration, as well as frequency of use and age of 20 first use. So, we do get information on We get information on route of 21 And the strengths of TEDS are that it collects 22 data on drug substance, but it’s fairly limited in terms PRECISE REPORTING, LLC jcp 110 1 of classes. So, it’s not always very specific, and it 2 also does provide some insight on the public health 3 burden on opioid analgesics. 4 The limitations are that these data are not 5 always completely nationally representative, and that 6 fact that 16 states report on specific opiates, but the 7 rest don’t. 8 9 Now, Monitoring the Future, which is conducted by the Institute for Social Research at the University 10 of Michigan, and it’s funded by NIDA, which is the 11 National Institutes on Drug Abuse, is an in-school 12 survey of drug abuse behaviors, attitudes, and values of 13 high school and college students, as well as young 14 adults, and it includes questions on attitudes and 15 perceived harmfulness of prescription drugs, including 16 opiates. 17 The strengths on Monitoring the Future are 18 that it examines drug abuse behaviors among a population 19 that's usually recently begun to start using and abusing 20 drugs. 21 conducted over many years. 22 It’s nationally representative, and it’s been The limitations are that respondents are asked PRECISE REPORTING, LLC jcp 111 1 on their use for drug classes, such as sedatives, 2 amphetamines, or narcotics other than heroin, not on 3 specific drugs, formulations, or brands. 4 asked on Vicodin and OxyContin, and usually adolescents 5 really can’t distinguish between brand or generic. 6 it’s not a perfect measure. 7 on youth that attend school. But they are So, And it only collects data 8 The Adverse Event Reporting System is a 9 database of FDA’s Post Market Safety Surveillance 10 Program for all approved drugs. 11 monitor adverse events and medication errors that might 12 occur with these products. 13 some adverse event and medical error reports directly 14 from health care professionals and consumers, but 15 manufacturers, by regulation, are required to send these 16 reports to FDA. 17 And it’s used to It’s voluntary and receives And the strengths of AERS is that it can 18 provide information on signals related to drug misuse, 19 abuse, and dependence, but it’s not complete reporting. 20 In fact, there is substantial underreporting, so, it 21 cannot be used as a surveillance tool. 22 Dr. Paulozzi actually cited most of these PRECISE REPORTING, LLC jcp 112 1 datasets, but he also presented data on overdose deaths 2 involving opioid analgesics from the National Vital 3 Statistics. 4 certificates, and it includes information on all deaths 5 in the United States. 6 It’s data extracted from death And the strengths of these data are these data 7 are not a sample. It is the true population. But the 8 limitations are that is provides data on opiates as a 9 class except for methadone, and that the data are 10 usually available a few years after the calendar year 11 has ended. 12 So, that's another long lag time. Now, I will discuss data sources that are 13 proprietary data sources, and these are just some 14 examples. 15 information, such as the Internet. 16 nationally-representative, although they are increasing 17 their coverage. 18 of both Sponsors’ proposals, they will be probably 19 giving more detailed prescriptions. 20 going to be brief. 21 22 Some are newer, may use different sources of Some may not be And since these data sources are part So, again, I’m NPDS, which is the National Poison Data System, provides data on poison exposure phone calls PRECISE REPORTING, LLC jcp 113 1 into poison control centers. 2 poison control centers across the United States, and 3 includes both calls on drug exposure as well as calls on 4 information for specific drugs. 5 They collect calls from It is a large data source, and it can provide 6 some data that is specific, even down to the level of 7 formulation and sometimes brand, but this information is 8 limited, and there is lag time for the annual reports. 9 RADARS was developed in 2002 by Purdue, but in 10 January of 2006, RADARS became a non-profit and 11 operation administered by the Rocky Mountain Poison and 12 Drug Center, and a representative from RADAR will be 13 presenting today, so, I’m not going to go into the 14 details of these data. 15 that it does provide timely data that can be brand- 16 specific, it’s published their findings in numerous peer 17 review journals so others in the research community have 18 reviewed their work, and that's also true for actually 19 most of the datasets. 20 components of their data are nationally representative. 21 22 But one strength of RADARS is And one limitation is not all NAVIPPRO is a system that collects information on prescription opioid abuse, and it’s also included in PRECISE REPORTING, LLC jcp 114 1 the Sponsor’s proposal, so, I’m not going to be 2 providing a lot of detail. 3 One of their strengths is also a limitation 4 because they gather a great deal of detailed information 5 from people seeking treatment for their drug dependence. 6 So, these are people who actually have quite a bit of 7 knowledge about the abuse of opiates, and that provides 8 information, and that's probably, like I said, their 9 biggest strength and their biggest limitation, but that 10 might not be generalize-able to the population as a 11 whole. 12 So, now I will present denominator data. When 13 providing the number of drug abuse events, the default 14 denominator is the U.S. population. 15 be refined by getting the number of events per 100,000 16 population, by age group, as well as the geographic 17 region, such as state or ZIP code. 18 information from the amount of drug utilization either 19 from ARCOS, which is the Automation of Reports 20 Consolidated Order System, which is ARCOS, and that's 21 administered by the EA, and that was also presented by 22 Dr. Paulozzi. And these data can We can also get But FDA also purchases excess to drug PRECISE REPORTING, LLC jcp 115 1 utilization data sources. 2 So, the U.S. Census data is readily available, 3 it’s easily understood, and it provides data on the 4 public health burden of these events and provides data 5 on groups that are at risk. 6 provide data on drug utilization. 7 be considered exposure; how much of the population is 8 exposed to the drugs that are being examined. 9 It does not, however, Drug utilization can So, ARCOS, which is DEA's dataset, it tracks 10 all schedule drugs that are in Schedules I, those are 11 the illegal drugs, and II, and both morphine and 12 oxycodone are C2s, as well as all narcotics from all the 13 schedules. 14 all levels. 15 the time they are delivered to the pharmacies. 16 most part, it reports the number in kilograms sold by 17 drug. 18 formulation and substance. 19 And opiates are considered narcotics, so, And it is from the point of manufacturer to For the This system also has some information on This is not a projection. This is all drugs 20 sold. The limitation is that it does not provide 21 information on the numbers of prescriptions sold. 22 it’s once it’s reached the pharmacy, we don’t have any PRECISE REPORTING, LLC So, jcp 116 1 2 further information with this data source. So, FDA has purchased access to many data 3 resources that provide estimates on the amount of drug 4 sold by substance, formulation, and brand, and this is 5 detailed information that includes data on the 6 prescriber, the patient, and the indication for which 7 the drugs are prescribed. 8 information on concurrent use of multiple drugs. 9 And it also provides So, these data are used as denominators, and 10 it does put drug abuse events into context. 11 these data to assess the amount of risk within the U.S. 12 population. 13 drug-prescribing patterns and abuse, and, also, we use 14 it to assess risk management, plans, and practices such 15 as labeling. 16 And we use We are also using these data to understand So, it does provide very specific information 17 on substance formulation and brand. The limitations are 18 that they are projections. 19 And we still don't know how the drug is taken by the 20 individual patient. 21 in their medicine cabinet and that could have resulted 22 in a family member taking it, or they could have sold It is not the amount sold. The patient may have left the drug PRECISE REPORTING, LLC jcp 117 1 it. And all we need to keep in mind that these data are 2 in no way linked to the outcomes associated with drug 3 abuse. 4 Okay, now I will discuss the Prescription Drug 5 Monitoring Plans, which are the PDMPs. And PDMPs are 6 statewide electronic databases that collect data on 7 controlled substances that are dispensed in pharmacies 8 by state. 9 are currently 34 states that receive federal funding. They were first started in 2002, and there 10 And there are two federal funding sources for the PDMPs. 11 The first is the Harold Rogers Prescription Monitoring 12 Plan, and that's administered by the Department of 13 Justice. 14 National All Schedules Prescription Electronic Reporting 15 Act administered by HHS, the Department of Health and 16 Human Services. 17 create PDMPs or enhance existing ones. The second source is NASPERA, which is the And this program enables states to 18 And these programs were just started, they’re 19 very new, and they were started to make sure that there 20 was access to legitimate drugs that work, controlled 21 substances, and, at the same time, identify people that 22 have multiple prescriptions for the same substance, and PRECISE REPORTING, LLC jcp 118 1 to intervene and offer treatment for people who are 2 addicted to prescription drugs. 3 for finding drug abuse trends. 4 the issue around doctor shoppers. 5 are someone may go to different physicians to get the 6 same prescription multiple times for drug abuse 7 purposes. It also provides data It also began to address So, doctor shoppers 8 And FDA can use these data in a few different 9 ways to identify which drug substances and formulations 10 are targeted by doctor shoppers. The data though are 11 still very new, understand development, so, they 12 continue to evolve and change. 13 states have PDMPs. 14 we are still learning how to use this data. And, to data, not all So, that's still a challenge. And 15 So, in conclusion, as we think about all these 16 data sources and how we will use them, we are faced with 17 the challenges that new drugs and formulations that 18 address the issues associated with drug abuse are 19 currently being developed or are currently under review. 20 And when we’re looking at new ways to evaluate these 21 abuse-deterrent formulations, we are using current and 22 new data sources, which are just being developed, and PRECISE REPORTING, LLC jcp 119 1 how these data sources will be used to sustain a 2 labeling claim of abuse deterrence is something that 3 we’re working very hard on. 4 Thank you. 5 DR. KIRSCH: 6 We’re now going to go to the question and 7 answer session again. 8 by Dr. Wolfe. 9 Thank you. Our first question will be given Clarifying Questions 10 DR. WOLFE: I was out of the country in 11 September of 2009, when these committees met to decide 12 whether or not there was enough evidence of some 13 improvement with OxyContin in the formulation to go 14 forward. 15 yesterday just to see what happened, and there were a 16 few people, including Dr. Flick, Dr. Kirsch, who were 17 concerned about going ahead, and I’ll just quote this 18 because it’s a comment and I have a question afterwards. 19 Dr. Kirsch voted against this because he said it’s 20 “unconscionable to move forward without well-defined 21 REMS.” 22 the purposes of this meeting today and tomorrow is so And so, I read the transcript of this meeting I am mainly an optimist, and I think that one of PRECISE REPORTING, LLC jcp 120 1 that the next time one of these drugs comes forward, 2 there will be a REMS at the beginning rather than a year 3 later. 4 But I want to move back one step and ask 5 questions of really the speakers this morning. as you 6 look at the submitted ideas for epidemiological studies 7 by the company, FDA’s ideas, and so forth, are there not 8 some of these studies that could be done prior to 9 approval? I mean, when you’re sort of struggling for a 10 comparator group and knowing that OxyContin old is not 11 around anymore, is it not possible to--I mean, I am all 12 in favor of post-market surveillance. 13 and a lot of these databases need to be utilized. 14 so, my question is to any of these people who spoke this 15 morning. 16 It’s necessary, But, Can you see any studies that could be done 17 prior to approval or more definitively answer the 18 question of whether there is a reduction in abuse 19 potential. 20 Embeda before naltrexone was embedded in it. 21 have naltrexone, and the comparison would be useful. 22 That's really the question open to anyone who spoke. Again, OxyContin, but the same is true for PRECISE REPORTING, LLC It did not jcp 121 1 Any kinds of studies, whether it’s using these processes 2 phenotypes that Dr. Anthony described or any other 3 means? 4 done prior to approval so we’d have a better idea about 5 whether there appears to be a risk reduction, as not to 6 say that you still don’t need the post-marketing 7 studies. 8 9 Cross-sectional studies. DR. KIRSCH: Anything that could be Dr. Rappaport, would you like to take that question or appoint it to somebody? 10 (Laughter.) 11 DR. RAPPAPORT: Well, I can try to speak to 12 some of it, perhaps. 13 as you mentioned, if there's baseline work or comparator 14 work that could be done prior to approval, and then once 15 a drug gets on the market, we could do those 16 comparisons, but I think we’d have to be careful about 17 secular trends. 18 there's a lot of other things going on during this 19 timeframe other than just FDA’s actions. 20 to be very careful if those comparisons span a number of 21 years. 22 I think there may be components, I think Dr. Lapteva pointed out that DR. RAPPAPORT: So, we’d have I would just add, if you’re PRECISE REPORTING, LLC jcp 122 1 trying to measure whether there's a reduction in abuse 2 in the community based on the change in formulation, I 3 don’t see how that could be possible if a drug hasn’t 4 been marketed out in the community. 5 DR. WOLFE: I agree fully with that, I’m just 6 simply saying the question a year ago was: Is some 7 reason to think that the new formulation is better? 8 all I’m suggesting, that the community is essential, the 9 longitudinals are essential because people may get wise And 10 to some of these tamper-resistant methods and 11 everything, but prior to approval, not in the community 12 in some kinds of studies, whether they’re observational 13 studies or whatever, cross-sectional studies, randomized 14 trials between the two. 15 of these studies before, some of them? 16 substitute. 17 18 19 DR. HERTZ: Is it not possible to do more Not as a So, to do observational studies of a non-approved product, what exactly do you mean? DR. WOLFE: You literally could do a 20 randomized trial. I mean, prior to approval, you have 21 old OxyContin, it is around, still approved, you have 22 the new version, which has not yet been approved, but PRECISE REPORTING, LLC jcp 123 1 it’s obviously being subject to other kinds of studies 2 prior to approval. 3 to the post-market studies what the evidence is that it 4 really has some improvement over the old product. 5 To get some better idea in addition DR. HERTZ: So, you are envisioning a multi- 6 thousand patient study of new and old OxyContin, looking 7 for aberrant drug behavior? 8 9 10 11 12 DR. WOLFE: That would be one way of doing it. There might be other ways of doing it. the question. I’m just raising It just seems as we’re struggling-- DR. HERTZ: I mean, I think that it’s very easy to throw out a concept. 13 DR. WOLFE: Right. 14 DR. HERTZ: Without having thought it through 15 because the concept of trying to look at abuse 16 aberrant drug-taking behaviors in a clinical trial of 17 patients is extremely difficult because they don’t tell 18 you that their intent is to misuse or abuse. 19 typically, we get pretty low rates of this type of 20 behavior, particularly, I would imagine, if that's what 21 the intent of the study would be. 22 love to be able to come up with a design to look at this So, So, I think we would PRECISE REPORTING, LLC jcp 124 1 pre-approval, but if you have actual thoughts on how it 2 can be done, that would be helpful, other than to say 3 something, because that's where we’re here for. 4 DR. KIRSCH: Well, I think the question has 5 been asked and answered. 6 question. 7 Dr. Mendelson? 8 DR. MENDELSON: 9 We’ll go on to the next Yes, hi. Thank you. This comes to about three different people who presented. 10 But the question for the Prescription Drug Monitoring 11 Programs, a big question would be: 12 from the mail-in pharmacies like Merck-Medco? 13 Increasing numbers of patients fill their prescriptions 14 in three-month intervals through long-term pharmacies, 15 and I think this may be fueling some of the supply of 16 licit opioids that become illicit. 17 important for epidemiologists to begin capturing who 18 pays for the medications and how many dose units are 19 dispensed at a time. 20 Do they capture data It would seem to me Dr. Lapteva, your wonderful slide there that's 21 the scariest slide I’ve seen as a practicing doctor, 22 says that most of the diverted opiates are coming from PRECISE REPORTING, LLC jcp 125 1 single physicians prescribing to single patients. 2 that sort of suggests that a large number of dose units 3 are going out to those individuals, and that is probably 4 occurring because o the way they’re paid for. 5 gets to Dr. Bickel's point that economics are important, 6 but they may not be the economics we think. 7 So, my question would be: And So, this How would we 8 capture payment sources for medications and how those 9 relate to the amount of abuse-able drug there is, and 10 this is something the FDA could possibly take a stand on 11 to say that there is going to be a limit on the number 12 of dose units dispensed in a per unit of time per 13 patent. 14 will drive costs up as people will need to be seen more 15 often for management. 16 think. 17 far, it’s who’s paying for these drugs, for the most 18 part, and I think it’s insurance is paying for all the 19 abused, and, therefore, we are all paying for them, 20 everyone who pays a premium for an insurance plan is 21 paying for medication. 22 the epidemiologists as to how would they capture who’s This will not please the cost people because it But it’s an important point, I It’s not covered in any of the presentations so So, I’d like some comments from PRECISE REPORTING, LLC jcp 126 1 paying for the drugs and how would that be captured in 2 Prescription Data Monitoring Systems, as well, for out- 3 of-state prescriptions, which are probably not showing 4 up, at least on my prescription monitoring reports. 5 DR. STAFFA: Well, I can address that at the 6 population level, the drug utilization data that Dr. 7 Dormitzer talked about does capture mail-order 8 prescriptions, it captures all different sources, 9 whether it’s retail, and so, that type of use is being 10 captured if we use that type of denominator. 11 familiar with the Prescription Drug Monitoring Plans. 12 don't know if someone else can address that. 13 MR. PAULOZZI: I’m not as I can comment on that. I Some 14 Prescription Data Monitoring Programs in states do 15 capture mail-order prescriptions to their state 16 residence. 17 out of the state elsewhere. 18 incomplete. 19 in each camp. 20 it does need to be captured. 21 capturing it, of course, is to underestimate the total 22 number of prescriptions, the number of doctor shoppers They don’t necessarily capture them mailed Others do not. So, it’s I don't know what percentage of states fall I think that it’s an important point that The net effect of not PRECISE REPORTING, LLC jcp 127 1 if people getting multiple prescriptions through that 2 route. 3 months’ supply controlled substances can be prescribed, 4 but even single-month supplies could be ordered through 5 mail-order pharmacies. 6 7 Of course, there are limits in terms of how many If there’s somebody from the DEA, they might be able to comment more on that topic. 8 DR. KIRSCH: 9 DR. OMOIGUI: Dr. Omoigui? A couple of things. The first 10 question is to Dr. Paulozzi. There's a significant 11 difference in the percentage of injectable OxyContin 12 compared to injectable morphine. 13 because the injectable formulation of morphine is 14 commercially-available or has that been broken down into 15 what is being injected from commercial formulations and 16 what’s being injected by kitchen chemists with respect 17 to the morphine? 18 any commercially-injectable formulation. 19 first question. Is that a difference We know that OxyContin does not have That's the 20 And the second thing I wanted to point-- 21 DR. KIRSCH: 22 Let's take one question at a time, okay? PRECISE REPORTING, LLC jcp 128 1 MR. PAULOZZI: Yes. Thank you. That data 2 that was not broken down by the formulation of the kinds 3 of drugs. So, I don’t have really the answer to your 4 question. And it may be related to the injectable form 5 of morphine, but I don't know. 6 DR. KIRSCH: 7 DR. OMOIGUI: Okay. Your second question? The second question, there has 8 been some reference to a behavioral economics. As a 9 physician on the frontline, I can see that one of the 10 studies not being done here and which I believe would be 11 a leading indicator of the success of the new 12 reformulated OxyContin is what is the price point at 13 which OxyContin is being sold on the street? 14 Right now, one of the greatest problems we 15 have with OxyContin is 80 mg tablets, which is being 16 sold at $1 per mg, you’re getting like $80 per pill, and 17 maybe with a little bit of discount, the person doing a 18 diversion gets $50 a pill. 19 prescription of 90 tablets, you’re getting $50 a pill 20 for a month’s supply. 21 tax income of $100,000 a year. 22 Now, if you’re talking about You’re essentially making after So, I think if we’re going to analyze the PRECISE REPORTING, LLC jcp 129 1 success of this abuse-deterrent formulations, we also 2 need to know are the street prices dropping because that 3 would be an indicator of the desirability of this new 4 formulations. 5 interested in incorporating that into any of the 6 studies. 7 8 9 I don't know if anybody would be DR. KIRSCH: Would someone from the FDA like to address that or should I call on somebody? DR. RAPPAPORT: Let me just say, I mean, it’s 10 an interesting concept, but I think really that's part 11 of the discussion that we’re asking you all to have 12 tomorrow in response to the questions. 13 looking for right now is that you get clarification from 14 the various speakers. 15 DR. KIRSCH: 16 Dr. Flick? 17 DR. FLICK: What we’re Thank you. Dr. Woodward, I just want to 18 better understand some of the datasets here. 19 about in the TEDS dataset that most states do not 20 collect identifiers, which suggests that some states do 21 collect identifiers, and is there an opportunity in 22 those states to do longitudinal studies? PRECISE REPORTING, LLC You talked jcp 130 1 DR. WOODWARD: We've looked into that. It’s 2 difficult because the data isn’t really that well 3 collected. 4 of analysis, the confidentiality privacy issues that I 5 alluded to. 6 not really in that group, so, I can’t really respond as 7 fully as I might want to or you might need, but I know 8 the others in that group have looked into it, and it’s 9 been difficult logistically to get those kinds of We also face restrictions on doing that kind The states have their own regulations. I’m 10 studies going on. 11 quality is that good that it would be that useful. 12 know certain states do have good data. 13 their own analysis, done their own reporting on that. 14 We haven't been sure that the data DR. FLICK: I They’ve done I would think that the ability to 15 collect identifiers and the ability to do longitudinal 16 studies is crucial to having good data with regard to 17 this question, and I would think or I would hope that we 18 could answer that question clearly. 19 opportunity to do those studies and select the states, 20 that we avail ourselves of that. So, if there is an 21 I also wanted to ask with regard to DAWN, this 22 is a probability sample of short-stay federal hospitals. PRECISE REPORTING, LLC jcp 131 1 Does that include rural hospitals? 2 cutoff for that? 3 where I think a good proportion of this problem exists? 4 And do we include children’s hospitals in that dataset? 5 Is there a bed size Are we eliminating rural hospitals DR. WOODWARD: We include rural hospitals. 6 Obviously, they’re not going to show up as part of the 7 metro area over-sampling; they’re going to be part of 8 the remainder sample. 9 versus non-rural, but EDs, the sample is strong enough We can do some reporting on rural 10 for that. We would only include children’s hospitals if 11 they have an emergency department that's open 24 hours, 12 7 days a week. 13 short-term general hospitals, so, I can’t think of any 14 in the 200, 250 hospitals that would be children’s 15 specialty hospitals. 16 DR. FLICK: But, generally, I mean, we only use All right. I would think that is 17 going to be a problem from an epidemiologic standpoint 18 if the age that we’re considering here is 12 and over, 19 especially in metropolitan areas, where a lot of the 20 pediatric emergency care takes place in a children’s 21 hospital. 22 patients. You will miss all of those or many of those PRECISE REPORTING, LLC jcp 132 1 DR. WOODWARD: Well, the DAWN does pick up 2 under 12; it picks up all age groups. 3 that's 12 and over. 4 DR. FLICK: It’s the NSDUH If it doesn't sample, then they 5 won't be represented or they won't be accurately 6 represented. 7 hospitals, then you’re sampling metropolitan areas. 8 They’ll be missed. 9 If you’re not sampling children’s DR. WOODWARD: No, it's a limitation. I 10 agree. 11 like to sample especially psychiatric facilities, too, 12 which that would be more consistent with SAMHSA’s 13 missions. 14 15 It’s also expensive to add to--I mean, we’d also DR. FLICK: I certainly understand the limitations. 16 DR. WOODWARD: 17 DR. FLICK: 18 DR. WOODWARD: 19 DR. FLICK: Yes. And I’m not being accusatory here. Yes. I’m just saying if we’re going to 20 plan studies, we need to understand what the datasets 21 contain and what they more importantly don’t contain. 22 If I could just ask one more question of you. PRECISE REPORTING, LLC jcp 133 1 DR. WOODWARD: 2 DR. FLICK: Sure Do any of these datasets 3 communicate with one another or can the data from one be 4 merged with another to gain a more accurate picture? 5 6 DR. WOODWARD: elaborate a little more on your prior question? 7 DR. FLICK: 8 DR. WOODWARD: 9 Could I, if it's okay, Sure. One of the things that we have been looking into, the staff has been looking into is 10 working with other ED databases. 11 and also ARHQ has the SIDS and SAIDS part of the HCUP, 12 the Hospital Cost Utilization Project. 13 largely focused on general hospitals, but they do get 14 into specialty hospitals more than we do. 15 trying to supplement. 16 DR. FLICK: Right. NCHS has the NHAMCS Again, they’re So, we’re The Age CUP database 17 contains an additional sub-database called the KID 18 dataset, which, if I’m not mistaken, has identifiers and 19 can accurately represent pediatric care. 20 be an opportunity to utilize that dataset to better form 21 a picture of pediatric care. 22 DR. WOODWARD: So, it might Yes, we're starting to talk PRECISE REPORTING, LLC jcp 134 1 with AHRQ, and we’ve started to talk to NCHS. 2 3 DR. KIRSCH: We're going to go to the next presentation. 4 DR. WOODWARD: 5 DR. KIRSCH: Okay. For a point of clarification, Dr. 6 Morrato is shaking her head no, and so, I’ll give one 7 second. 8 9 DR. MORRATO: KID. Yes, I've done research with KID is admissions, it’s the identified. If you’re 10 really interested in pediatric, there’s a semi- 11 proprietary PHIS dataset which would allow you to get to 12 that. 13 14 DR. KIRSCH: So, our next speaker is Dr. Kornegay. 15 Study Designs to Assess Prescription Drug Use 16 Design Considerations in Epidemiological Studies of 17 Abuse-Deterrent Opioids 18 DR. KORNEGAY: Good morning. My name is 19 Cynthia Kornegay, and I’m an epidemiologist in the 20 Office of Surveillance and Epidemiology at FDA. 21 going to spend the next several minutes discussing some 22 general design considerations related to epidemiological PRECISE REPORTING, LLC I’m jcp 135 1 studies of opioid abuse. 2 First, I will outline the purpose of this 3 talk. I am also going to briefly review the definitions 4 of abuse and misuse. 5 background information, and, finally, discuss 6 comparators and issues related to measuring change in 7 abuse-related outcomes in studies. 8 summarizing the concerns highlighted in my presentation. 9 The purpose of this presentation is to provide Then I will provide some I will conclude by 10 general comments based on preliminary proposals 11 submitted by the Sponsors and to provide a framework for 12 considering the industry presentations later this 13 afternoon and tomorrow. 14 study design, while the next talk will address some of 15 the statistical considerations. 16 not a detailed critique of the specific proposals, but 17 general comments based on the submitted documents. 18 My talk will relate mostly to This presentation is To review, abuse is defined as the non-medical 19 use of a drug repeatedly or sporadically for the 20 positive psychoactive effects it produces, while misuse 21 is defined as the use of a drug outside label 22 directions or in a way other than prescribed or directed PRECISE REPORTING, LLC jcp 136 1 by a healthcare practitioner. 2 For abuse, the intent is non-therapeutic, that is to get 3 high, while the intent is still therapeutic for misuse. 4 These definitions are independent of anything that may 5 be done to the drug, for example, crushing, dissolving, 6 chewing, et cetera, to achieve the desired effect. 7 The difference is intent. The proposals that you will hear about will 8 use differing approaches, measuring a particular 9 population versus studying several different populations 10 and measuring the severe end of the abuse spectrum 11 versus including occasional or recreational use. 12 proposals attempt to measure change, but that raised the 13 question of change from what? 14 Both Prior to measuring change, the baseline abuse 15 potential for a product would need to be established. 16 This would also be important in targeting studies and 17 interventions to where they will have the largest 18 impact. 19 is at greater risk for abusing a particular product, 20 including demographic, social, economic, and geographic 21 considerations in defining at-risk populations may help 22 define the limits of abuse deterrence efforts. One of the basic considerations is defining who PRECISE REPORTING, LLC In jcp 137 1 addition, other risk factors such as family history and 2 certain psychological conditions may also be important 3 in determining the at-risk population. 4 Along with who at-risk individuals are, how 5 they are manipulating the product or not is also 6 important. 7 deterrent properties designed to affect the typical 8 routes of abuse? 9 outcomes can be measured, the baseline abuse level In other words, are a product’s abuse- Before a change in abuse-related 10 population and risk factors must be known. 11 information can be determined from historical data on 12 the same or similar products. 13 derived from real world, that is post-approval, product 14 use. 15 This This baseline should be The next few slides will discuss the important 16 issue of comparators. The appropriate comparator is 17 vital to providing the big picture or context for 18 changes in abuse-related outcomes. 19 different choices for meaningful comparators, but, 20 usually, multiple comparators are required. 21 comparisons can be historic by class, drug schedule, 22 and/or individual product. There are several Multiple Specific attributes of PRECISE REPORTING, LLC jcp 138 1 comparators will be discussed in more detail shortly. 2 Although multiple comparisons on many levels are 3 necessary to understand the big picture. 4 confusing, and conversely make it more difficult to 5 understand what is going on. They can be 6 In some cases, it will be possible to compare 7 non-abuse-deterrent and abuse-deterrent formulations of 8 the same product from either a historic or concurrent 9 perspective. This could be a particular problem, 10 however, for novel drugs introduced with abuse-deterrent 11 formulations. 12 how will it be possible to determine if abuse-deterrent 13 properties are effective? Once an abuse baseline is established, 14 Although these challenges are familiar, when 15 deciding on what comparators to use, this is a partial 16 list of some of the attributes that should be 17 considered. 18 population, indication, active ingredient, time-released 19 formulation, single ingredient or combination, and time 20 on market. 21 of administration, and the scheduling. 22 will provide some thoughts on interpreting a change in Ideally, comparators should be similar in One should also consider the strength, route PRECISE REPORTING, LLC The next section jcp 139 1 abuse. 2 Although none of these questions have easy 3 answers, some are more abstract and may require a 4 paradigm shift to address effectively. 5 question is what is a significant change and how can 6 that be defined? 7 of change are most meaningful from a regulatory 8 perspective, a change in the common methods of abuse or 9 an overall reduction as an example? The most obvious Close on heels of that is what types Other questions are 10 over what time period should the change occur and in 11 what population? 12 A second consideration in measuring change is 13 how time will be incorporated. 14 to assume that the abuse profile of a drug would be 15 static once established. 16 method of abuse are both subject to change. 17 It would be unrealistic The abusing population and the To address this, should updates be triggered 18 at some pre-defined point or be scheduled on a product- 19 by-product basis? 20 updates be? 21 to detect a shift in the baseline assessment or in abuse 22 levels? In addition, how comprehensive should That is, what level of detail is necessary PRECISE REPORTING, LLC jcp 140 1 A third issue is a population that is studied. 2 How should high-risk populations be weighted in a 3 baseline abuse assessment? 4 populations always be included, and what about the 5 general population? 6 require careful consideration of when data are available 7 relative to when they are collected and of unrelated 8 events that any affect baseline abuse levels. 9 also require the ability to modify or augment currently 10 Should traditional high-risk Addressing these concerns will It may existing data resources. 11 In addition to prior population 12 considerations, it is not clear how results from 13 numerator-only sources should be weighed in the national 14 context. 15 non-epidemiologic studies? 16 improving or refining the abuse profile and safety 17 issues of a particular product? 18 Also, what is the value of small, focused, How can they contribute to In an effort to increase the quality and scope 19 of data available to assess abuse-related outcomes. 20 Several non-traditional data resources are under 21 consideration. 22 as Kaiser, convenient samples or surveys, and add-ons to These include health care systems such PRECISE REPORTING, LLC jcp 141 1 population-based surveys and questionnaires. 2 these data sources can provide detailed information, 3 data resource validation, as well as sample size and 4 power considerations will still need to be identified 5 and appropriately addressed. 6 to determine the relative importance of study results. 7 Should they be limited to reporting findings from more 8 commonly-used data resources or are they robust enough 9 to be viewed as an independent study? 10 While It will also be necessary A final consideration is somewhat less 11 concrete, but still carries important implications for 12 determining the impact of abuse-deterrence measures. 13 Most studies implicitly assume that less hardcore users 14 means less casual users, but it is not always clear that 15 that is a simple relationship, and, from a public health 16 perspective, is a most effective abuse reduction 17 strategy when the target’s new initiates, hardcore 18 users, or individuals who are transitioning from one 19 extreme to the other? 20 To summarize, this presentation presented 21 several ideas to keep in mind while listening to the 22 upcoming presentations. Has a good baseline picture of PRECISE REPORTING, LLC jcp 142 1 the at-risk population mechanism been established prior 2 to measuring the effect of abuse-deterrent mechanisms? 3 Are the abuse-deterrent mechanisms designed to address 4 major or problematic routes or methods of abuse? 5 If multiple studies are proposed, will they be 6 completed in sequence or concurrently? How much 7 influence should numerator or sub-national studies have 8 on the overall abuse profile? 9 comparators been selected? Have the appropriate Have timing issues, the time 10 to create a baseline abuse profile, the time to 11 determine appropriate study length, and decide to 12 include time-dependent outcomes been considered and 13 incorporated into the study design? 14 resources or collection methods are proposed, how will 15 they be validated? 16 If novel data And finally, there were several population- 17 related questions. Is the proposed population similar 18 to the at-risk population? 19 and geographic issues been appropriately incorporated? 20 Should the study target high-risk individuals? 21 hardcore users, recreational users, or both be included? 22 Also, what are the implications of performing the study Have demographic, co-morbid, PRECISE REPORTING, LLC Should jcp 143 1 in a non-traditional data resource? 2 Thank you. 3 DR. KIRSCH: 4 Thank you. The next speaker is Dr. Keeton. 5 Statistical Considerations for Epidemiological Studies 6 of Abuse-Deterrent Formulations 7 MS. KEETON: Good morning. My name is 8 Stephine Keeton, and I am a statistical safety reviewer 9 in the Office of Biostatistics at FDA. Today, I will 10 present the statistical considerations for 11 epidemiological studies of abuse-deterrent formulations 12 of opioids. 13 The purpose of this talk is to highlight and 14 discuss generals statistical issues based on the 15 preliminary study proposals submitted by the Sponsors. 16 I will not provide a detailed critique of the 17 proposed statistical analyses, but, instead, will raise 18 general issues for consideration. 19 conjunction with Drs. Dormitzer and Kornegay’s talks 20 that you’ve just heard. 21 22 This talk is in Here is an outline of the presentation. I will start by briefly discussing the trend and Cross- PRECISE REPORTING, LLC jcp 144 1 Sectional Approaches described in the proposals. 2 then discuss some general modeling issues, followed by 3 issues related to data sources. 4 sample size and power considerations. 5 will discuss multiplicity and replication issues before 6 summarizing the statistical considerations. 7 I will I will also discuss And finally, I The overall goal of the Epidemiological 8 Program proposals is to show a reduction in the rates of 9 death, overdose, or abuse by comparing drug products and 10 formulations. 11 Trend Approach. 12 the OxyContin Program. 13 rates before and after the introduction of a new 14 formulation. 15 rates across time after introduction of a new 16 formulation. 17 One method to compare the rates is the This is the primary approach used in The Trend Appraoch compares This approach can also be used to compare The graphic on the left-hand side illustrates 18 the approach of comparing rates before and after 19 introduction of a new formulation. 20 represents time and the Y-axis represents the rate of 21 the outcome. 22 for a considerable time such as OxyContin, the trend The X-axis For a product that has been on the market PRECISE REPORTING, LLC jcp 145 1 approach compares the rates before and after 2 introduction of the new formulation. 3 represents the time of introduction of the new 4 formulation. 5 First, a change in the mean level of the rates before 6 and after introduction of the new formulation, and 7 second, a change in the slope of the rates before and 8 after introduction. 9 happens to rates for some period of time before and 10 The red dash line Two effects on the rate are considered. It is important to observe what after introduction of a new formulation. 11 The graphic on the right side illustrates the 12 Trend Approach for other opioid products. 13 the rates within a population over time may be 14 occurring. 15 factors such as the introduction of risk, evaluation, 16 and mitigation strategies or changes in cultural 17 behavior related to abuse. 18 such factors should be accounted for in the analysis. 19 Changes in These changes may be influenced by other These changes in rates from Another approach proposed is the Cross- 20 Sectional Approach. This is the primary approach used 21 in the Embeda Program. 22 compares rates for a new formulation to other products The Cross-Sectional Approach PRECISE REPORTING, LLC jcp 146 1 at a specific point in time or over a short period of 2 time. 3 Sectional Approach. 4 The graphics on this page illustrate the Cross- Again, the X-axis represents time, and the Y- 5 axis represents the rates of the outcome. Time 1 on the 6 X-axis denotes some period of time in which a comparison 7 between drug A and other drugs is made. 8 can also be made at an additional time point, time 2 to 9 evaluate whether the effect remains constant. The comparison 10 In this slide, I will talk about some 11 considerations of the trend and Cross-Sectional 12 Approaches. 13 graphics and analytic summaries. 14 is important to control for changes over time and 15 secular factors related to the outcome of interest. 16 Trend Approach can also be used to compare rates across 17 time after introduction of a product. 18 The Trend Appraoch offers intuitive In this approach, it The For example, for Embeda, the Trend Approach 19 could be used to look at the effect of the drug over 20 time after introduction of the new formulation. 21 absence of randomized treatment assignment, other 22 factors may confound association between the product and PRECISE REPORTING, LLC In jcp 147 1 the rates. 2 snapshot of abuse in time. 3 time may not be constant and must be considered. 4 in the absence of random treatment assignment, other 5 factors may confound association between a product and 6 the rates. 7 The Cross-Sectional Approach provides a However, the effect over Again, In the next couple of slides, I will discuss 8 some issues related to modeling of outcomes, death, 9 overdose, and abuse. The analysis should control for 10 potential confounders such as possible patient selection 11 biases. 12 abuse-deterrent formulation to patients suspected to 13 abuse opioids, and this may result in higher rates of 14 abuse. 15 across time. 16 and propensity score methods may be used to adjust for 17 differences in patient characteristics across 18 formulations and time. 19 Physicians may preferentially prescribe an Additionally, patient characteristics may differ Techniques such as multivariate analysis When considering the model for these studies, 20 it is important to note whether drug availability is 21 included in the model as a covariate or not. 22 example, in one of the Embeda studies, several models PRECISE REPORTING, LLC For jcp 148 1 are proposed to compare rates. The primary model does 2 not adjust for drug availability. 3 drug availability, the model provides estimates of the 4 proportion of abuse by unlabeled routes of 5 administration among all users. 6 does address for drug availability. 7 rates of abuse by any route of administration among all 8 users. 9 should be considered when selecting a model. Without adjusting for The secondary model The model provides 10 The difference in interpretation of the models I will now discuss some issues related to data 11 sources. 12 population of interest, for example, abusers versus all 13 users, and enriched population, such as abusers from 14 substance abuse treatment centers, may be easier to 15 study since it will provide more events of interest. 16 will discuss this topic more later when I discuss power 17 in sample size issues. 18 The data source should represent the Some of the studies propose, such as studies 19 conducted from data using data from substance abuse 20 treatment centers and special populations are based on 21 convenient samples. 22 interpreting results from these studies since Caution must be taken when PRECISE REPORTING, LLC I jcp 149 1 generalizing the results to a population that has 2 clinical relevance may be difficult. 3 these studies, the results of such studies, one should 4 characterize how the study subjects differ from the 5 population of interest, in particular, pay attention to 6 who might be left out or underrepresented in the data. 7 When interpreting The appropriateness and accuracy of 8 denominator information should also be considered to 9 ensure meaningful interpretation of the data. 10 Statistical surveys provide denominator estimates. 11 Well-designed and conducted surveys provide valid 12 denominators. 13 internal denominators. 14 represent the population of interest. 15 provide only numerator information, for example, poison 16 control data, the denominator must come from other 17 sources. 18 accuracy of the denominator source should be considered. Administrative claims data provide However, these databases may not For studies that 19 For such studies, the appropriateness and Power and sample size determination are 20 necessary if the study is to be used for statistical 21 inference. 22 conducting the study. The sample size impacts the feasibility of The statistical power and sample PRECISE REPORTING, LLC jcp 150 1 size depend on the event rate which I will discuss in 2 more detail in the next slide. 3 size and length of time period for trend analyses. 4 stated in Dr. Kornegay’s talk, defining a significant 5 reduction is a critical component of designing studies 6 to measure change. 7 clinical relevance. 8 wants to detect, the larger the sample size required. 9 For trend analysis, the length and number of time As well as the effect As The required effect size is based on The smaller the effect size one 10 periods must be taken into consideration. 11 the time period, the more information, and, therefore, 12 more likely it is to precisely identify patterns of 13 change. 14 The longer Statistical power depends highly on the number 15 of events captured in a study. 16 greater statistical power. 17 population such as studies of patients entering 18 substance abuse treatment centers will likely offer a 19 relatively high number of events, and, thus, easier to 20 study. 21 more frequent outcomes are easier to study. 22 More events provide Studies of enriched The choice of outcome also impacts power, as The duration of the study also impacts power PRECISE REPORTING, LLC jcp 151 1 in that the longer the duration, the more events 2 observed. 3 comparative period needs to be considered. 4 events provide greater statistical power. The event rate of the comparator drug or 5 Again, more Finally, product availability impacts a number 6 of events. 7 use, it may take a considerable amount of time to get 8 events. 9 size are incorrect, the study may be underpowered and 10 11 If a product is not readily available to If any assumptions used to calculate the sample not capable of demonstrating an effect. The various studies proposed addressed 12 different populations and questions, but may provide 13 overall information on abuse. 14 data may be required to replicate the results. 15 abuse deterrence over time raises statistical issues. 16 As stated in Dr. Kornegay’s talk, the abuse profile of 17 the new formulation may change after some period of time 18 of being on the market. 19 of the same data sources and adjustments for multiple 20 testing may be required. 21 22 However, other sources of Testing This may require multiple uses In summary, the Sponsors have proposed two general approaches to measure the abuse-deterrent effect PRECISE REPORTING, LLC jcp 152 1 of new formations. For both the Trend and Cross- 2 Sectional Approaches, there are several important 3 statistical issues to consider when reviewing the abuse- 4 deterrent studies. 5 potential confounders, such as possible patient 6 selection biases and consider methods to adjust for 7 differences in patient characteristics across 8 formulations and time. 9 to the population of interest should be considered, as The analysis should control for The relevance of the data source 10 some studies will be conducted using data based on 11 enriched populations or convenient samples. 12 appropriateness and accuracy of Denominator information 13 may have an impact on study quality. 14 The The statistical power and sample size depend 15 on the choice of outcome population, duration, and drug 16 availability. 17 feasibility. 18 These factors does affect the study And lastly, the abuse profile of the new 19 formulation may change after some period of time of 20 being on the market. 21 over time should be accounted for in the statistical 22 inference. Multiple uses of the same data PRECISE REPORTING, LLC jcp 153 1 Thank you for your attention. 2 DR. KIRSCH: Thank you. Before we break for 3 lunch, I want to ask for any of the morning speakers 4 whether anybody who spoke this morning is not going to 5 be here this afternoon. 6 further questions to this afternoon, but if any of the 7 speakers from this morning plans to leave at the 8 lunchtime and not come back, I’d like to direct any 9 questions. There’s one. My preference is to hold the So, we have one speaker who’s 10 not coming back, and that's the person from the DAWN 11 Group. 12 committee has for that individual speaker before we 13 break for lunch? So, are there any questions any member of the 14 Seeing no hands, I’ll assume that there will 15 be no questions for that individual, and we will break 16 for lunch now. 17 for lunch. 18 room at 12:55. 19 with you at that time. 20 It’s currently 11:55. I will now break We will reconvene again in one hour in this Please take your belongings you may want Committee members, please remember that there 21 should be no discussion of the meeting during lunch 22 amongst yourselves, with the press, or any member of the PRECISE REPORTING, LLC jcp 154 1 2 3 audience. Thank you. (Whereupon, at 11: a.m., a luncheon recess was taken.) 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 A F T E R N O O N S E S S I O N PRECISE REPORTING, LLC jcp 155 1 (12:55 PM) 2 3 DR. KIRSCH: seats. I’d like everybody to take their 4 We’re going to reconvene the meeting. The first session of this afternoon is to 5 address additional clarifying questions. I want to 6 remind the members of the committee that really the 7 purpose that we’re here for for these two days is really 8 to help FDA come up with relevant studies to encourage 9 the companies to use in order to look at the outcomes 10 that we think are important. 11 necessarily criticize one company or another; it’s to 12 really help the FDA come up with studies, which I find 13 it to be a very exciting opportunity. 14 It’s really not to So, in that context, we’re going to go back to 15 our question list, which is--I think, actually, I cut 16 off Dr. Flick when he was talking before. 17 make sure he asked all of his questions. 18 I want to Clarifying Questions 19 DR. FLICK: I did have one other question, and 20 the question that wasn’t answered was: Do we have the 21 capacity to merge information from one dataset to 22 another? And I don't know which one of the presenters PRECISE REPORTING, LLC jcp 156 1 can address that question. Which of any of the datasets 2 can merge their data to make a more robust dataset? 3 MR. PAULOZZI: 4 answer to the question. 5 identifiers. 6 coroner information, or vital statistics information for 7 deaths, and that information has been, can be merged 8 with-- 9 I guess I can try a partial You need, of course, personal So, that would mean medical examiner, DR. FLICK: No, I think we're maybe talking 10 about two different things. 11 before, the National Hospital Discharge Survey, for 12 example, uses the same methodology as the National 13 Survey of Ambulatory Surgery. 14 the sampling universe can be merged to make an accurate 15 representation of surgical procedures in the U.S., for 16 example. 17 things that we’re-- 18 19 So, the methodology and So, I think those are a little bit different MR. PAULOZZI: Yes, yes, and I don't have an answer to that question. 20 DR. FLICK: 21 DR. KIRSCH: 22 For example, as I’ve used Okay. Okay, and then Dr. Morrato had a question that she was going to ask during the break and PRECISE REPORTING, LLC jcp 157 1 I cut her off because it wasn’t in the public forum, so, 2 I asked her to ask her question in the public forum. 3 DR. MORRATO: Okay. It was a question related 4 to CDC, and maybe a little bit of what we’ve discussed 5 before in terms of what can be done before a drug is 6 approved. 7 they’ve taken approach of system dynamics modeling. 8 They’ve used this, for instance, to model national 9 cocaine prevalence, and, most recently, there was a I know in other areas of public health, 10 special issue July of the American Journal on Public 11 Health, where they were talking about this is a way of 12 looking at tobacco control. 13 And, so, my question was whether or not anyone 14 has done sort of this theoretical framework that looks 15 at sort of the abuse of prescription opioids as a 16 starting point. 17 interventions, you might be able to predict which might 18 work better than others. So, as we're looking at these different 19 So, for instance, so, it’s a system dynamics 20 modeling, it allows you to model multiple interactions 21 of diseases, risk, delivery systems, populations that 22 we’ve been talking about, and how it relates to national PRECISE REPORTING, LLC jcp 158 1 and state policy. 2 but then it gives guidance for what might be 3 interventions and/or measurements, et cetera. 4 don't know the field well enough whether or not CDC 5 folks are aware of it or anyone. 6 So, it’s more of a modeling exercise, MR. PAULOZZI: And I I'm sorry; I'm not familiar 7 with that modeling approach. I would think though it 8 depended on the availability of multiple variables about 9 individuals involved, and it seems like the constant 10 problem in this is that you don’t have a lot of in-depth 11 information about individuals using the drugs. 12 DR. MORRATO: Yes, so, it's more of a 13 conceptual framework where it’s social sciences create 14 the model, and then you put information in it as you 15 acquire it, but at least you’re starting with a 16 framework. 17 MR. PAULOZZI: 18 DR. ANTHONY: Yes. The RAND Group probably launched 19 the best starting points for those systems analysis 20 models having to do with cocaine and other drugs. 21 best work that's been done since then is probably 22 Jonathan Caulkins’ work. He works with a group in PRECISE REPORTING, LLC The jcp 159 1 Vienna, as well. Jonathan is a Carnegie Mellon. 2 And then the elaboration with some more 3 sophisticated statistical models is being done by 4 Georgiy Bobashev at Research Triangle Institute. 5 there are developments, but it’s pretty cartoon-like. 6 You can make some predictions. 7 assumptions is difficult, but I think it’s a very good 8 suggestion and a line to work. 9 good work on that. So, Validating the They’ve done some pretty 10 DR. KIRSCH: Okay. Dr. Bickel? 11 DR. BICKEL: I have two questions. I’ll ask 12 them one at a time. And this one is for everybody who 13 has different datasets. 14 strong monotonic association with social economic status 15 that is whether you measure it by educational attainment 16 income or occupational status. 17 education, the lower of your income, your lower 18 occupational status, the more likely you are to use 19 drugs and to be addicted. 20 different datasets to model the fact that that is the 21 case and use that to make inferences about who is 22 susceptible? So, given that addiction has a The lower your Has anyone used these PRECISE REPORTING, LLC jcp 160 1 DR. ANTHONY: I'm sorry. I have to start by 2 challenging your assumption. It depends really heavily 3 on stage of introduction of the drug and the population, 4 and earlier in the stages, there’s often an excess among 5 high SES individuals, and in later stages, things 6 straighten out. 7 excellent example of that in the United States right 8 now. 9 who’s published on this topic in relation to cocaine Tobacco smoking epidemic is an cocaine, there’s someone sitting in this room 10 using the National Surveys on Drug Use and Health and 11 the national household surveys. 12 done, but I don’t think I’d have as a starting point the 13 assumption that you started with. 14 DR. BICKEL: So, yes, work has been Okay. Fair enough. My second point, 15 and, Jim, you started to address this, right? 16 are the early adopters, and should we have a model of 17 them and see whether they’re predictive of later use 18 patterns so that we could actually use them as like a 19 canary in the coal mine? 20 DR. ANTHONY: So, who In my slide on future 21 directions, I planted a little seed, and I’m happy that 22 you’ve given me a chance to see if it can blossom. PRECISE REPORTING, LLC If jcp 161 1 the regulations allow limited release to subpopulations, 2 there would be some groups that would be useful to try a 3 limited release form of experiment, and those would be 4 hospitals where you know quite a bit of opioid diversion 5 is going on with existing established products. 6 under fairly carefully controlled conditions, you could 7 release the new product and see what happens in a 8 population where you know there are hospital personnel 9 who will figure out ways to divert new products. 10 And So, I think I’d encourage FDA to think about 11 this as a way of staging the introduction of the product 12 in different subpopulations that might vary in their 13 vulnerability. 14 when Dr. Wolfe and I didn’t have any gray hair, I 15 recommended FDA randomly assigning regulatory regimes to 16 different states when new products were introduced. 17 pretty sure the Boards of Pharmacy and the like would be 18 open to this, particularly if it would lead us to better 19 evidence about the effect of different regulations and 20 the impact of different compounds. A long time ago, and this is probably 21 DR. KIRSCH: Dr. Denisco? 22 MR. DENISCO: Thank you. I’m A couple of comments PRECISE REPORTING, LLC jcp 162 1 along with a question to possibly Dr. Paulozzi and 2 anybody else. 3 One is we looked at populations. We briefly 4 discussed subpopulations that we needed to pay attention 5 to in pediatric populations were mentioned. 6 population that I think needs to be mentioned is the 7 prisoner population. 8 population is incarcerated, and especially with the 9 NSDUH study, that doesn’t sound like 1 percent would 10 affect the statistics very much, but a little back of 11 the napkin calculation, when you look at the high 12 prevalence rate of addiction in that population, it can 13 really throw things. 14 any way better than the back of the napkin to sort of 15 address that. We have almost 1 percent of the So, I was wondering if there was That was one thing. 16 The second thing is -- 17 DR. KIRSCH: 18 first question first. 19 Paulozzi? 20 Another Well, let's hold off for the Was the question pointed at Dr. MR. PAULOZZI: So, you're asking whether or 21 not the surveys can be affected by not including 22 incarcerated populations and how that can be affected? PRECISE REPORTING, LLC jcp 163 1 Yes, I can see your point, 1 percent with a 2 tenfold great prevalence of addiction might really make 3 a difference in the statistics. 4 though in terms of how to address that particular 5 problem since that would be an issue really for the 6 SAMHSA people to try to address how to involve that 7 population. 8 perpetrators who were arrested, ADAMS System, which gets 9 at drug use and screens people at intake after arrest, I don’t have an answer They do have surveys of drug use of 10 which gives you some sense of drug prevalence or use of 11 drugs in that population. 12 DR. KIRSCH: Dr. Anthony? 13 DR. ANTHONY: 14 get to make many sins. 15 proposing that we should do. 16 Catchment Area studies in the early 1980s where we had 17 coordinated samples of household, institutional, and 18 homeless individuals, and institutions included prisons 19 and jails and nursing homes and the like, mental 20 institutions. 21 the estimates, although, of course, there are 22 concentrations of the cases where you would predict. So, if you live long enough, you We actually did what you’re In the NMIH Epidemiologic We really didn’t find much variation in PRECISE REPORTING, LLC jcp 164 1 But it turns out the relative proportions in those 2 subpopulations are small enough that they don’t really 3 perturb the estimates when you take the confidence 4 intervals of the estimates into account. 5 single point estimate is not going to be all that 6 helpful, but if you put a confidence interval around it 7 and then ask how often are you moving the estimate 8 around outside of the confidence interval, it didn’t 9 really happen very much. 10 Because any Subsequently, I don’t believe any of the 11 investments, and we’re talking about more than $1 12 billion in the early 1980s in these surveys. 13 think any of the subsequent studies have included 14 coordinated household, non-household populations as we 15 did there. 16 all that much difference, but, as Len pointed out, there 17 are special studies of prisoners and special studies of 18 the homeless and so on. 19 20 21 22 I don’t Perhaps because we didn’t find that it made DR. KIRSCH: Okay. Go to your second question. MR. DENISCO: Yes, the second question is: Since the prescription drug problem is closely PRECISE REPORTING, LLC jcp 165 1 intertwined with the pain medicine treatment and under 2 treatment of pain. 3 unintended consequences, and I would do an almost 4 proposed. 5 that any time these issues of prescription drugs are 6 studied, there should be some attention paid to what it 7 does to the availability of legal prescriptions to 8 appropriately screen patients. 9 10 Certainly, there's going to be It’s not really a question, more a comment, DR. KIRSCH: I would bet that our open public hearing tomorrow, we’ll hear a lot about that. 11 Dr. Bilker? 12 DR. BILKER: Yes, I was wondering whether it 13 would be important and if any of the databases are 14 capable of assessing abuse by new or previously naïve 15 abusers of illicit drug versus continuing abuses. 16 possible, that would allow estimation of barriers to 17 abuse as a deterrent to initiation of new abusers versus 18 continuing users. If 19 I had one more question after that. 20 DR. KIRSCH: Dr. Anthony, did you want to-- 21 DR. ANTHONY: You’ll be happy to know it’s 22 been done. So, within the limits of the assessment, you PRECISE REPORTING, LLC jcp 166 1 can restrict your estimates to people who have used no 2 other drug before the one of interest and people who’ve 3 started with tobacco or alcohol and people who have used 4 cocaine and so on. 5 and the context of new products is a new one, and it 6 hasn’t been done there because usually these surveys are 7 contracted out in five-year increments. 8 plans are laid out fairly well in advance, and the 9 capacity of the survey team to move quickly when a new So, that routinely is being done, 10 product is introduced is slow. 11 it’s slow and expensive. 12 The assessment I mean, it’s there, but Now, if you had asked me is it worth it to 13 make the expense of putting a piggyback assessment on a 14 standardized assessment in order to learn something of 15 value for the public’s health, I’d say in many cases 16 we’re going to find that it’s worth the cost. 17 hasn’t been done yet. 18 DR. BILKER: Not to my knowledge. Are any of the databases that 19 we’re talking about capable of doing this? 20 assessment? 21 22 DR. ANTHONY: But that Making this In the existing data, you have a trace for each individual of the age of onset of each PRECISE REPORTING, LLC jcp 167 1 drug, and you also, for people who’ve used and started 2 to use in the past 24 months or so, you have month of 3 first use. 4 grained, temporal sequences for each individual, and 5 that would allow you to study these variations that 6 you’re pointing toward. So, you can actually get some pretty fine- 7 DR. BILKER: Okay. Great. 8 DR. KIRSCH: Did you have a second question? 9 DR. BILKER: Yes. Just one point I wanted to 10 make about the Trend Approach, in looking at the trends 11 over time. 12 introduction of the drug to consider nonlinear effects. 13 You’re looking at linear effects, but there might be 14 bumps as people figure out how to get the drug out. 15 You’ll see bumps in the road. 16 not to just look at linear effects. It might be very important early on after 17 DR. KIRSCH: 18 Dr. Wolfe? 19 DR. WOLFE: So, it’ll be important Thank you. This is for my gray-haired 20 colleague, Dr. Anthony, which was I’d just like to hear 21 a little bit more about your interesting concept of 22 processed phenotypes both in terms of how that could be PRECISE REPORTING, LLC jcp 168 1 incorporated into post-approval surveillance or even 2 arguably into pre-approval looks. 3 are using amongst others as subjects for studies people 4 who are beyond that, who are clearly addicted or at the 5 other end of the spectrum who are new, but here’s the 6 halfway, and it would seem that this is a sensitive 7 group to be able to assess whether or not X is going to 8 more rapidly move them towards the other end of the 9 spectrum. I mean, we obviously 10 Could you just comment on that, please? 11 DR. ANTHONY: Thanks for the opportunity. So, 12 let’s think about the patient who’s given one of the 13 opioids that has a relatively high side effect profile 14 for a dysphoric response. 15 reasons or other reasons is generating dysphoria. 16 So, for pharmacokinetic In that patient for that product, you wouldn’t 17 expect more than one pill or so to disappear from the 18 dispensed container. 19 you might find a population that's very low susceptible 20 to the repeated use of the drug that would lead to a 21 dependent syndrome or subsequent problems. 22 you’ve got people who are given prescriptions say for And by doing a pill count study, PRECISE REPORTING, LLC And then jcp 169 1 dental surgery, maybe a run for 14 days or something, 2 but after 2 days, the pain has subsided and the rest of 3 the product sits in the medicine cabinet. 4 are going to be individuals who you’d suspect to be less 5 likely to develop dependence. 6 Those, again, Now, what that leads me to then is to suggest 7 that very early monitoring of the dispensed 8 prescriptions to do pill count studies and to see how 9 often people are ramping them up, whether they ramp them 10 up more quickly for an established product versus a new 11 product, that's the kind of thing that I don’t think has 12 been done, but could be done. 13 from industry that it has been done, but I haven't seen 14 the published studies. 15 steps that you’d think that are leading toward this 16 outcome of dependence that would give us some early 17 clues about what might happen next. 18 Now, maybe we’ll learn But those are the very early And then time from the first use of the drug 19 to the second use of the drug is not routinely assessed 20 in any of these studies, but it would seem to me that 21 that interval is a very informative interval with 22 respect to the likelihood that someone’s going to become PRECISE REPORTING, LLC jcp 170 1 a repetitive user. 2 DR. WOLFE: 3 DR. KIRSCH: Thank you. Okay. We don't have any more 4 people who want to ask questions, so, we’re going to go 5 on to the presentation by the Sponsor. 6 MS. BHATT: 7 DR. KIRSCH: 8 DR. WALSH: 9 Dr. Walsh. I’m sorry. Thank you. I’m sorry. Dr. Walsh? Yes, actually my question in part, I have two questions, but it falls on 10 from what Dr. Wolfe was just asking about. 11 like we’re being asked to be creative in thinking about 12 what needs to be captured in advance, but we’re asking a 13 question that hasn’t been answered before: 14 detect a signal that's a change either historically or 15 to some kind of comparator? 16 directed to Dr. Anthony. 17 creativity, and in the processed phenotype, thinking 18 about intent. 19 It seems How do we And my questions are I was intrigued by your You said that you were asking about what have 20 you done and then what do you intend to do? 21 heard from people about changes over time that may take 22 place, where people either determine that they can PRECISE REPORTING, LLC And we have jcp 171 1 compensate for the deterrent properties and you would 2 see an increase in use or alternatively, they may decide 3 that it really isn’t worth the time and effort, and then 4 it really does have abuse-deterrent properties because 5 it’s not popular among the abusers. 6 trying to get at the questions of why, what can we learn 7 about the motives between first and second use and 8 strategies for misuse, what can we incorporate into 9 these kinds of studies to ask the questions about why And I’m wondering, 10 and actually get valid information about that that could 11 help us for future drug design questions. 12 DR. ANTHONY: Thanks for the question, but I 13 think there I might have been misunderstood because for 14 reasons you probably know, I don’t usually ask people 15 why they do things because I don’t usually believe the 16 answer. 17 DR. WALSH: Well, that's why I'm asking 18 because I think it’s important to know why here because 19 there's a lot to be learned from it. 20 DR. ANTHONY: 21 DR. WALSH: 22 DR. ANTHONY: Yes. So, how do we construct that to-No, I think it's a great PRECISE REPORTING, LLC jcp 172 1 question, but this isn’t the guy to answer that question 2 because I don’t study things like that, but there are 3 people who do in the social psychology realm, 4 ethnography, anthropology, and so on. 5 DR. WALSH: Well, I mean, the question is not 6 the existential why, it’s really why with this 7 medication, why would you continue to use the original 8 product of OxyContin and why did you not determine that 9 you would use it a second time? 10 DR. ANTHONY: What were the barriers? You're asking great questions, 11 but you’re looking at a large sample, shallow, 12 quantitative epidemiologist. 13 has a little bit more depth in terms of what's going to 14 be probed into the answer to that question. 15 may be someone in the room, but it’s not going to be me. 16 DR. WALSH: And you need someone who And there Well, let me ask you the second 17 question because I think that this one you probably have 18 thought about. 19 presentation from Dr. Keeton that someone raised the 20 fact that we really need to be thinking about the 21 relative risk or abuse, misuse versus exposure, and one 22 measure of exposure is sales, and it’s very possible So, it wasn’t until, I think, the last PRECISE REPORTING, LLC jcp 173 1 that as a new product rolls out that is abuse-deterrent 2 that the sales are going to fluctuate, and that's a 3 changing background, and I’m wondering if you could just 4 say a few things about how to control for those changing 5 things. 6 And then the other thing that I started 7 thinking about was how would sales change? Let’s say 8 that something really is successfully abuse-deterrent, 9 if you just in a very simplistic model think that 10 there’s two categories of sales, and there's more than 11 that, obviously, but one is for legitimate patients and 12 then one is for the pseudo patients who are getting 13 legitimate prescriptions, but then are misusing it or 14 selling it. 15 You’d expect that those two pockets of sales 16 would really be differentially affected once it was 17 determined that the drug was abuse-deterrent and had a 18 lower street value, for instance. 19 control for those factors in the background as we go 20 forward? 21 DR. ANTHONY: 22 introduce three ideas. So, how do you Okay, so, I think I have to PRECISE REPORTING, LLC jcp 174 1 One of them is that with respect to the last 2 point about the heterogeneity and the consumer 3 population, econometricians and economists have great 4 models for that heterogeneity. 5 buying the Volvo for safety considerations or because of 6 its looks or because it used to be from Sweden, I mean, 7 those are human behavioral economic modeling of the type 8 that Dr. Bickel was mentioning is really appropriate 9 there, but I don’t think it’s ever been applied in the 10 11 And whether someone is context of do FDA-type regulation of drug products. Working backward toward the sales, I’ve always 12 really been a critic of 13 denominator when gauging event outcomes like the number 14 of overdoses and so on because the sales, as you just 15 pointed out, are a manifestation of demand of the drug. 16 Again, what you really require here is a multivariate 17 model, not a simple ablative ratio that is going to 18 destroy information that's contained in the sales and in 19 the event rate values. 20 using the sales as a So, I think the answer there is a multivariate 21 model where you’re modeling sales; at the same time, 22 you’re modeling outcomes sometimes with a lag that will PRECISE REPORTING, LLC jcp 175 1 take into account that some of the outcomes are being 2 determined by a product that's been sitting in that 3 medicine cabinet for 6 to 12 months or more, and then 4 finally is getting out into a consumer pool that it 5 otherwise shouldn’t get into. 6 And then back to the exposure question, you 7 know I love to ask people about if they’ve had a chance 8 to try a drug, and I would think with these new 9 products, knowing transition probabilities from having a 10 chance to try, and you could actually show them a 11 picture of the pill and whether they’ve actually used, 12 that transition probability is a really important one. 13 one might expect that a product that's favorable, that's 14 protective will have a longer lag time between chance to 15 try and actual use as compared to one that has a 16 reputation on the street as being the greatest thing 17 since sliced bread. 18 Okay, thanks. 19 DR. KIRSCH: 20 Okay, so, next, we’re going to get to the presentation of the Sponsor. 21 Both the Food and Drug Administration and the 22 public believe in a transparent process for information PRECISE REPORTING, LLC jcp 176 1 gathering and decision making. 2 transparency at the Advisory Committee Meeting, the FDA 3 believes that it is important to understand the context 4 of an individual’s presentation. 5 To ensure such For this reason, FDA encourages all 6 participants, including the Sponsor’s non-employee 7 presenters, to advise the Committee of any financial 8 relationship that you may have with the firm at issue, 9 such as consulting fees, travel expenses, honaria, and 10 interest in the sponsor, including equity interests and 11 those based upon the outcome of today’s meeting. 12 Likewise, FDA encourages you at the beginning of your 13 presentation to advise the committee if you do not have 14 any such financial relationships. 15 address this issue of financial relationships at the 16 beginning of your presentation, it will not preclude you 17 from speaking. 18 There’s been in a change in the organization 19 of the presentation by Purdue. 20 speaker as they come up. 21 Landau. 22 If you choose not to So, I’ll announce each The first speaker is Dr. Industry Presentation: Purdue PRECISE REPORTING, LLC jcp 177 1 Introduction 2 DR. LANDAU: Thank you. Good afternoon. I’m 3 Craig Landau, Purdue’s chief medical officer. 4 of our company, I want to thank the Agency and the 5 combined advisory committees for the opportunity to 6 present and share with the group our Epidemiologic Study 7 Program. 8 whether or not we’ve been successful and to what degree 9 we’ve been successful on effecting abuse and it’s 10 11 On behalf We believe and hope we’ll shed light on consequences with the reformulation. We’ve used reformulation science to lessen 12 OxyContin’s attractiveness to abusers, while retaining 13 the benefits intended for patients. 14 OxyContin in such a way that we view it as a risk 15 mitigation tool to deter its abuse. 16 recognize the role OxyContin has played. 17 vulnerability in the ease with which it could be crushed 18 and how attractive that feature has made it to abusers. 19 So, we’re very happy to be here today to attempt to 20 learn how to do something about that and measure it. 21 The reformulation was approved earlier this year, and 22 we’re actually in the midst of transitioning to the new We reformulated And we certainly PRECISE REPORTING, LLC It’s a jcp 178 1 2 formulation as we speak. Following my introduction, we’ll provide an 3 overview and sort of a description for our rationale in 4 designing this eight-study program. 5 into a presentation of individual studies, a detailed 6 presentation. 7 favor by notifying everyone that the agenda, the 8 sequence of the presentations is different from the 9 original agenda, but the slides that have been provided We’ll then move The Chairman, Dr. Kirsch, has done me a 10 to you are in the correct order. 11 individual study presentations, we’ll conclude by 12 summarizing and offering some closing remarks. 13 So, following the Before the scientific presentations begin, 14 I’ll speak to four topics: 15 reformulating the product, the transition to the new 16 formulation, our thoughts on how formulations intended 17 to deter abuse should be characterized, and our position 18 on label claims for abuse deterrence. 19 our rationale for Millions of patients have been treated with 20 OxyContin since it was approved in 1995. 21 and effective when used appropriately by legitimate 22 patients, we recognize the original formulations PRECISE REPORTING, LLC And while safe jcp 179 1 controlled-release delivery system could be easily 2 crushed, easily defeated. 3 with no more than two spoons or a bottom of a beer 4 bottle could defeat the controlled-release delivery 5 system and convert a controlled-release, twice-daily 6 product to an immediate-release dose form of oxycodone. 7 The result in material could then be swallowed, snorted, 8 or even injected. 9 the purpose of reducing or addressing this vulnerability 10 11 Within seconds, an abuser So, we reformulated the product for and reducing its abuse. The next few slides provide a visual 12 impression and highlight some of the features between 13 the original formulation and the reformulated medicine. 14 And, as you can see here, the two formulations are 15 similar in appearance, but not identical. 16 obvious difference is in the indicia. 17 tablet on the top of the slide, displaying an OC 18 indicia, and the reformulated product on the bottom 19 displaying OP. 20 The most The original And, like most strengths of the formulation, 21 including the 40 mg tablet represented here, most of 22 them are slightly larger and slightly thicker than the PRECISE REPORTING, LLC jcp 180 1 2 original formulation. But, despite their similar visual appearance, 3 the formulations have very different physical chemical 4 characteristics. 5 half of the slide can be converted to a fine powder in a 6 matter of seconds, as I’ve mentioned. 7 tablets require much more effort, much more time, and, 8 in some cases, specific tools and energy to reduce the 9 tablets into smaller particle sizes. The original formulation on the top The reformulated We understand from 10 our experience that inadvertent misuse by patients and 11 intentional abuse by abusers often, but not always, 12 starts with attempts to manipulate the tablet. 13 Of course, the image on top is the original 14 formulation, crushed between two spoons. 15 final powder. 16 reformulated tablet that contains the same tablet 17 strength after five minutes of vigorous manipulation in 18 a mortar and pestle. 19 fragments there. 20 You can see a On the bottom half of the slide, a And you can see large tablet When mixed with a volume of water-soluble for 21 intravenous injection through a tuberculin or an insulin 22 syringe, the original tablet, when hydrated, is easily PRECISE REPORTING, LLC jcp 181 1 drawn up and easily injected. 2 very attractive to those who seek to abuse it via the 3 intravenous route. 4 other hand, becomes quite viscous, difficult to syringe 5 or draw up into a syringe, if not impossible to draw up 6 and inject using commonly-available syringes, 27 or 7 higher gauge needles. 8 9 It’s what makes it so The reformulated tablet, on the These are the properties we introduced into the tablet to make them more difficult to manipulate and 10 less attractive to abusers. It was specifically 11 designed to deter crushing, snorting, and intravenous 12 injection. 13 formulation, and, therefore, it’s considered 14 therapeutically interchangeable for patients. 15 Agency approved the reformulated product on April 5 of 16 this year, and we’re in the midst, as I mentioned a 17 moment ago, of the transition to the reformulation. 18 Our goal is to do this as quickly as possible to reduce 19 confusion, reduce overlap of the two formulations, but 20 to do it not at the expense of patient access. 21 understand the concept of physical dependence, and want 22 to avoid a condition where a patient would go to a It’s bioequivalent to the original PRECISE REPORTING, LLC The We jcp 182 1 pharmacy to fill their prescription and not be in a 2 position to receive one. 3 issue. 4 annually, this transition is quite significant. 5 This could become a safety So, with over 1.2 million patients treated Here on this slide, we see a plot of weekly 6 prescriptions dispensed for OxyContin at retail 7 pharmacies. 8 formulation over time starting in June of this year, and 9 in the orange color, the reformulated OxyContin product. 10 In the yellow color, we see the original A few things to point out here. We stopped 11 shipping the original formulation on Thursday, August 5, 12 and we began to ship exclusively to wholesalers the 13 reformulated product on Monday, August 9. 14 1, the only formulation a retail outlet could obtain 15 from a wholesaler was the reformulated product. 16 the wholesaler level, the supply chain was saturated 17 with only reformulation. 18 As of October So, at For the week ending October 1, 65 percent of 19 all OxyContin prescriptions dispensed were for the 20 reformulation, and, in fact, at the break, I just 21 learned that within the week ending October 8, that 22 number is now 70 percent. By the end of this month, we PRECISE REPORTING, LLC jcp 183 1 expect it to approximate 90 percent, and sometime before 2 or at or about the end of the year, we expect the 3 transition to be nearly complete at the patient level. 4 So, in order to characterize its potential 5 abuse, we worked with the Agency, Abuse Surveillance, 6 and drug safety experts to develop what we think is a 7 very rational, four-step plan, much like the plan or 8 approach described earlier today included in certain 9 guidance documents and referred to by Dr. Rappaport in 10 his memorandum to the combined advisory committees. 11 The first step is in vitro testing in a 12 laboratory, and where we look to define the physical 13 chemical properties of the formulation and go further to 14 define its failure limits. 15 course, and the data was the subject of discussion at 16 the September 2009 meeting of these combined advisory 17 committees. 18 Now, we’ve done that, of Our experiments were designed by external 19 experts in abuse and extraction, and the experiments 20 were designed to reflect methods that abusers currently 21 and could use to defeat a controlled-release delivery 22 system. The large majority of the experiments were done PRECISE REPORTING, LLC jcp 184 1 on the outside through third-party CROs to reduce 2 potential bias. 3 on the outside, as well, by experts. 4 The data were QAed and then interpreted The results of the studies, for those not 5 familiar, suggest the reformulation to be more difficult 6 to purposefully or inadvertently crush. 7 difficult to insufflate if one does to abuse by 8 intravenous injection. 9 presence of alcohol, and that it’s inefficient to abuse 10 Certainly more That it doesn’t dose dump in the by smoking. 11 The second level of testing, pharmacokinetic 12 testing, is fairly straight forward. 13 determine the bioavailability of the intact tablet and 14 tampered tablets along with relevant controls in the 15 volunteer setting. 16 We looked to Studies on the third level, abuse liability, 17 go one step further, and looked to introduce subjective 18 measures or an evaluation of subjective measures. 19 Again, alongside pharmacokinetics, studying intact and 20 tampered reformulated OxyContin and relevant controls. 21 22 So, the in vitro data were submitted to the NDA. They were discussed and reviewed by the Agency, PRECISE REPORTING, LLC jcp 185 1 discussed at the Advisory Committee in September of last 2 year. 3 completed and submitted to the division for review. 4 Each of the levels, the first, second, and third level, 5 as mentioned earlier this morning, are informative for 6 sure, but they're insufficient to predict what will 7 happen when a product is introduced in a real-world 8 setting. 9 needed, and these studies comprised the fourth level of 10 The testing on levels two and three were recently To do this, post-marketing outcome data are testing, epidemiologic testing. 11 Purdue believes all newly-approved opioid 12 products should possess some degree of abuse-deterrent 13 features, whether they be pharmacologic or physical 14 chemical. 15 on each one of these four axes at the relevant time in 16 their development. 17 And all such products should undergo testing The testing performed to date with the 18 reformulated OxyContin product tells us it’s an 19 incremental improvement over the original formulation 20 with respect to its resistance to manipulation. 21 formulations are not a complete solution. 22 that prescription drug abuse is complex, it’s a multi- PRECISE REPORTING, LLC But new We understand jcp 186 1 factorial problem, and it can’t be solved simply by 2 addressing a defect in a single formulation, that is the 3 ease with which OxyContin product could be crushed and 4 its controlled-release delivery system defeated. 5 introducing a new product. 6 be abused whether by swallowing intact tablets or after 7 tampering, after an abuser looks to spend the required 8 time and effort to manipulate the tablet. 9 Or by Our reformulation will still Nonetheless, transitioning to the 10 reformulation as we did, or as we begun in August, 11 represents an opportunity to make a positive impact, and 12 that's what we’re here to do. 13 of the formulation requires diverse approaches, and 14 that's what this meeting is about. 15 Understanding the impact The transition and the unprecedented challenge 16 of designing the studies we’re discussing provides us a 17 unique opportunity. 18 to measure if and how a change in a formulation can 19 impact a clinical outcome. 20 this opportunity before. It’s an opportunity to research and I don't know that we’ve had 21 We’ve also learned that no single 22 epidemiologic study can adequately assess the impact a PRECISE REPORTING, LLC jcp 187 1 formulation can make. 2 studies, the multi-study program we’ve proposed will 3 give us a lens and help us to understand if and how one 4 can be successful; we or other Sponsors. 5 We’re hopeful that the eight We’re pursuing the studies to meet a post- 6 marketing requirement issued to Purdue as an approval 7 requirement, and, of course, to learn what impact, if 8 any, a formulation could have on abuse and its 9 consequences. We did not design these studies with the 10 goal of pursuing a label claim, and we’re not currently 11 seeking one. 12 Given the complexities of the studies and the 13 context in which we’re discussing them, we feel a 14 conservative approach to both interpretation and 15 communicating the results of the study is warranted. 16 start with an assumption that within the class or within 17 a given schedule of drug, the abuse liability of all the 18 products should be considered the same or similar. 19 claim of reduced abuse liability must be based on 20 substantial, sustained, and consistent effects measured 21 over time in a real-world setting on a number or axes of 22 evaluation. We understand the complexities and the PRECISE REPORTING, LLC We Any jcp 188 1 interrelationships of some of the outcomes we’re going 2 to be presenting in a few minutes. 3 Even if the evidence is deemed sufficient to 4 support a label claim at some point going forward, there 5 were risks to making one. 6 this morning talk about them. 7 members of this panel and previous meeting have 8 surfaced, creating a false sense of security in the 9 minds of prescribers, dispensers, and patients is We heard some of the speakers For one thing, and 10 something we’d look to avoid. 11 cause reduced vigilance. 12 eliminate or at least reduce any of the potential 13 benefits of formulation like our reformulation could 14 introduce, and it could certainly undermine the goals of 15 the class REMS currently under consideration within FDA. 16 It could introduce or Reduced vigilance could I said more plainly reduced vigilance could 17 worsen the already significant public health crisis we 18 spoke about this morning, that of prescription opioid 19 abuse and prescription drug abuse, something we wish to 20 avoid. 21 generate a level of convincing evidence substantial and 22 sustained and consistent across axes, I think at that If we, in the future, or any sponsor does PRECISE REPORTING, LLC jcp 189 1 point, we would hope that we press pause and a benefit 2 risk assessment be made. 3 we can that the benefits of introducing language in a 4 product’s label offset the risk we talked about in the 5 context of the larger public health. 6 We’d want to be as certain as To help us design and understand the results 7 of the studies we’ll be presenting in a moment, we’ve 8 enlisted the help of outside experts. 9 with us today, and you’ll from Dr. Richard Dart in a A couple are here 10 moment. 11 Andrews, Dr. Greg Burkhart, and Dr. Richard Dart, Dr. 12 Ken Rothman, and Dr. Ed Sellers. 13 The experts are listed here, Dr. Elizabeth In a moment, we’ll move forward with the 14 agenda. 15 Purdue’s new head of Risk Management in Epidemiology. 16 He’s also an adjunct assistant professor at the 17 University of Pennsylvania School of Medicine in the 18 Department of Clinical Biostatistics in Epidemiology. 19 20 21 22 You’ll hear from Dr. Paul Coplan next. He’s Paul will provide the overview and the rational for the eight-study program. Following Paul, five external experts and one internal expert, Dr. Howard Chilcoat, who’s new to PRECISE REPORTING, LLC jcp 190 1 Purdue, will go through each of the individual studies 2 in more detail. 3 associate professor at Johns Hopkins Bloomberg School of 4 Public Health, and was also chief of the epidemiology 5 research branch at the National Institute on Drug Abuse. 6 And when we conclude the study presentations, Paul will 7 come back with some concluding remarks and direct 8 responses to any questions the committee or anyone has 9 on our work. 10 Prior to joining Purdue, Howard was an Thank you. 11 Overview and Rationale of Study Program 12 DR. COPLAN: Thank you, Dr. Landau. Good 13 afternoon. Thank you for the opportunity to address 14 this committee. 15 provide an overview and rationale for the Epidemiologic 16 Study Program that Purdue plans to conduct to assess the 17 effect of the new formulation on abuse. 18 designed to meet FDA’s requirements for post-marketing 19 studies. 20 FDA, members of the Advisory Committee, and other 21 experts to ensure that the studies answer the most 22 important public health questions about the new The purpose of this presentation is to This program is We very much appreciate the insight of the PRECISE REPORTING, LLC jcp 191 1 formulation, particularly in the light of the 2 limitations of each of the available data sources that 3 we heard about earlier today. 4 In order to evaluate the effects of a new 5 formulation on the epidemiology of prescription opioid 6 abuse, it is important to first characterize the 7 background epidemiology of prescription opioid abuse and 8 its adverse consequences. 9 presentations, we have worked to provide the relevant For each of the upcoming 10 background epidemiology so that you can evaluate the 11 relevance of the study designs, the study populations, 12 and the outcome measures. 13 including baseline data, may also assist in predicting 14 the hypothesized effect of the new formulation. 15 This background epidemiology, OxyContin is the extended-release formulation 16 of oxycodone. 17 original formulation was easily circumvented by 18 crushing, breaking, or dissolving. 19 earlier question by Dr. Omoigui, this is crushing the 20 tablet and dissolving it for purpose of snorting or 21 injecting. 22 The extended-release system of the And to address an A characteristic of oxycodone in general, and PRECISE REPORTING, LLC jcp 192 1 extended-released oxycodone in particular, is the 2 diversity of routes by which it is popularly abused. 3 get a baseline picture of these routes of abuse from the 4 National Surveillance System of 500 abuse treatment 5 centers around the U.S. 6 than 7,000 people who enter treatment in one of the 2 7 centers of the network reported abusing OxyContin during 8 the intake clinical examination by at least 1 of 4 9 routes. In the past 3.5 years, more Snorting was the most common route of abuse, 10 and 34 percent reported abuse by injecting. 11 numbers don’t add up to 100 percent because subjects 12 could endorse more than one route of abuse, as Dr. 13 Paulozzi referred to earlier. 14 We These Other frequently-abused opioid drugs have 15 fewer primary routes of abuse. 16 primarily abuse hydrocodone via the oral route and abuse 17 morphine through injecting, as we’ll see in a later 18 presentation by Ms. Cassidy. 19 For example, people Oxycodone’s diversity of route of abuse may 20 have increased its popularity among people who initiate 21 abuse via the oral route and who progress to more 22 frequent abuse via injecting and snorting. PRECISE REPORTING, LLC jcp 193 1 This pattern is reflected in data from a study 2 of people abusing OxyContin in Kentucky. 3 Appalachian region that has one of the highest rates of 4 death from opioid overdoses in the U.S. 5 people entering a treatment program of an average of 6 19.7 months of abusing the drug. 7 initial route of administration when they started 8 abusing OxyContin and is based on the medical chart 9 information. 10 Located in the It looks at The left bar shows the The right bar shows the stated routes of 11 administration upon admission to the treatment center. 12 There were substantial shifts in the routes of abuse 13 after 19 months of abusing. 14 percent to 58 percent and injecting increased from 1 15 percent when initiating abuse to 21 percent upon 16 treatment admission. 17 how routes of abuse can change as the stages of abuse 18 progress and suggest the likely benefits the 19 reformulation may provide. 20 Snorting increased from 16 These data provide one example of For legitimate users, the hypothesized benefit 21 is to provide bioequivalent delivery of oxycodone to 22 treat moderate to severe pain while reducing the PRECISE REPORTING, LLC jcp 194 1 inadvertent areas in usage that the patients and nursing 2 staff sometimes commit through breaking, crushing, or 3 chewing of OxyContin, which was previously referred to 4 by the FDA definitions as misuse. 5 OxyContin overdose associated with such areas have been 6 reported to the Purdue Case Report Database. 7 To date, 155 cases of For abusers of OxyContin, the hypothesized 8 benefits if the physical chemical properties of the 9 reformulation impede tampering are reduction in abuse 10 through breaking, crushing, or chewing due to the 11 tablet’s hardness. 12 hydro-gelling of the tablet when dissolved in water, as 13 Dr. Landau mentioned earlier, snorting will reduced due 14 to properties of the new formulation tablet that, when 15 crushed, they crumble into large chunks rather than a 16 fine powder created by crushing the old formulation, and 17 the large chunks form a gel in the nasal passages. 18 is requiring measurable endpoints for the epidemiology 19 studies. 20 these endpoints. 21 22 Injecting will be reduced due to the FDA The hypothesized benefits need to be mapped to In FDA’s background document for this meeting, FDA stated future studies need to address abuse and PRECISE REPORTING, LLC jcp 195 1 misuse and its consequences, overdose, addiction, and 2 death. 3 potential impact of the new formulation can be studied 4 by assisting changes in the prevalence of abuse, demand 5 for purposes of abuse, and abuse via routes of 6 administration that require tampering. To measure the impact on abuse and misuse, the 7 It is possible that reducing abuse via routes 8 that require tampering may have a downstream benefit of 9 reducing the clinical outcomes of abuse, including 10 overdose, addiction, and death. 11 referred to as the consequences of abuse. 12 tasked with designing epidemiology studies that will 13 measure the effects of the new formulation. 14 These are what FDA has We have been Because of the multifaceted nature of the 15 prescription opioid abuse problem in the overall 16 community, no single study could assess all the aspects 17 of the problem, such as the subpopulations affected, the 18 influence of routes of administration, and the stages of 19 addiction that we heard in the discussion by Dr. Anthony 20 earlier. 21 and limitations relating in part to the strengths and 22 limitations of available data sources. Each of the studies have specific strengths PRECISE REPORTING, LLC As a result, we jcp 196 1 engaged a panel of experts to help us design multiple 2 studies to measure the different aspects of the problem. 3 based on the input, we’ve developed a program of studies 4 designed to comprehensively measure the effects of the 5 new formation. 6 collaborating evidence of the real-world effects across 7 studies and multiple data sources. 8 totality of the program is designed to leverage the 9 strengths and address the limitations of the individual 10 The individual studies will provide Taken together, the studies. 11 The key questions that we will be seeking to 12 address in these eight studies will be based on these 13 design principles. 14 could be phrased as a question or a hypothesis, is: 15 Will the methods for tampering and extraction become 16 widespread? 17 be developed, but if they require too much effort and 18 are too burdensome, they’re unlikely to become 19 widespread. 20 The first question is, and this We know that certain routes of abuse will Is the reduction in abuse via routes that 21 require tampering? 22 community? Is there a reduction in abuse in the Is the reduction in demand for purposes of PRECISE REPORTING, LLC jcp 197 1 abuse? 2 reduction in clinical endpoints, including in specific 3 subpopulations of pain patients, the general population, 4 and the people entering treatment for addiction? 5 questions will then be matched to specific studies to 6 address the questions. 7 And, lastly, and most importantly, is there a These The Internet discussion will be conducted to 8 address the first question. A study of an abuser cohort 9 in Kentucky and surveillance of addiction treatment will 10 be conducted to address the second question. 11 such as the NSDUH, and other surveys will be conducted 12 to assess the third question. 13 RADARS System and doctor shopping and Prescription 14 Monitoring Programs will be conducted to assist the 15 fifth question. 16 Surveys, Law enforcement in the And lastly, the Kaiser Overdose Study and the 17 Poison Center Program--the Kaiser Overdose Study will be 18 conducted to assist outcomes in pain patients. 19 assist outcomes in the general population, the Poison 20 Center Program, and, in addition, the Kaiser Overdose 21 Study will be used. 22 routes of abuse and types of abuse in people entering To And, lastly, to monitor changes in PRECISE REPORTING, LLC jcp 198 1 treatment for addiction, we’ll be using a study of 2 surveillance of addiction treatment centers. 3 A table showing the eight studies and the 4 outcomes they’re designed to measure would be helpful 5 for summarizing the program. 6 eight studies cover the outcomes that FDA is requesting 7 to be addressed. 8 background document for this meeting, only seven studies 9 were listed. This helps to see that the I should mention that in produced We added an eighth study, a study of law 10 enforcement event in the RADARS System. 11 added because when we were preparing for this meeting, 12 some external experts told us that a study of law 13 enforcement events would be helpful. 14 already conducting such a study as part of an ongoing 15 risk management activity, we added this to the study 16 program. 17 This study was Since Purdue was The duration of baseline data for each of 18 these studies will be relevant to the ability of the 19 studies to measure changes over time, and we’ll be 20 coming back to this slide repeatedly and using it as a 21 structure for organizing each of the eight studies so 22 that the eight studies don’t seem like a morass of PRECISE REPORTING, LLC jcp 199 1 diffused studies, but that their purpose in measuring 2 specific outcomes is apparent or hopefully apparent. 3 The duration of the baseline varies from 6 4 months for the abuser cohort in Kentucky to 7 to 8 years 5 in the Drug Diversion Study, and the Poison Center 6 Study, in 10 years in the Kaiser Overdose Study. 7 the studies already collecting the data needed for 8 validating the effect of the new formulation. 9 All of For the Poison Center and Drug Diversion 10 Programs, Purdue has been receiving quarterly reports 11 from the Rocky Mountain Poison Center and will continue 12 to get these in an ongoing fashion, and Dr. Rick Dart 13 will present more on that in a moment. 14 The Kaiser data is an Electronic Medical 15 Record System that collects data on patients using 16 OxyContin in an ongoing way. 17 endpoints for the study are being collected. 18 So, as we speak, the The time to see an effect for the new 19 formulation, we estimate, will vary depending on the 20 nature of the endpoint that the study’s designed to 21 detect. 22 measure routes of abuse, according to our best estimate, It’ll take six to nine months for studies that PRECISE REPORTING, LLC jcp 200 1 one to two years to see an effect for studies that 2 measure changes in usage and demand, and one to two 3 years, probably closer to two years, to see an effect 4 for studies that measure changes in clinical outcomes. 5 The planned analysis must take into account the baseline 6 trend, the pre post changes, and compared to comparator 7 drugs, as was mentioned earlier by the FDA speaker. 8 9 The planned analyses will use an interrupted time series approach where baseline trend is identified, 10 a change shortly after the introduction of the new 11 formulation is evaluated, and, secondly, the trend over 12 time after the introduction of the new formulation is 13 evaluated. 14 OxyContin as well as for comparator opioids. 15 The changes in trends will be assessed for The depiction on this slide is one of the ways 16 in which we hypothesize success will look. 17 the trend for OxyContin, but not for other comparator 18 opioids. 19 prescription opioids that can capture background trends 20 in abuse of prescription opioids. 21 22 A change in It is important to select comparator Immediate-release oxycodone provides a close comparison to OxyContin since they share a common active PRECISE REPORTING, LLC jcp 201 1 drug substance and only different extended-release 2 mechanism versus the immediate-release formulation. 3 Hydrocodone combinations are one of the most frequently 4 prescribed and abused types of opioids in the U.S., and, 5 therefore, provides a relevant comparator. 6 Extended-release opioids, such as extended- 7 release morphine sulfate, excluding naltrexone- 8 containing formulations, and the fentanyl transdermal 9 patch provide a useful comparator since they are 10 indicated for the same indication as OxyContin. 11 addition, these comparators are also included with class 12 REMS for longer-acting opioids. 13 control for the background effects of the class REMS. 14 In So, they provide a Lastly, methadone will be used as a comparator 15 because in many mortality studies, methadone, at least 16 in the past, did appear as the number one cause of 17 opioid overdose deaths. 18 As an example of why it is important to 19 measure background trends and comparator opioids, 20 changes in immediate release or I’ll call them IR, and 21 extended releases, or ER oxycodone are relevant. 22 Over the past 10 years, there have been large PRECISE REPORTING, LLC jcp 202 1 shifts in prescribing practices for extended-release 2 oxycodone and immediate-release, single entity oxycodone 3 that must be considered when utilizing immediate- 4 release, single entity oxycodone as a comparator. 5 data from SDI looking at retail prescriptions shows that 6 there’s been a 40 percent rise in the number of 7 prescriptions for extended-release oxycodone from 5.5 8 million prescriptions in 2000 to 7.7 million 9 prescriptions in 2009. 10 This In the same time period, there’s been a 660 11 percent rise in the number of prescriptions for IR 12 single entity oxycodone from 1.2 million prescriptions 13 in 2000 to 9.2 million prescriptions in 2009. 14 source of the data is the FDA’s Advisory Committee 15 Briefing Document from the last advisory committee of 16 this group. 17 The In selecting a comparator, it is also 18 important to differentiate between immediate-release 19 single entity oxycodone and immediate-release 20 combination oxycodone. 21 immediate-release combination of oxycodone prescribed in 22 the U.S. is for oxycodone-acetaminophen combinations. Ninety-nine percent of the PRECISE REPORTING, LLC jcp 203 1 Data from the DAWN study that the FDA 2 presented to this advisory committee on April 22 show 3 that the risk of emergency department visits for non- 4 medical use using a denominator of 10,000 prescriptions 5 that was obtained from the SDI data show that there's 6 substantially different risk of ED visits for immediate- 7 release single entity oxycodone, which is shown in the 8 violet line and immediate-release oxycodone in 9 combination shown in the green line. 10 In 2008, the risk of overdose at adverse 11 events was substantially higher for immediate-release 12 single entity oxycodone than for IR oxycodone 13 combination. 14 thousand prescriptions respectively. 15 and 2008, the risk of overdose adverse events was 20 16 percent for ER oxycodone, 62 percent for IR oxycodone 17 combined, and 150 percent for IR oxycodone single 18 entity. 19 trends into account when designing studies and 20 interpreting the results. 21 differentiate whether the cause of overdose or death is 22 due to IR or ER oxycodone could thus mass-effect with Forty-five versus thirteen per ten And between 2004 It would be important to take these background Data systems that do not PRECISE REPORTING, LLC jcp 204 1 2 tamper-deterrent formulation. This overview and rationale provides a context 3 for the presentation of the individual studies. The 4 first study utilizes the existing Electronic Medical 5 Records System of the Kaiser Permanente Health System 6 that records overdoses and poisonings due to opioids. 7 If the new formulation is successful, this study will 8 demonstrate reductions in the instance rate of OxyContin 9 overdoses and poisonings per prescriptions for OxyContin 10 amongst the membership of the Kaiser population. 11 first study will be presented by Dr. Nancy Perrin, who’s 12 a senior investigator at Kaiser Permanente Northwest in 13 Portland, Oregon. 14 Overdose Rates in OxyContin Patients and Non-Patients at 15 Kaiser Permanente 16 DR. PERRIN: Good afternoon. The I’m Nancy 17 Perrin, a senior investigator at the Center for Health 18 Research at Kaiser Permanente Northwest. 19 expertise is biostatistics and research design. 20 My area of Prior to coming to the Center, I was professor 21 and director of the Statistical Corps in the School of 22 Nursing at Oregon Health and Science University. PRECISE REPORTING, LLC I have jcp 205 1 no personal financial interests in the outcome of this 2 meeting, and I have been paid by Purdue for my time. 3 I am leading this study, exploring the rates 4 of adverse events in OxyContin patients and non-patients 5 at Kaiser Permanente. 6 relevant because it focused on clinical outcomes in a 7 broad population. 8 Medical Records to establish the trend and adverse 9 events prior to the introduction of the new formulation. The study is particularly We have 10 years of Electronic 10 The study will provide data on the impact of the new 11 formulation on opioid-related adverse events reported 12 within the Kaiser Permanente System. 13 We have already gathered data to determine the 14 baseline trend. 15 related poisonings in Kaiser Permanente Northwest from 16 1998 to 2009. 17 with opioid dispensings in the six months prior to the 18 event. 19 This is our initial estimate of opioid- We identified poisonings and linked those The graph shows the number of poisonings among 20 people with various dispensings of opioids. 21 and extended-release oxycodone are shown in orange, 22 immediate-release oxycodone in pink, and other Schedule PRECISE REPORTING, LLC OxyContin jcp 206 1 II opioids in yellow. 2 The trend in the poisonings is increasing over 3 time for all the groups, which increases our power to 4 detect a change with a new formulation. 5 many of the opioid-related poisonings are among patients 6 that did not have a dispensing of any opioid in the six 7 months prior to the adverse event, as shown in the blue 8 line here at the top of the graph. 9 people who are intentionally misusing opioids. Interestingly, These are likely 10 As Dr. Coplan mentioned, we are continuing to 11 assess adverse events since the introduction of the new 12 formulation. 13 the rate of overdose adverse events associated with 14 OxyContin decreases with the new formulation. 15 population for this study are Kaiser Permanente Health 16 Plan members, both with and without dispensing of 17 opioids. 18 and overdose adverse events derived from Electronic 19 Medical Records. 20 it will take 2 years after the introduction of the new 21 formulation of OxyContin to determine its effect. 22 The objective of our study is to assess if The Using a cohort study, we’ll examine poisoning We have 10 years of baseline data, and Kaiser Permanente, or KP, as we like to call PRECISE REPORTING, LLC jcp 207 1 it, has multiple regions, many of which will be included 2 in our study. 3 KP Northwest, which has over 475,000 members annually 4 and will guide the scale of the full study. 5 The pilot work is being conducted at the For the full study, we have access to data 6 from over 8 million members across the regions of KP. 7 The regions are linked by the Virtual Data Warehouse, 8 which is a unique data resource that combined 9 comprehensive membership, demographic, in-patient 10 utilization, outpatient utilization, dispensed 11 prescriptions, laboratory tests, and imaging data dating 12 back to 1996 from multiple health plans. 13 from Electronic Medical Records, not insurance claims. 14 It’s derived The KP population and the Virtual Data 15 Warehouse provide us with an opportunity to also examine 16 adverse events among family members of patients 17 dispensed opioids. 18 individuals with dispensings of opioids and conduct 19 separate analyses for this sub-sample. 20 opportunity to examine accidental use and misuse of 21 OxyContin. 22 We will identify family members of This is a unique As I mentioned, we are studying the trend in PRECISE REPORTING, LLC jcp 208 1 adverse events over time. We plan to compare cohorts 2 with dispensings of different opioids, as Dr. Coplan 3 mentioned, with an interrupted time series approach. 4 The interrupted time series design is an optimal method 5 for conducting naturalistic studies of the effect of 6 system level changes, such as the introduction of the 7 new formulation of OxyContin. 8 the trend over time and the rate of adverse events 9 before and after the introduction of the new This approach compares 10 formulation. 11 we can test if it follows the same path as the baseline 12 trend or if there has been a significant change in the 13 trend. 14 As we begin to observe the trend forward, Critical to the validity of the study is the 15 proper identification of adverse events in the 16 Electronic Medical Records. 17 derived from ICD-9 and 10 codes. 18 classified as poisonings or overdoses. 19 we are capturing changes in the use of the ICD codes 20 over time, we are currently conducting chart audits to 21 examine patterns of codes used to document opioid- 22 related adverse events in the KP Northwest System. Adverse events can be These events can be PRECISE REPORTING, LLC To be sure that jcp 209 1 Additional audits are being used to validate our 2 algorithms to extract adverse events from the Electronic 3 Medical Records. 4 The analysis is based on rates of adverse 5 events. We will actually look at multiple rates. The 6 first is the rate of OxyContin-related adverse events 7 for which the numerator is the number of adverse events 8 among patients prescribed OxyContin and the denominator 9 reflects the extent of the use of OxyContin. The rate 10 will be calculated for each time period in the time 11 series and compared to rates of adverse events among 12 patients prescribed other opioids. 13 comparator rates for OxyContin immediate release and 14 other Schedule II opioids, and we can vary these 15 denominators to capture different subpopulations such as 16 people with new prescriptions of opioids. 17 We will compute Comparing the rates allows us to control for 18 changes in the number of members and prescribing 19 patterns across time. Regression will be used for the 20 statistical analyses. The model we will use yields 21 unbiased estimates of the level and slope of the trend 22 in the adverse events prior to the introduction of the PRECISE REPORTING, LLC jcp 210 1 new formulation to the left of the dotted line here and 2 estimates of the change in the level and slope after the 3 introduction of the new formulation. 4 The change in level provides an estimate of 5 the immediate effect. 6 estimate of the difference in the trends between the two 7 time periods. 8 9 The change in slope provides an The main comparison is between the slope of the trend prior to and post the new formulation. This 10 means that statistical power of the study is a function 11 of the change in slope. 12 provide greater statistical power to detect changes 13 after the implementation of the new formulation. 14 rates based on large sample sizes and the same 15 populations over time improve stability. 16 regions of KP provide very large populations to estimate 17 these trends. 18 Long, stable baseline periods And The multiple In our pilot work, we have observed in KP 19 Northwest over 580 opioid-related poisonings per year in 20 recent years, and estimate there will be approximately 21 125 to 150 additional overdose events per year. 22 calculated preliminary estimates of power based on the PRECISE REPORTING, LLC We have jcp 211 1 pilot data for poisonings in KP Northwest. 2 We are still working on our algorithms to 3 extract overdose events. Approximately 1 percent of 4 patients in KP Northwest with dispensings of OxyContin 5 has a poisoning event. 6 patients needed to detect various degrees of change from 7 the two years prior to the new formulation to the year 8 after the introduction of the new formulation. 9 Approximately 3,600 patients are needed to detect a 50 We calculated the number of 10 percent reduction in the rate of poisonings and with 80 11 percent power, and approximately 4,800 patients with 90 12 percent power. 13 estimates of power as they were based on comparisons of 14 two rates, not the differences in the slope of the 15 trends over time, and they were based on poisonings 16 only. 17 These are preliminary, conservative The main power analysis will be conducted when 18 we conclude our pilot work, and we’ll use simulations 19 approaches to determine the sample size needed to detect 20 various changes in slopes. 21 estimates, we’ll determine which KP regions to include 22 in the main study to assure an ample sample size. Based on these power PRECISE REPORTING, LLC jcp 212 1 One strength of our design is our ability to 2 statistically compare trends over time for different 3 opioids. 4 internal validity of our study. 5 changes and trends over time between OxyContin and 6 comparator opioids can be tested by incorporating 7 interaction terms into the regression analyses. 8 might find that there's no change in the trend of 9 adverse events pre and post the introduction of the new Including comparator groups improves the Differences in the We 10 formulation for either OxyContin or the comparator 11 opioid, as illustrated here. 12 the rate of adverse events for OxyContin after the new 13 formulation is introduced, but not the comparator 14 opioid, as illustrated on the left. 15 Or we may see a decline in Alternatively, there could be a decline in the 16 rate of adverse events for OxyContin and an increase in 17 the rate of adverse events for other opioids. 18 I have described today has its strengths and 19 limitations. 20 fact that not everyone fills their prescriptions at KP. 21 However, we do know from previous work that less than 10 22 percent of prescriptions are filled outside of the KP The study The limitations of the design include the PRECISE REPORTING, LLC jcp 213 1 System. We may not always be able to identify the 2 opioid uniquely associated with an adverse event since 3 people can be prescribed more than one type of opioid, 4 especially to manage pain within the observation window. 5 The study does have a limited socioeconomic profile as 6 it is conducted in an insured population. 7 The strengths of the study include the ability 8 to incorporate a comparator time series of other 9 opioids, improving the internal validity of the 10 research. We have a 10-year baseline period from which 11 to detect changes in trends after the new formulation. 12 Use of data derived from Electronic Medical Records and 13 access to geographically diverse regions of KP through 14 the Virtual Data Warehouse are additional strengths of 15 the study. 16 Thank you. 17 DR. COPLAN: 18 Our next study, the second study, utilizes the Thank you, Dr. Perrin. 19 existing network of poison centers in the U.S. 20 new formulation is successful, this study will 21 demonstrate reductions in the number of OxyContin 22 exposures reported to poison centers over time. PRECISE REPORTING, LLC if the This jcp 214 1 study will also assess whether the number of deaths from 2 OxyContin reported to poison centers declines as a ratio 3 of the number of reported exposures. 4 The presenter for this is Professor Rick Dart. 5 Exposures Reported to Poison Centers 6 DR. DART: Good afternoon, and thanks for the 7 opportunity to present to the committee and describe how 8 RADARS can be used to assess the new formulation of 9 OxyContin. 10 First, my disclaimer. I have no personal 11 financial interest in the outcomes of this meeting, 12 however, my travel expenses only were paid for by Purdue 13 for this trip. 14 I’d like to start with an overview of the 15 RADARS system. The idea here, of course, is we’re 16 trying to understand prescription medication abuse and 17 misuse, and, as several speakers have said today, you 18 need to look at it from multiple perspectives because 19 these people rarely present themselves. 20 principle behind RADARS is to do exactly that, is to 21 create a mosaic by looking at prescription medication 22 abuse from several different perspectives. PRECISE REPORTING, LLC So, the We currently jcp 215 1 have six different programs, and in response to one of 2 the comments made this morning, we actually have just 3 created the methodology and tested it to look at street 4 price of prescription opioids, as well, and we’ll be 5 adding that program to the RADARS System. 6 Purdue and most of the manufacturers of 7 opioids use RADARS for their post-marketing commitments 8 and to perform risk management. 9 evaluation of OxyContin though, we’re focusing on three Specifically for the 10 programs, and I’m going to talk about two of those 11 today. 12 criminal justice system through drug diversion and acute 13 events through the poison center. 14 The first two you see here, which is the First, the poison centers. The U.S. is very 15 fortunate, at least I think so, in that we have a system 16 of 60 poison centers across the country that cover every 17 part of the United States, and you can reach them 18 through a single toll-free number, the poison help 19 number, and that leads the caller to initial triage and 20 care advice for their poisoning or their exposure I 21 should say because not all of these are true poisonings. 22 That information, they’re helped by a health care PRECISE REPORTING, LLC jcp 216 1 professional, either a nurse of a pharmacist, and that 2 information is entered into a single database from all 3 poison centers. 4 software has certain data-checking elements to it to 5 make sure we don’t enter pregnant males and that type of 6 issue. We all use the same fields, and that 7 Eventually, the patient receives a 8 disposition, they're followed throughout their hospital 9 course by the poison center by telephone, and that leads 10 to reporting on QA and QC. 11 centers in the United States currently participate in 12 the RADARS System, and if you take their coverage areas, 13 their jurisdictions, if you will, that's 240 million or 14 almost 85 percent of the U.S. population. 15 Now, 49 of these 60 poison One nice thing about poison centers is you 16 really get 100 percent reporting rate every quarter 17 because we can bug them until they do it. 18 Now, an exposure is defined as an individual 19 taking a drug leading to a call to the poison center, 20 some type of event. 21 and signs. 22 medication. Often, there are medical symptoms Sometimes it is I took an extra dose of Will that lead to an adverse event? PRECISE REPORTING, LLC For jcp 217 1 just the opioids and RADARS’ participating poison 2 centers, there were over 42,000 exposures reported in 3 2009. 4 So, we have quite a bit of baseline data on 5 these medications. This shows the reported intentional 6 exposures using the denominator of population. 7 can use the denominator of individuals filling a 8 prescription for the drug, but for simplicity’s sake, 9 I’m showing this as the population denominator. We also 10 You can see in the purple line, which is 11 immediate-release oxycodone, that there has been a 12 relentless increase in the amount of just a single 13 entity immediate-release oxycodone over the past seven 14 years. 15 been more stable, but has a slight increase in 2008 and 16 2009. 17 oxycodone, which was introduced in 2004, and has 18 subsequently been largely withdrawn from the market. 19 So, it came, was detected by the Poison Center Program, 20 but then has gone back down as the drug was decreased. 21 If we focus just on OxyContin in this slide, 22 we have the Poison Center data just for OxyContin with The orange line is OxyContin. OxyContin has the yellow line is generic extended-release PRECISE REPORTING, LLC jcp 218 1 the rate of intentional exposures again. 2 line is the combination actually of OxyContin and the ER 3 oxycodone. 4 confidence intervals around that number, and then, as 5 you’ll see, and these data, I should point out, go 6 through June of 2010. 7 The orange The light blue lines are the 95 percent So, in RADARS, we report out our data three 8 months after the close of each quarter. 9 data through June. So, we have That's all real data up to that 10 dotted line. And then after the dotted line are the 11 potential effects of the introduction of a new 12 formulation of OxyContin. 13 hypothesis would be that it could continue unchanged. 14 In theory, it could even increase, but, for some reason, 15 the new formulation was more attractive and used more. 16 Or it could decrease, as shown by the orange triangles. 17 We're also going to look at the case fatality And, as you can see, the 18 rate for OxyContin. This slide is just a sample to 19 compare OxyContin to methadone. 20 because it has the highest case fatality rate in the 21 RADARS’ Poison Center Program. 22 into greater than 12 and less than 12-years-of-age, but Methadone was chosen I’ve broken the ages PRECISE REPORTING, LLC jcp 219 1 we get the patient’s age in each case. 2 analyze this by any age group that you would like. 3 So, we can For methadone, if you take the younger 4 patients, you can see that there are still some deaths, 5 but over the period of 2003 to 2009, there were only 7, 6 although, in my world, 7 is a lot. 7 get older, you can see the rate goes up to 241 deaths. 8 So, really 1.5 percent of cases coming to a poison 9 center involving methadone end up as a fatality. 10 And then as people For OxyContin, the rates are lower. I’d say 11 there’s relatively high, but lower than methadone. 12 greater than 12, it was .58 percent. 13 For If we plot the trends of these over the years, 14 this shows methadone, fentanyl, OxyContin, and 15 hydrocodone. 16 group in the top two lines and OxyContin and hydrocodone 17 in the bottom two lines. 18 relatively stable, which will help us see a difference 19 if one occurs after the introduction of OxyContin. 20 You can see that fentanyl and methadone And the baseline data here is To summarize that study, what we’ll be looking 21 at will the incidence rate, the first slides I showed, 22 change after the introduction of the new formulation, PRECISE REPORTING, LLC jcp 220 1 and will the case fatality rate change after the 2 introduction of the new OxyContin? 3 These calls come from the general population. 4 In fact, all poison centers have to solicit calls from 5 the entire population of their service area. 6 observational time series. 7 be the case fatality rate, but also the incidence rate, 8 as I mentioned. 9 and we think that we will see an effect in six to nine 10 It’s an And the primary outcome will We have seven years of baseline data, months if there is an effect. 11 I mentioned that a case is an intentional 12 exposure, and an incidence rate will be calculated 13 simply by taking the cases per quarter for a specific 14 medication and dividing it by the population that was 15 actually covered by the poison centers covering that. 16 That’s the 85 percent I was describing earlier. 17 Case fatality rate is a measure of toxicity, 18 and this is exposures resulting in death divided by the 19 total exposures. 20 of OxyContin could result in a decreased case fatality 21 rate. 22 like because we collect this information on all of the The idea is that a change in toxicity We can use any opioid comparator that you would PRECISE REPORTING, LLC jcp 221 1 2 opioid medications. For our analytic approach, the case fatality 3 rate is an interrupted time series. We have to realize 4 that event rates are relatively low, and, so, the data 5 will be analyzed using a Poisson distribution. 6 The time series data tend to be auto 7 correlated. 8 current inferences. 9 before and after the formulation change. 10 11 So, data will be modeled to allow the Separate trend lines will be fit All of this is accommodated by using a generalized linear mixed model. The main limitation to poison center data is 12 that calling a poison center is not mandatory. 13 voluntary act by someone or their friends or family who 14 feel a need after an exposure to call the poison center. 15 However, it’s a large system, and it seems unlikely that 16 systematic changes will occur across all 49 centers 17 simultaneously, and our long track record of baseline 18 data will help us show that. 19 It’s a The main strength is that it has large 20 national coverage of the general population, and we have 21 consistent data collection. 22 data fields, as I mentioned, but RADARS has a specific All centers use the same PRECISE REPORTING, LLC jcp 222 1 change unique to RADARS. 2 If you think about NPDS, the National Poison 3 Data System of the AAPCC, American Association of Poison 4 Control Centers, all these centers participate in both 5 systems. In fact, the data feed for RADARS is identical 6 to NPDS. The difference is that we collect more fields, 7 and, in particular, we collect what’s called the case 8 notes for each case. 9 specialist fills out during the case. These are notes that the We use that 10 information to check fields like product coding and 11 route of administration to make sure that that data is 12 accurate and internally valid. 13 As I mentioned, the data are available within 14 three months of the close of each quarter, and another 15 advantage that's my particular favorite is that cases 16 involving children can be analyzed separately to see if 17 there's any unintended effects of the introduction. 18 DR. COPLAN: Thank you. 19 So, the first two studies addressed the 20 outcomes of overdose and death. The third study, now 21 we’ll move on to studies that will address the first 22 three outcomes. The next study will address routes of PRECISE REPORTING, LLC jcp 223 1 abuse and usage and demand. 2 from Inflexxion, director of Epidemiology at Inflexxion 3 will present the next study. 4 And Ms. Theresa Cassidy Thank you, Theresa. 5 OxyContin Abuse Among Entrants to Substance Abuse 6 Treatment Programs 7 MS. CASSIDY: 8 Good afternoon. 9 Thank you, Dr. Coplan. I am Theresa Cassidy, director of Epidemiology at Inflexxion. Inflexxion is a 10 private public health research and technology company 11 with expertise in substance abuse research. 12 risk management and post-marketing surveillance services 13 to pharmaceutical companies through the NAVIPRO System. 14 I have no personal financial interest in the outcome of 15 this meeting, but I have been paid by Purdue for my 16 time. 17 We provide We will be conducting the study of abuse of 18 the reformulated OxyContin among adults entering 19 substance abuse treatment programs. 20 ASI-MV Connect data, we have over three years of 21 baseline on abuse of OxyContin. 22 baseline route of administration data for three drugs Through NAVIPPRO's This graph shows PRECISE REPORTING, LLC jcp 224 1 from the ASI-MV Connect Treatment Center Network since 2 2007. 3 variety of routes of administration are reported by 4 adults in treatment, with the most frequently being 5 snorting at 58 percent, followed by oral administration, 6 and then injection. For OxyContin, shown farthest to the left, a 7 In contrast, injection is reported most 8 frequently from morphine extended release products, 9 while snorting is a little bit lower in frequency. And 10 the most common route of abuse reported for hydrocodone 11 is oral administration at 92 percent. 12 Monitoring changes in the route of 13 administration profile will be a key element to 14 evaluating the question of tamper-resistance for the 15 reformulated OxyContin. 16 data just shown, specific route of administration 17 profiles exist for different prescription opioid 18 products. 19 As observed from the baseline From our data, we have observed that these 20 profiles can be used to characterize a drug’s pattern of 21 abuse because the profiles are distinct, they can be 22 differentiated from one product to another, and they PRECISE REPORTING, LLC jcp 225 1 tend to be stable over time. 2 For example, these baseline data show the 3 different routes of administration reported for 4 OxyContin by adults entering substance abuse treatment 5 are generally consistent and have been stable since 6 2007. 7 The objective of our study is to assess both 8 the route and frequency of abuse for the reformulated 9 OxyContin pre and post launch in comparison to other 10 prescription opioids. 11 adults entering treatment programs from a defined 12 network of centers using data from ASI-MV Connect. 13 observational surveillance study will measure recent or 14 past 30-day abuse and specific routes of administration 15 reported by abusers of OxyContin and other opioid 16 products. 17 The study population includes This The ASI-MV Connect has more than three years 18 of baseline data to use for comparison, and we 19 anticipate being able to observe a change within six to 20 nine months after the full introduction of the 21 reformulated OxyContin. 22 To provide a little background on our data PRECISE REPORTING, LLC jcp 226 1 source, the ASI-MV Connect collects information in near 2 real-time from a network of about 500 treatment programs 3 across the United States located in 36 states. 4 are collected using a computerized interview, which is a 5 standardized and validated instrument required during 6 clinical intake to treatment. 7 drugs that they’ve abused through pictures, drug names, 8 and street names, and the data are self-reported and 9 identified so that they are HIPAA-compliant. 10 The data Individuals identified This study will use statistical modeling 11 approaches that apply regression analysis to assess both 12 the frequency of abuse and the routes of administration 13 over time. 14 proportion of patients reporting past 30 day abuse of 15 OxyContin pre and post reformulation. 16 compare the difference in the continued rate of abuse by 17 measuring the number of days that a patient has reported 18 abuse within the past 30 days prior to treatment. 19 We will compare the difference in the We will also The modeling approach will take into 20 consideration adjustment for factors such as 21 prescription volume and geographic location. 22 of administration, we will compare the differences in PRECISE REPORTING, LLC For route jcp 227 1 the proportion of patients reporting abuse through 2 different routes with specific breakdowns for the oral 3 category that includes swallowing whole, chewing, and 4 other oral routes of administration. 5 One limitation to the study is that the sample 6 is now representative of all those entering substance 7 abuse treatment nationally, but rather are collected 8 from the 500 centers located in 36 different states. 9 Also, the ASI-MV Connect Network does not necessarily 10 11 collect data from individuals who do not seek treatment. The study does, however, exhibit strengths, 12 and these include the use of consistent measurements 13 over time that allow for reliable detection of 14 differences in the route of administration profile for 15 the reformulated OxyContin, and this among a sentinel 16 population that is at high risk of abuse for 17 prescription opioids. 18 Additionally, the online data collection 19 methodology that we use allows for timely analysis and 20 for prospective outcome monitoring with a high level of 21 specificity. 22 DR. COPLAN: Thank you, Ms. Cassidy. PRECISE REPORTING, LLC jcp 228 1 The next study, we’ll use surveys, in 2 particular, the NSDUH survey, and we’ll be looking at 3 routes of abuse, usage and demand, and addiction in the 4 study, and it’ll be presented by Dr. Howard Chilcoat, 5 director of Epidemiology at Purdue. 6 Using Surveys to Assess the Impact of a New Formulation 7 of OxyContin 8 DR. CHILCOAT: Thank you, Dr. Coplan. 9 Data from national surveys on drug use will be 10 a valuable tool to help us understand the impact of the 11 reformulation of OxyContin on non-medical use in the 12 U.S. population. 13 earlier in several presentations is the National Survey 14 on Drug Use and Health, or NSDUH, which is often used to 15 examine trends in drug use. 16 provide baseline data on non-medical OxyContin use. 17 One survey that you’ve seen presented We can use NSDUH data to As shown on this slide, these NSDUH data show 18 the percentage of the U.S. population that used 19 OxyContin non-medically from 2004 to 2009. 20 a slight increase in non-medical use of OxyContin during 21 this period. 22 data show an uptick to 0.7 percent from levels of around There’s been In particular, the recently-released 2009 PRECISE REPORTING, LLC jcp 229 1 .5 percent from 2004 to 2006. 2 the 2010 data become available to see if the increase in 3 2009 represents a trend. 4 happens to this trend after the introduction of the 5 reformulation. 6 We’ll have to wait for Then we will look to see what Our plan is to use data from NSDUH and three 7 other national cross-sectional surveys to compare trends 8 in the prevalence of non-medical use of OxyContin and 9 other prescription opioids. These surveys are measured 10 in a systematic way year by year and cover the vast 11 majority of the population. 12 assessed non-medical OxyContin use for up to six years 13 prior to the introduction of the reformulation, they 14 provide a useful baseline. 15 Because these surveys have However, due to the time needed to collect the 16 data each year and prepare datasets for public use, it 17 will be at least two years before we see an effect from 18 these studies. 19 We plan to use data from four different 20 national surveys, as outlined on this slide. 21 surveys include the NSDUH, as well as the Monitoring the 22 Future Study, Partnership Attitude Tracking Survey, and PRECISE REPORTING, LLC These jcp 230 1 2 the RADAR System College Survey. As you can see on this table, the studies have 3 different age ranges and sample sizes. 4 and RADARS enroll school-based samples ranging from 5 middle school to college. 6 today, I will focus on our strategy for using NSDUH data 7 because it covers the broadest scope of the U.S. 8 population and has more extensive measures of OxyContin 9 than other surveys. 10 The MTF, PATS, In the interest of time The NSDUH started in 1971, and has been 11 conducted annually since 1990 by the Substance Abuse and 12 Mental Health Services Administration. 13 over 60,000 respondents each year, and the survey 14 methods have not changed since 2002, allowing trend 15 comparisons for years since then. 16 collected data specific to OxyContin since 2004. It interviews The NSDUH has 17 Our strategy for using NSDUH data will be to 18 examine trends for several outcomes of non-medical use 19 of OxyContin and other prescription opioids. 20 look at the period prevalence, frequency of use, recency 21 of use, and the presence of DSM-IV dependence diagnosis. 22 We will depict prevalence used trends graphically and PRECISE REPORTING, LLC We will jcp 231 1 then we will compare trends across the time periods 2 before and after the introduction of the reformulation. 3 The most commonly used measure in surveys such 4 as NSDUH is period prevalence, which captures the 5 percentage of the U.S. population who use a drug in a 6 specified timeframe such as in the year prior to survey, 7 as I showed earlier. 8 9 The orange line represents the population of past year OxyContin users. However, it combines all 10 users from experimenters who have used the drug just 11 once orally all the way to daily injectors who are 12 addicted. 13 possible that the reformulation of OxyContin might have 14 varying effects on abuse and different populations, 15 depending on route of administration and stage of use. And as Dr. Coplan described earlier, it’s 16 To get a better understanding of the different 17 populations this orange line represents, we have divided 18 OxyContin users by frequency of use. 19 is associated with route of administration and stage of 20 drug use, and we expect that the formulation would have 21 a greater impact on more versus less-frequent users. 22 Frequency of use The blue line shows the trends for low PRECISE REPORTING, LLC jcp 232 1 frequency use, defined here as less than once a month. 2 And the yellow line shows a trend for high frequency 3 users, those who use monthly or more on average. 4 disaggregating by frequency of use, we begin to see 5 different trends emerge. 6 indicates that the overall increase in the overall 7 prevalence is accounted for by increases in frequent 8 use, whereas low frequency uses remain relatively 9 stable. 10 By Existing data through 2008 We’ll extend this analysis by using future 11 NSDUH data. 12 OxyContin use before and after the introduction of the 13 reformulation. 14 of the slide represent hypothesized changes following 15 the introduction of the reformulation. 16 whether there's an overall decrease in non-medical use, 17 and then explore whether the changes are specific to 18 frequent users or occasional users. 19 We will start by comparing trends in any The dotted lines on the right-hand side We’ll examine I’ve highlighted frequency of use in the 20 presentation, but we will look at several other 21 indicators of abuse. 22 of the reformulation is greater for persistent versus We will explore whether the impact PRECISE REPORTING, LLC jcp 233 1 recent onset use, as well as by history of prior other 2 drugs. 3 dependence changes among those using OxyContin non- 4 medically, and we can even possibly explore whether 5 there’s a switch to heroin once the reformulation is 6 available by assessing the occurrence of heroin use 7 among former OxyContin users. 8 9 We will see whether the occurrence of DSM-IV In this way, we can gain greater insight into the pathways through which the reformulation might 10 affect abuse at the level of the population. 11 understand it is possible that the changes that we 12 observe in OxyContin use after the introduction of the 13 reformulation could be caused by overall trends in non- 14 medical use of prescription opioids that are unrelated 15 to OxyContin. 16 parallel set of analyses for all prescription opioids to 17 see if the changes are specific to OxyContin. 18 We It will be necessary for us to do a As we looked at all the national surveys, we 19 noted that while they have a number of strengths that 20 will assist our works, they also have some limitations. 21 22 The limitations of large-scale survey data are well-known. It’s been discussed today. PRECISE REPORTING, LLC Retrospective jcp 234 1 self reports are subject to underreporting and under 2 reliability. 3 However, because these surveys use consistent 4 methods each year, it is unlikely that the changes in 5 trends would be due to differential underreporting over 6 time. 7 the surveys did not measure different routes of 8 administration of OxyContin. 9 data will allow us to look at frequency of use, as well With the exception of the RADARS College Survey, However, the available 10 as other indicators of abuse that might be related to 11 route of administration. 12 In addition, the surveys did not capture 13 certain populations. 14 earlier the NSDUH doesn’t include those in institutional 15 settings. 16 students who have dropped out or not attending school. 17 But we don’t expect this limitation to affect trends. 18 For instance, it’s been discussed The school-based surveys don’t include Among strengths, these surveys capture 19 patterns of non-medical use in the general population 20 rather than specialized samples, such as those entering 21 treatment. 22 dependence ever receive treatment, and, so, we need to Only about 10 percent of those with opioid PRECISE REPORTING, LLC jcp 235 1 2 go beyond this group. The surveys also ask OxyContin-specific 3 questions, which will be vital for our work. 4 give us a well-established baseline upon which to 5 compare trends following the introduction of the 6 reformulation. 7 formulation come in, we will analyze existing data to 8 better understand how OxyContin and other opioids are 9 used and develop a baseline against which we will 10 compare findings found in the introduction of the 11 reformulation. 12 It will So, until the data from the new So, although it may take awhile, we believe 13 that the data from national surveys will significantly 14 contribute to our understanding of the impact of a 15 formulation designed to make it more difficult to 16 manipulate OxyContin for the purpose of abuse. 17 Thank you. 18 DR. COPLAN: Thank you, Dr. Chilcoat. The 19 next study of law enforcement in the RADARS System, 20 it’ll be measuring demand for purposes of abuse, and 21 will be presented by Professor Rick Dart. 22 Law Enforcement Events in the Drug Diversion Program of PRECISE REPORTING, LLC jcp 236 1 RADARS System 2 DR. DART: Thank you. 3 The Drug Diversion System in RADARS is a 4 network of about 300 reporters in 50 states that are 5 either in local law enforcement agencies or in some 6 statewide taskforces that report each quarter about the 7 new cases in their area. 8 the investigator in that area submits a report using a 9 standard report tool into the database, and that's So, for new case of diversion, 10 reported out quarterly in terms of incidence rates for 11 that jurisdictional area. 12 This program is run by the Center for Drug and 13 Alcohol Studies from the University of Delaware, and it 14 currently covers 658 of the 3-digit zip codes in the 15 country, and there are about 960 or 70 of those. 16 about 68 percent of the total population, and for the 17 first 8 years, reported over 77,000 events of diversion. 18 So, this system has a lot of baseline data, as It’s 19 well. This is the same structure as my previous slide 20 which shows the incidence rate on the right per 100,000 21 population. 22 oxycodone, the orange line is OxyContin, and the yellow The purple line is immediate-release PRECISE REPORTING, LLC jcp 237 1 is the generic extended-release form of oxycodone. 2 As you can see, it’s a very similar trend in 3 both of the RADARS’ programs. We’ve experienced 4 dramatic increases in immediate-release oxycodone over 5 the last few years, and some increase in OxyContin, as 6 well. 7 number of people filling a prescription for those 8 medications. These in general have been quite related to the 9 What we will study here is the question will 10 diversion of OxyContin and comparators change after the 11 introduction of the new formulation? 12 unique; it’s drug dealers and diverters. 13 standardized surveillance from a long-established 14 network of centers, and the outcome really is the number 15 of new drug diversion events in that jurisdiction each 16 quarter. 17 think we will see an effect in the system in six to nine 18 months. 19 The population is It’s We have eight years of baseline data, and we Just to clarify a little bit about what a case 20 is, specifically, it involves a new written report 21 investigated during the prior three months. 22 be documented in the legal records and their arrest PRECISE REPORTING, LLC This has to jcp 238 1 records. And this is based on attempt or actual 2 diversion based on legal prescriptions, physician or 3 pharmacy reports of prescriptions, empty prescription 4 bottles, or actual drugs seized, such as in a buy. 5 The incidence rate for drug diversion events 6 is just simply the number of cases reported divided by 7 the population for that jurisdiction, and we add all 8 those together to get the total. 9 collects street price. This system also I hadn’t planned on including 10 that in the presentation, but if there's questions, I 11 can answer them during the Q and A. 12 For our analytical approach, this is very 13 similar to what I presented for poison centers, an 14 interrupted time series with low rates. 15 have to model that to allow correct inferences and we’re 16 going to include covariates, such as local prescription 17 availability and geographic location. 18 We’re going to This shows a slide of reported diversion 19 events from 2004. The data through mid-2010 is actual 20 data from the system, showing OxyContin, and it’s 95 21 percent confidence intervals. 22 right of the dotted line labeled “New Formulation” are The orange lines to the PRECISE REPORTING, LLC jcp 239 1 the potential outcomes here. 2 before, we’ll be looking for either no change or a 3 substantial decrease in the diversion of OxyContin. 4 And, as I described As I mentioned, we can also look for a change 5 in the price on the street of OxyContin. 6 decreases, we should see a decrease in the price. 7 If demand The primary limitation of law enforcement data 8 is potential for reporting bias, and intensity of 9 enforcement focus can vary somewhat. These 10 professionals are responding to the needs of their 11 community, and sometimes prescription drug abuse is 12 primary, and sometimes it isn’t. 13 from the baseline that we have long-term data over eight 14 years, and we don’t think that that will change 15 substantially, with over 300 investigators 16 simultaneously. 17 However, you can see Another point I want to make is that the data 18 don’t represent pain patients, and a previous speaker 19 mentioned this, but it’s really important to understand 20 that we may think they’re pain patients, but, in 21 reality, most of these are not pain patients. 22 dealers who are making a profit, entrepreneurs, if you PRECISE REPORTING, LLC They’re jcp 240 1 will, trying to make a profit. 2 The strength of our RADARS System is that we 3 create a mosaic by reporting from multiple stakeholders 4 and perspectives on the same phenomenon, and a positive 5 strength of the Drug Diversion System is that the 6 product is available usually for accurate identification 7 because they seized it in the arrest. 8 other systems, the data available within three months of 9 the close. And, as with our 10 Doctor Shopping for OxyContin as Measured by 11 Prescription Monitoring Programs 12 DR. COPLAN: The next study is a measure of 13 Doctor Shopping Prescription Monitoring Programs and 14 this is a measure of usage and demand. 15 Doctor shopping occurs when individuals visit 16 numerous physicians to obtain multiple prescriptions. 17 The excess drug can be abused or diverted. 18 Monitoring Programs were developed to track abuse and 19 diversion of prescription drugs at a state level. 20 Thirty-four states have operational PMPs. 21 recently been added. 22 correct. Prescription Delaware has The slide by Dr. Dormitzer was To date, two state PMPs have agreed to share PRECISE REPORTING, LLC jcp 241 1 data for analyses, Ohio and Connecticut. 2 been in discussions with other PMPs, such as 3 Massachusetts, Maine, North Carolina, and Utah to 4 participate in sharing data for an analysis. 5 We’ve also The objective of the study is to assess 6 whether the number of people who doctor shop for 7 OxyContin decreases with the new formulation. 8 population of people receiving opioid prescriptions in 9 the states covered by the PMPs. The study The design is an open 10 cohort study comparing changes in doctor shopping for 11 OxyContin and compared to opioids using data collected 12 by the PMPs. 13 the number of individuals who doctor shop. 14 And the outcome measure for the studies, The baseline data is approximately two to 15 three years. 16 PMPs, and we predict a time to see an effect of 17 approximately 12 months. 18 It has been collected in the existing There are two phases of the study. The first 19 phase of the study will develop and validate an 20 algorithm to measure doctor shopping by combining the 21 data elements in the PMP databases to measure doctor 22 shopping. And the second phase in the analysis will PRECISE REPORTING, LLC jcp 242 1 analyze changes in rates of doctor shopping using the 2 identified data. 3 The data elements that are available to detect 4 doctor shopping in the database are the number of 5 prescribers per time and the number of pharmacies per 6 time. 7 for example, by Dr. Ned Katz, to look at whether there’s 8 a change in the rate of measures of doctor shopping over 9 time. These have been used in the published literature, 10 Other researchers, primarily in Europe, in 11 France, have used overlapping dispensing periods of 12 repeated prescriptions or fills for opioids as a measure 13 of doctor shopping. 14 The measure of quantity and dose of 15 prescriptions can sometimes be indicative of a doctor 16 shopping. 17 because people who are doctor shopping tend not to use 18 insurance since the insurance can pick up the multiple 19 prescriptions. 20 benzodiazepine prescriptions, that also can be 21 indicative and would also be captured by the PMP. 22 will calculate a rate of doctor shopping, which is the In addition, cash payments can be indicative In addition, if the patient is receiving PRECISE REPORTING, LLC We jcp 243 1 number of doctor shoppers divided by either the number 2 of prescriptions or the Census population in that area. 3 One limitation of the study is that there is 4 no gold standard to measure doctor shopping. This could 5 lead to a high false positive rate of doctor shoppers. 6 However, the false positive rate should be relatively 7 consistent over time when measuring trends. 8 And also, this could be addressed by varying 9 the sensitivity and specificity of the doctor shopping 10 measures, as was indicated by Dr. Paulozzi in his 11 presentation and seeing how that changes the trends. 12 In addition, the study is somewhat limited by 13 the geographic coverage of the PMPs who participate in 14 the study. 15 provides an assessment of the desirability of OxyContin 16 for purposes of abuse, and it complements a study of 17 substance abuse treatment centers because the PMP study 18 population is not limited to those who seek treatment. 19 One strength of the study is that it The next study will look at Internet 20 discussions, and this is a measure of routes of abuse 21 and usage and demand. 22 Internet Discussion About Reformulated OxyContin Abuse PRECISE REPORTING, LLC jcp 244 1 MS. CASSIDY: Thank you. 2 This study also uses data from the NAVIPPRO 3 System to study Internet discussion about abuse of the 4 reformulated OxyContin. 5 A number of Web Sites exist that support 6 active discussion forums solely devoted to recreational 7 illicit drug use. 8 Sites freely offer their ideas and beliefs, discuss 9 trends and preferences, and they offer information and Drug users who frequent these Web 10 warnings about prescription medications. 11 provide suggestions and instructions in the form of 12 recipes for the physical and chemical extraction of 13 active ingredients. 14 on Web Sites for a discussion related to a number of 15 different topics, including routes of administration for 16 specific products, product comparisons, pill 17 identification, and methods for obtaining drugs for 18 illicit use. 19 Many of them We can examine these conversations Monitoring these Internet discussions allows 20 review of unfiltered opinions on various prescription 21 drugs among a sentinel population of abusers. 22 discussions provide early indications on whether abusers PRECISE REPORTING, LLC These jcp 245 1 attempt to tamper with a product for abuse by alternate 2 routes of administration. 3 discussions are available for wider dissemination among 4 non-participants who view the information, but do not 5 actively participate in these discussions. 6 And, more importantly, these The objective of this study is to assess 7 differences in the pattern of Internet discussion among 8 drug abusers regarding the new formulation of OxyContin. 9 The population for this study are drug abusers on the 10 Internet and the study uses an observational 11 surveillance approach. 12 Monitoring Data from a number of drug discussion Web 13 Sites, we will examine the proportion of discussion 14 related to conversations about tampering, routes of 15 administration, and overall sentiment by drug abusers 16 regarding the new formulation. 17 Using NAVIPPRO Internet This approach is currently being used by a 18 number of companies as part of their FDA post-marketing 19 surveillance requirements. 20 Internet discussion for the original OxyContin, and have 21 over three years of baseline data. 22 this type of surveillance, the data are available in a We have been monitoring Given the nature of PRECISE REPORTING, LLC jcp 246 1 quick timeframe, and we estimate that it would take 2 approximately three to six months after the introduction 3 of the new formulation to determine an effect of whether 4 there is a change among abusers in the nature of their 5 discussion in their interest in abusing this drug. 6 Although, it is not possible to quantify all 7 of the Internet discussion regarding the new 8 formulation, we are able to gain an understanding of the 9 types and the level of conversation occurring among drug 10 abusers within a stable community of individuals who are 11 on these selected Web Sites. 12 characterize differences in online discussion between 13 the old and the new formulation of OxyContin by 14 quantifying the proportion of posts, threads, and unique 15 authors that contribute to the conversation about the 16 drug. 17 This study is designed to One example of quantifying the level of 18 discussion is to calculate the proportion of posts 19 pertaining to a particular product over the total number 20 of posts on a selected Web Site. 21 approach to evaluate the number of individuals that 22 contribute to the conversation and to quantify the level We can use this PRECISE REPORTING, LLC jcp 247 1 2 of discussion for specific topics. To characterize Internet discussion by topic 3 area, we will obtain and review a random sample of drug- 4 specific posts. 5 by research staff using standardized coding procedures 6 to assess the sentiment of the message, including 7 whether the author endorses, discourages, or has mixed 8 comments about abusing the drug. The posts are then reviewed and rated 9 For example, if an author would post that they 10 enjoyed getting high from OxyContin, this would be coded 11 as endorsing the product for abuse. 12 if an individual references being addicted, which is 13 typically referenced in a negative connotation, or warns 14 against using OxyContin, this would coded as 15 discouraging the product for abuse. But, alternatively, 16 To ensure consistency between coders and over 17 time, inter-rater reliability is assessed using blinded 18 co-samples. 19 topic area including the routes of administration and 20 recipes for tampering with the drug. 21 22 And these methods are used to code posts by These baseline data shows sentiment in Internet discussion for three categories from a sample PRECISE REPORTING, LLC jcp 248 1 of posts for the original OxyContin and hydrocodone 2 products since 2007. 3 shown here in the orange bars, is more frequently 4 discussed in an endorsing manner among drug abusers 5 online at 38 percent of the sample of posts. 6 contrast, sentiment towards abuse of hydrocodone 7 products shown here in the blue bars is ambivalent, with 8 an equal percentage of posts endorsing and both 9 discouraging the product. 10 The data indicate that OxyContin, In We have observed some early Internet 11 discussion on the reformulation, and, in general, the 12 early discussion indicates dislike and frustration with 13 the reformulation by abusers. 14 here by the committee earlier if we are aware of any 15 information about the street price of the drug, and we 16 have seen some evidence of increases in the street price 17 of the original OxyContin, as shown in this example. 18 We’ve also seen intention by abusers to switch to other 19 opioid products preferred for abuse. 20 The question was asked Like any data source, there are certain 21 limitations to the study of Internet discussion, and the 22 nature and size of the Internet make it impossible to PRECISE REPORTING, LLC jcp 249 1 quantify and report on all discussions specific to the 2 new formulation. 3 data may relate to increase or decreases in population- 4 based trends of abuse is uncertain. 5 as the availability of a drug in any particular location 6 may also influence what is being abused there, and then 7 subsequently discussed online. 8 9 In addition, the extent to which these Other factors, such Study of Internet discussions, however, have a number of strengths. Analyses from the Internet 10 discussion can act as a rapid sentinel surveillance 11 system among sentinel population of abusers who are 12 motivated to tamper with the product for abuse. 13 Monitoring these discussions over time allows to detect 14 changes in how, why, who, and sometimes even where 15 diversion in formulation tampering can occur. 16 strength is that we have more than three years of 17 baseline data to use as a comparison and that the data 18 collection and analytic procedures that we are using are 19 consistent over time to allow for changing patterns. Another 20 DR. COPLAN: 21 The next study is abuser cohort in Kentucky. 22 Thank you. This study will follow-up individuals over time and will PRECISE REPORTING, LLC jcp 250 1 assess the measures of routes of abuse, demand for 2 reasons of abuse, and also the outcome measure of 3 addiction. 4 5 Professor Carl Leukefeld from the University of Kentucky will present. 6 Changes in Abuse Patterns in a Cohort of People Abusing 7 OxyContin in Rural Kentucky 8 MR. LEUKEFELD: Thank you. 9 I’d like to start by saying I have no personal 10 financial interest in the outcome of this meeting. 11 have been paid by Purdue for my time, and I have no 12 interest in the outcome of the study. 13 Our research team has been studying drug abuse 14 for prescription drug abuse as well as illegal drug 15 abuse in rural Kentucky for several years. 16 we have a good understanding of the baseline of 17 OxyContin abuse in this population. 18 I As a result, Let me fist give you some background on this 19 region and the cohort, and then I’ll discuss the study’s 20 objectives and methods. 21 22 Research indicates that prescription opioid abuse and dependence is more prevalent in some rural PRECISE REPORTING, LLC jcp 251 1 areas, thus, we believe we have been working with a 2 sentinel population for this type of research. 3 In addition, government and media reports have 4 pronounced that prescription opioid misuse is at 5 epidemic levels in the Appalachian regions of Kentucky, 6 Virginia, and West Virginia, where it is thought that 7 long and labor-intensive work, such as mining and 8 logging, has helped to create what has been called a 9 pain culture. 10 11 This supposition is supported by 2004 to 2008 national data. In our own 2007 study, more than 40 percent of 12 those indicating past 30-day prescription opioid abuse 13 that also injected during their lifetime, and our cohort 14 every OxyContin abuser reported injecting OxyContin, 15 which was surprising, and this is in the same study, 16 same county that we focused our current study on. 17 contrast, significant and a number of injectors who had 18 said they had Hepatitis C was significantly greater than 19 those who did not inject. 20 In Another study described the routes of 21 administration for prescription opioids. 22 previously studied these 101 opioid abusers in Perry PRECISE REPORTING, LLC We have jcp 252 1 County. 2 behavior across all drugs, and a high rate of injection 3 behavior with OxyContin specifically. 4 mind, I will now move on to the overview of the study we 5 are about to undertake. 6 What we found was a high rate of snorting With this in The objective of our study is to describe 7 changes in use and abuse patterns of OxyContin following 8 the introduction of the reformulation. 9 interview and follow 200 OxyContin abusers to examiner 10 self-reported changes in routes of administration and 11 preparation methods. 12 We will Let me say something about Perry County, 13 Kentucky. It has been popularized in the media. There 14 are about 30,000 folks who live there. 15 Hazard, of about 5,000 people, with a population that is 16 30 percent below the poverty line, and there are limited 17 economic opportunities there after the coal mining 18 industry collapsed. 19 qualitative and quantitative data. 20 who have been abusing OxyContin from a variety of 21 sources. 22 participants in an ongoing study of a current study of The main city is We’re going to collect both We’ll recruit people More than half will likely have been PRECISE REPORTING, LLC jcp 253 1 prescription opioid abusers in Perry County. 2 baseline enrollment to be completed by early next year. 3 We’ll also use qualitative extensive face-to-face 4 interviews with 15 randomly-selected participants about 5 the impact of the new formulation in their own drug use, 6 about the patterns of their drug use, and about what 7 drug use means to their family, friends, as well as 8 others in the community. 9 interviews will be about three to six months after the 10 11 We expect The follow-up structured baseline to assess any changes. These interviews will be conducted by a 12 trained interviewer, and will be quite detailed to gain 13 a picture of how these people abuse drugs. 14 Our structured interviews will assess what you 15 see here across and within individuals, preparation, 16 administration, abuse, opioid abuse. 17 addiction severity measures to look at symptoms and look 18 at symptoms of abuse and dependence, rates at which 19 people change their use of OxyContin, other prescription 20 opioids, and other drugs after the reformulation. 21 22 We’ll use the Our particular interest will be determining the extent to which OxyContin abusers switched to other PRECISE REPORTING, LLC jcp 254 1 opioids, such as IR oxycodone methadone tablets or 2 heroin. 3 limitations, and our findings will need to be considered 4 in light of these, including the single geographic area, 5 some limited generalized ability, reliance on self- 6 reports. 7 As with any study, this one has certain Along with these limitations are: the 8 extensive qualitative and quantitative data will provide 9 an opportunity to conduct exploratory data analysis, as 10 well as to apply more complex statistical approaches. 11 We believe we will be able to examine changes in the 12 abuse of other opioids as a sentinel population. 13 follow-up with individuals who are abusing OxyContin 14 when the reformulation is introduced provides a way to 15 directly assess changes in abuse behavior. 16 Our Finally, this study represents an 17 extraordinarily opportunity for us to examine the abuse 18 of opioids. 19 an abused drug might make a measurable impact in drug 20 abuse behaviors which we have not been able to do in our 21 studies for the last 20 years. 22 Thank you. We can now hypothesize at a reformulation PRECISE REPORTING, LLC jcp 255 1 DR. COPLAN: 2 Thank you, Professor Leukefeld. Summary and Conclusions 3 Taken together, the eight studies are designed 4 to make the most of the existing data sources to piece 5 together a clear picture of changes in patterns of 6 abuse. 7 address FDA’s request help to disaggregate the drivers 8 of abuse and its outcomes. 9 on studies routes of administration to assess the impact The mosaic of studies that we’ve assembled to We have a particular focus 10 of the new formulation, and we’ve attempted to 11 incorporate wherever possible different geographies and 12 populations. 13 impact, if any, of the reformulation in the real world. 14 The results will inform us and FDA of the Interpretation of the eight studies will be 15 focused on answering the five key questions that we 16 introduced earlier. 17 address each question. 18 estimate effects rather than test formal, statistical 19 hypothesis. 20 that is sustained and consistent across the studies. 21 22 There are one or two studies that Our goal is to describe and We’ll be looking for substantial effect The data collection has already begun within most of the data sources, and it’s estimated to take PRECISE REPORTING, LLC jcp 256 1 between 6 to 9 months and 24 months to see an effect. 2 These timeframes are our best guess. 3 diligent in continuing to monitor the effects in case a 4 way to circumvent the new formulation is developed after 5 the two-year horizon, or if additional cases are 6 required for study precision and power, particularly in 7 the Kaiser study. 8 9 We will remain The estimated time to see an effect can be used to determine a proposed duration of the studies for 10 a post-licensure commitment. 11 two years we’ll be able to see an effect. 12 select data, it make take two years to become available, 13 such as the NSDUH data, two years after the 14 observational period. 15 observational period is for two years to be initiated, 16 and we have initiated data collection beginning in 17 August 2010. 18 event rates are lower than expected to increase study 19 power. 20 It is anticipated within However, for Therefore, the proposed However, the duration may be lengthened if Purdue will submit annual reports to the FDA, 21 and investigators will independently report their 22 results of their studies. The overall interpretation of PRECISE REPORTING, LLC jcp 257 1 the results will consist of an internal assessment by 2 Purdue staff and an independent evaluation by the expert 3 panel that Dr. Landau mentioned in the introduction. 4 The Epidemiologic Study Program is designed to 5 address the five outcomes of interest using eight 6 studies, and this mosaic of studies will provide an 7 opportunity to address the impact of the new 8 formulation. 9 In conclusion, the new formulation is expected 10 to reduce injecting, snorting, and smoking routes of 11 administration of OxyContin by impeding the ease of 12 tampering with OxyContin. 13 associated with more frequent and independent abuse. 14 Multiple studies are required to demonstrate an effect 15 of the new formulation on various populations, stages of 16 abuse, and outcome measures. 17 studies to provide a comprehensive picture of the impact 18 of the new formulation. 19 These are the routes that are We’ve designed eight If the approach of the tamper-deterrent 20 formulation is demonstrated to be effective for 21 OxyContin, the approach may be generalizable to other 22 prescription opioids. Our goal today is to develop the PRECISE REPORTING, LLC jcp 258 1 best possible studies to address the questions about the 2 potential impact of the new formulation, and we welcome 3 the input of the panel in designing these studies. 4 5 Thank you for your attention to this rather long presentation. 6 DR. KIRSCH: Thank you. 7 We will now take a 15-minute break. Committee 8 members, please remember that there should be no 9 discussion of the meeting topic during the break amongst 10 yourselves or with any member of the audience. 11 resume at 3:15. 12 13 14 We will (Break.) Clarifying Questions DR. KIRSCH: All right, we're going to restart 15 the last part of the session for today. 16 session, if everyone could take their seats, for 17 clarifying questions to the Sponsor. 18 question, raise your hand, and we will recognize you. 19 Dr. Fletcher? 20 DR. FLETCHER: Thank you. And this is a So, if you have a I’d like to 21 congratulate the Sponsor and also FDA for kind of laying 22 out the various databases and the information about PRECISE REPORTING, LLC jcp 259 1 prior exposure and information about existing and 2 stability of some of the measures that are proposed 3 going forward prospectively to see changes in, but I 4 wondered if particularly the Sponsors the individual 5 speakers could address the issue of effect size in these 6 various databases and changes. 7 Being a clinician, I’m interested in that, and 8 while FDA’s questions don’t specifically address effect 9 size, they imply that that's an important aspect of it 10 in our deliberation tomorrow, and I’m wondering, for 11 example, given the baseline levels of data and 12 variability known, what kind of an effect size would be 13 able to be seen in the timeframe of the one to two years 14 for these various programs? 15 I'm particularly interested for Dr. Perrin 16 from the Kaiser Permanente, and, perhaps, Dr. Cassidy 17 and Inflexxion if they could comment about what size of 18 an effect in their measures could they be able to see a 19 difference in based on the information they have 20 already. 21 what’s clinically important difference for the group 22 tomorrow. Just to put in perspective the question about So, if anyone would want to address that. PRECISE REPORTING, LLC jcp 260 1 DR. COPLAN: Okay. We'll present our three 2 outcomes studies looking at the effect size of each of 3 those. 4 Study, please? 5 Could we have backup slide 45 for the Kaiser Forty-five. And, so, the power calculation for the Kaiser 6 addressed the effect size somewhat. We’ve tried to 7 avoid setting a mechanistic threshold above which we 8 would call an observed effect a success because we don’t 9 really have an evidence base on which to evaluate what 10 would be a priority, what would be a success. 11 tried to focus more on estimation, and look at the 12 precision to estimate trends or changes. 13 So, we’ve So, in this slide, to detect a 50 percent 14 reduction in the Kaiser Study that Dr. Perrin showed, we 15 would need 4,800 patients for 90 percent power or 3,600 16 patients for 80 percent power, which we’re easily able 17 to get, especially if we combined Kaiser Northwest with 18 Kaiser in Southern California and Kaiser in Northern 19 California. 20 patients a year if we combine those 6.6 million roughly 21 from a guess. 22 They’re approximately 6,000 to 7,000 Dr. Perrin, do you want to add anything? PRECISE REPORTING, LLC jcp 261 1 DR. PERRIN: (Off microphone.) 2 DR. COPLAN: Yes, so, to detect an effect size 3 of 75 percent, we would be very well powered. 4 below 50 percent, we’re really starting to run into an 5 inability to detect effect within one year. 6 was 40 percent, we probably would be okay with 80 7 percent power, but below that in one year, we wouldn’t 8 be able to detect. 9 then, obviously, we would be getting three times that 10 13 So, if it If we went out two or three years, amount of patients, and power would increase. 11 12 However, Could we have backup slide 54, please? Fifty- four. So, if we look, another system would be the 14 Poison Control Center, which is another way of measuring 15 outcomes, and, as Professor Dart showed, these are the 16 95 percent confidence intervals around the poison 17 control reports over time for OxyContin, so, it provides 18 an estimate of the precision. 19 and fairly tight, so, if we saw a 50 percent reduction 20 or a 30 percent reduction, I think that we’d have 21 adequate power to detect that. 22 The confidence intervals Professor Dart, did you want to add? PRECISE REPORTING, LLC I think jcp 262 1 the power calculation that you did, you had 80 percent 2 power to detect a 24 percent reduction. 3 4 Theresa, did you want to address the Inflexxion? 5 Did you have anything to add? 6 DR. DART: Sure. I just wanted to mention that I 7 think this question goes both ways because with a 8 number of events being reported, I think we’ll be able 9 to report a change, but whether that's a change that you 10 would recognize as important or the panel or society is 11 one of the issues I have is because we have so many 12 events that if we show a 5 percent decrease, but it’s 13 statistically significant, is that really a meaningful 14 change and what that conclusion might be from the 15 advisory committee. 16 DR. FLETCHER: Yes, I greatly appreciate that. 17 It’s not your role to necessarily say what are 18 clinically-important differences. 19 committee, because they’re the experts here. 20 DR. DART: 21 DR. FLETCHER: 22 high I just wanted the Right. To say what the power is to see what kind of a change, and then the deliberation might PRECISE REPORTING, LLC jcp 263 1 be informed tomorrow about what the size of these 2 effects are clinically-valid. 3 DR. DART: 4 DR. FLETCHER: 5 6 7 8 9 Sounds good. So, I completely agree with your conclusion there. DR. COPLAN: Also, could we have back up slide 65 to address the Inflexxion System? Sixty-five. So, one of the slides shown by Ms. Cassidy was looking at the baseline data of OxyContin route of 10 administration over time by various rates. 11 put the Confidence Intervals on this graph because it 12 would be too busy. 13 is approximately 7,000 patients who are reporting 14 OxyContin abuse. 15 ability to detect a statistically-significant difference 16 quite easily, but whether that would be clinically- 17 significant would be a different story. 18 We didn’t But they’re very small because this So, again, I think we would have the DR. FLETCHER: Thank you very much. I think, 19 to me, that's quite helpful for our discussion tomorrow 20 after we’ve heard all of the presentations. Thanks. 21 DR. KIRSCH: Dr. Walsh? 22 DR. KEETON: My question is for Dr. Landau. PRECISE REPORTING, LLC jcp 264 1 So, I’m just asking for a point of clarification. 2 the early part of your initial presentation, you said 3 that the company is not seeking any claims, and in 4 addition, on the slide, it says that any claims for 5 abuse liability should require substantial evidence to 6 support the claim. 7 never seeking any claims for this new formulation or is 8 it your hope that maybe these studies will yield the 9 substantial data that are needed to support a claim? 10 In So, is it your position that you are DR. LANDAU: So, we’d be very happy, 11 obviously, if the studies we proposed were able or 12 provided us a look and we’re able to detect a 13 significant change that we in the Agency would agree is 14 meaningful. 15 we’re not pursuing a claim, if on the other hand the 16 studies bear out and the Agency believes it’s in the 17 best interest of the public health to have this type of 18 information in a package insert, we’d certainly be 19 willing to have the discussion. 20 territory for us, and it’s not a path we’re ready to 21 pursue at this time. 22 It’s our current position, and not that DR. KIRSCH: This is unchartered I'm going to ask the next PRECISE REPORTING, LLC jcp 265 1 question, and it’s for Dr. Coplan and Dr. Perrin. 2 You’ve chosen to use the database from Kaiser, and one 3 of the weaknesses of that database, as was pointed out 4 by Dr. Perrin, is that it’s an insured population. 5 questions related to that. 6 Two There are many organizations now that have 7 Electronic Health Records similar or identical to the 8 one that's used at Kaiser. 9 which is not representative of most of the United Why go to their population 10 States? 11 system to pick up patients who are cared for primarily 12 in that system, but who go to other institutions when 13 they have an overdose or have a complication from the 14 treatment that's given to them at Kaiser? 15 And, second, what does one do within that DR. PERRIN: I think one of the reasons for 16 the selection of the Kaiser System is because we have 10 17 years of Electronic Medical Records to establish that 18 baseline. 19 that we actually work on with other studies do not have 20 a long enough baseline period with Electronic Medical 21 Records. 22 A lot of the newer community health centers It’s a more recent event for them. So, then I think to address your second PRECISE REPORTING, LLC jcp 266 1 question, Kaiser does record outside claims. 2 somebody goes to an emergency room that is not Kaiser- 3 run, those are registered into our databases if the 4 health plan is billed. 5 the differences from our initial looks at the data of 6 outside claims that come in, claims of poisonings from 7 outside of the system versus inside of the system. 8 that's a key variable that we’ll be looking at. 9 go outside the system and they ask for Kaiser not to be 10 So, if So, we can already see trends in So, If they billed, we will not be able to pick them up. 11 DR. KIRSCH: 12 DR. WOLFE: Dr. Wolfe? I have two questions. You asked 13 the second one, except I’ll just add a little piece for 14 Dr. Perrin again. 15 someone is found at home or someone goes to an emergency 16 room or whatever, how do you capture those deaths? 17 Deaths that occur outside the system, DR. PERRIN: We do have death data with codes, 18 Death ICD-10 codes, actually, in our system. 19 their primary care physician was the person who signed 20 the death certificate, they also register that into our- 21 -we have an Internal Death Database. 22 DR. WOLFE: So, if But that, again, assumes it PRECISE REPORTING, LLC jcp 267 1 happens within the system. 2 system, go somewhere other, not necessarily getting 3 billed, they just are found dead at home, go there, how 4 do you capture that? 5 DR. PERRIN: If they die outside the We only can capture those deaths 6 through the state death records, which we do check 7 regularly. 8 cause of death is easily ascertained on those records is 9 going to be difficult. Now, to be honest, the extent to which the 10 DR. WOLFE: 11 The other question was for Dr. Dart, and has 12 13 Right. It’s a problem. to do with his slide 53, if you could put that up. My memory is that during the period before the 14 generic prescribing stopped that a significant 15 proportion, 10, 20, 30 percent or more of all 16 prescriptions for oxycodone extended-release were 17 generic, and yet, this slide makes it appear that 18 consistently during this time, the last few dots are 19 probably with almost no prescriptions. 20 understandable, but consistently during this time, the 21 rate of intentional exposures per 100,000 population is 22 much less, much lower per 100,000 than with the PRECISE REPORTING, LLC That's jcp 268 1 OxyContin. 2 And the question is: Why is that? I mean, 3 I’ve heard previously, and it may have nothing to do 4 with this, that on the street or other places, the OC is 5 recognized as OxyContin, and people pick it up, they use 6 it, they sell it, whatever, that might not be true for 7 the generic, but I’m just really curious as to why there 8 are these huge differences between the generic extended- 9 release oxycodone and OxyContin, if you have any ideas 10 about that. 11 DR. DART: I don't have a lot of ideas. I 12 think yours is a reasonable one. 13 harder to identify. 14 orient and educate the poison centers, in other words, 15 new coming products, what they look like and that type 16 of thing, but because it didn’t have a clear indicia on 17 it, a lot of times, it’s harder to identify those. 18 Those products were We have a program where we actually So, I would say that that's part of it. Part 19 of it was they have somewhat lower sales, although, like 20 you said, they had substantial sales before they 21 started-- 22 DR. WOLFE: Yes, I think sometime in the last PRECISE REPORTING, LLC jcp 269 1 2 2 to 3 years, it was 20, 30, 40 percent. DR. DART: Yes, yes. It got up that high, and 3 I think the main reason is that. The other is that Dr. 4 Jim Inciardi, who has passed away, but worked with 5 RADARS, has pointed out many times about the brand 6 loyalty and has published some papers on brand loyalty, 7 and that's a very strong phenomenon among drug abusers. 8 It takes time for them to switch. 9 some work to show that it takes about three years for In fact, he even did 10 them to kind of get the idea that the generic is the 11 same as the branded product and switch to it. 12 think one of the problems there is that they never 13 actually totally became convinced that abusing the 14 generic form was as good as abusing OxyContin itself. 15 16 DR. WOLFE: So, I Maybe the OC or an OP should be removed from the pill so that it could back to the-- 17 DR. KIRSCH: 18 MR. WOLFE: 19 DR. COPLAN: Dr. Flick, next question. That's all. Thank you. Actually, could I just add a 20 quick note to that? We did look at the overdose rates 21 in the Kaiser System generic extended-release oxycodone 22 and branded OxyContin, and during the time period, that PRECISE REPORTING, LLC jcp 270 1 actually was in an earlier version on the slide deck. 2 We took it out in order to cut down time. 3 have it in the backup, but I can show you that at the 4 break. 5 event rate went up slightly at the time of the extended- 6 release oxycodone introduction. 7 DR. WOLFE: 8 DR. COPLAN: 9 And I don’t And what is shows is that the overdoes adverse The generic? The generic, yes. Yes. I’ll show you at the break. 10 DR. KIRSCH: 11 DR. FLICK: Dr. Flick? Just a couple of questions to 12 better understand the databases. 13 regard to the poison center data, Dr. Dart, these are 14 incidence rates per 100,000. 15 DR. DART: 16 DR. FLICK: 17 DR. DART: 19 DR. FLICK: 21 22 That's right. Now, the poison centers, you don’t have all poison centers participating. 18 20 On slide 53, with their coverage areas. DR. DART: That's correct. And poison centers overlap in Is that right? No. The state has to designate the coverage area for a poison center. So, each one will PRECISE REPORTING, LLC jcp 271 1 have a discreet coverage area. 2 DR. FLICK: 3 So, you clearly know the population? 4 DR. DART: 5 DR. FLICK: 6 question with regard to RADARS. 7 described as “incidence rates,” but based on I think it 8 was ZIP codes or jurisdictions-- 9 10 DR. DART: Okay. And on slide 83, a similar Again, these are Right, we match the jurisdiction of the investigator two or three-digit ZIP codes. 11 12 That's true. DR. FLICK: So, do jurisdictions match ZIP codes? 13 DR. DART: They don't match perfectly, so, we 14 have to proportionalize the data sometimes. 15 correct. 16 17 18 DR. FLICK: Okay. That’s And one brief last question. Ms. Perrin, I think the death certificate data 19 is critical to your results, but you don’t sound like 20 you have a very clear and absolute link to death 21 certificate data in your population. 22 DR. PERRIN: We can get access to the national PRECISE REPORTING, LLC jcp 272 1 death certificates and the state death certificates. 2 also have in our own database cause of death for those 3 who have died. 4 DR. FLICK: Well, but certainly there are a 5 significant number of these deaths that occur out of 6 hospital. Those deaths never enter a hospital. 7 DR. PERRIN: 8 DR. FLICK: 9 We Yes. And, so, they would only be captured in death certificate data, and I would think 10 that you would not answer by saying we can get access to 11 that. 12 be part of the study. 13 DR. PERRIN: Right, and that makes sense, yes. 14 DR. KIRSCH: Dr. Nelson? 15 DR. NELSON: Thank you. 16 I think you must have access to that, and it must I actually have two questions. 17 Like Rick Dart, I’m intimately involved with a 18 poison center at the New York City Poison Control 19 Center, and we don’t contribute. 20 percent of centers that don't actually contribute data 21 to RADARS. 22 that she discussed, and I think this is going to be a We’re 1 of the 20 But one of the things in Dr. Perrin’s slides PRECISE REPORTING, LLC jcp 273 1 semantic issue, are that for most of the day, we’ve been 2 talking about abuse and misuse, and she spent a lot of 3 time talking about adverse events and poisoning and 4 overdose. 5 about overdose, we usually think about suicidality or 6 some other intent, which I guess my question really is: 7 How do we reconcile that as you try to figure out the 8 data? 9 The problem with that is that when we think How do you integrate the terminology? DR. COPLAN: So, as Dr. Perrin showed in her 10 slide, the adverse events really refer to overdoses and 11 poisonings associated with prescription opioids. 12 Specific ICD-9 codes 965.0 and EA-50, which is very 13 specific for poisoning and overdose. 14 refer to adverse events, we’re really referring to those 15 specific ICD-9 codes. 16 DR. NELSON: So, yes, when we Right, but I guess my question is 17 when you look up an ICD-9 code for an overdose, you 18 might not be looking at an abuse or a misuse. 19 be looking at a suicidal patient, and when you look up 20 poisoning, it’s a very vague term. 21 usually generally means therapeutic misadventure, a 22 therapeutic problem. You might Adverse event PRECISE REPORTING, LLC jcp 274 1 DR. COPLAN: Yes, yes. 2 DR. NELSON: It could also mean a medical 3 error. 4 well. 5 misuse and abuse as a construct, I’m not sure how these 6 terms are going to easily equate to the terms that we’re 7 interested in hearing about. 8 9 I mean, these are terms that don’t match very And I guess since we’re really talking about DR. COPLAN: Well, we categorized our outcomes into measures of abuse, measures of routes of 10 administration, and changes in the overdose rates and 11 poisoning rates over time in essentially a cohort study 12 of the Kaiser membership population. 13 looking at is changes over time. 14 misclassification, that should be consistent over time, 15 and, therefore, the trends in OxyContin versus a trend 16 in comparator opioids should be reasonably meaningful. 17 So, what we’re So, if there is some I agree with you that if that was as 18 foolproof, as bulletproof as a randomized, clinical 19 trial, where you specifically look for these endpoints 20 then we wouldn’t have to do eight studies, we would just 21 do that one study. 22 DR. NELSON: Yes, I don't want to belabor it, PRECISE REPORTING, LLC jcp 275 1 but I guess my question is: Why don’t you look at 2 misuse and abuse as concepts within the KP data instead 3 of looking at these codes because they’re not really 4 reconcilable? 5 reconcilable. At least I don’t think they’re easily 6 My other question-- 7 DR. COPLAN: That is something we will DR. NELSON: Yes. 8 consider. 9 My other issue is, and I 10 want to second the comments about medical examiner data, 11 about death data because, I mean, clearly, one of the 12 big issues that we see with abuse and misuse as a very 13 defined endpoint is death. 14 and I’ll comment at another time about the use of 15 medical examiner data to define abuse and misuse because 16 there are a lot of issues with that. 17 I’ve commented in the past, One of my concerns is that the two systems 18 you’re going to use to look at death, one of them 19 involves the KP data. 20 center data, and as Rick Dart will elaborate on, if 21 you’d like, most deaths are not called into poison 22 centers. The other one involves the poison Right? PRECISE REPORTING, LLC jcp 276 1 I mean, when we compared our data to medical 2 examiner data, we have only a fraction of the poisoning- 3 related deaths. 4 overdose-related deaths occur outside of a hospital, 5 we’re going to miss the vast majority of overdoses. 6 without looking at some other database, meaning the 7 medical examiner database or some vital statistics 8 database, I think we’re going to miss a lot of deaths. 9 Just looking at the two that you have, as some people 10 And since many, if not most opioid So, have already pointed out, is going to be very limiting. 11 DR. COPLAN: I fully agree with you. That's 12 one of the reasons why we use this case fatality rate 13 top of metric in the Poison Control Center, which hasn’t 14 been used for an analytic type of study before, 15 precisely to address this concern because the case 16 fatality rate looks at the number of deaths per 17 exposures, and looking at whether it’s changed in the 18 time of fatalities per exposures as one way of getting 19 at that. 20 I don’t think it’s a perfect way. DR. NELSON: Yes, my strong recommendation 21 would be to look at medical examiner data. 22 limitations, it still will account for the majority of PRECISE REPORTING, LLC With all its jcp 277 1 the deaths, which I think you’ll miss in this system. 2 3 DR. COPLAN: this? Yes. Nab, do you want to address We’re calling on someone from the other bullpen. 4 (Laughter.) 5 MR. DASGUPTA: My name is Nabarun Dasgupta. I 6 am with the University of North Carolina and Chapel 7 Hill, and I have no financial interest in this meeting, 8 but my way here has been paid for King. 9 The reason I’m up here is that we’ve been in 10 talks with Paul and the other folks at Purdue to do 11 exactly the study that you proposed, Dr. Nelson, and 12 what Dr. Paulozzi has proposed before, where we can link 13 the Prescription Monitoring Program to the medical 14 examiner data. 15 earlier this year to allow for that linkage to happen. 16 The State Health Department has done a pilot study doing 17 the linkages between the Prescription Monitoring Program 18 and the medical examiner data in three counties. 19 have a methodology for it, and we have agreement within 20 that state health government structure to go forward 21 with it. 22 We passed legislation in our state We That study is not presented here because it’s PRECISE REPORTING, LLC jcp 278 1 not formalized and hasn’t been advanced enough at this 2 point to warrant full scientific scrutiny, but it is 3 something that we are looking at and have some 4 experience with, and are confident that we can do. 5 DR. COPLAN: So, essentially, what we’re 6 planning to do is a repetition of the study Aaron Hall 7 and colleagues from the CDC that looked at linking state 8 medical coroner’s reports for deaths in West Virginia in 9 the year of 2006 with Prescription Monitoring Programs 10 and state toxicology reports, and we would look at that 11 by looking at changes over time. 12 As Nab pointed out, we haven't presented that 13 yet today. We did actually mention it when first 14 submitted something to the FDA that that was a study 15 that we would like to do, but because of the feasibility 16 of integrating multiple state agencies and third-party 17 groups to do the statistical analysis, it would take 18 some time, particularly when working as a for-profit 19 company, a drug sponsor trying to get this study to 20 happen, it would take some careful negotiations to bring 21 together various state parties. 22 that we are actively pursuing, and we do seem to have an But that is something PRECISE REPORTING, LLC jcp 279 1 avenue that's opening for that. 2 DR. KIRSCH: Before I call on the next person, 3 I want to remind the members of the committee we’re 4 calling on people in order from when they raised their 5 hand. 6 We’ll get to you. So, if you think we’re ignoring you, we’re not. 7 For the FDA, I actually have a question. With 8 a number of people who we have on the list to ask 9 questions, I think it’s very likely we’ll run over our 10 4:00 time, which, for me, as long as I get done by 9:00 11 to watch Oregon beat UCLA, I’ll be okay. 12 (Laughter.) 13 DR. KIRSCH: 14 will that be a problem? 15 16 But if we run a little bit over, DR. RAPPAPORT: Well, I have a show on earlier than that. 17 (Laughter.) 18 DR. RAPPAPORT: 19 discretion when to end the meeting. 20 DR. KIRSCH: 21 Ms. Krivacic? 22 MS. KRIVACIC: But I think it’s your Okay. Thank you. I have a couple of PRECISE REPORTING, LLC jcp 280 1 short questions. One is for Chilcoat. 2 there was reference made to the materials that we 3 receive about the PRIDE Survey, the acronym PRIDE, and 4 if that is part of the surveys that will be utilized 5 here. 6 DR. COPLAN: 7 DR. CHILCOAT: The question is Dr. Chilcoat? No, it's not. It actually only 8 grants one question about OxyContin. So, I think a lot 9 of the prevalence for that study is much higher than 10 what you see in other surveys, and, so, we decided not 11 to use it, even though we originally considered it. 12 MS. KRIVACIC: Thank you. My second question 13 refers back to slide 56, the case fatality rate, and I 14 was wondering if Dr. Dart can speak to teasing that data 15 out by way of age range, adolescence. 16 that you have access to? 17 it out by even socioeconomic status? 18 Thank you. 19 DR. DART: Is that something And then, also, can you tease That's a yes and a no. So, for 20 age, we can bring it out really of any single year you 21 want, any age range grouping you want, that can be done. 22 So, if we want to look at adolescence for new initiates PRECISE REPORTING, LLC jcp 281 1 or into your 20s, we can do that. But when it comes to 2 socioeconomic status, we don’t gather that information. 3 The reason is that this is an acute health care event. 4 So, the real reason of a poison center, of course, is to 5 give some advice to the caller or to the health care 6 provider who’s calling, and, so, we don’t ask a lot of 7 demographic questions since we’re basically taking care 8 of the patient. So, yes. 9 DR. KIRSCH: Dr. Zelterman? 10 DR. COPLAN: I’m sorry, could we have backup 11 12 55? Just to add to that response, if you wouldn’t mind. This is data broken down by, as Professor Dart 13 mentioned, this was somewhat arbitrary. 14 down into less than 12 and greater than 12 to see if we 15 could detect an effect specifically in the pediatric 16 population with the assumption that if a new formulation 17 didn’t allow the extended-release mechanism to be so 18 easily broken by a kid, an infant or a child, they would 19 have more time to get into emergency department and get 20 a shot of naloxone and be able to save the kid. 21 unfortunately, there are only three deaths of--not 22 unfortunately. Sorry. PRECISE REPORTING, LLC We broke it But, jcp 282 1 (Laughter.) 2 DR. COPLAN: For the purpose of this study, 3 there are rather few deaths to be able to make any 4 determination of that. 5 DR. KIRSCH: 6 DR. ZELTERMAN: 7 A comment about the Internet study and then the Kaiser Permanente Study. 8 9 Dr. Zelterman? My mother told me not to believe what I read, and that's especially true of the Internet. 10 (Laughter.) 11 DR. ZELTERMAN: If you really want it, I mean, 12 by Monday morning, we could have 100,000 posts of what 13 you can do with OxyContin that defy your imagination. 14 What you can get out of the Internet is if somebody 15 really finds a way using common household methods of 16 extracting the slow release, and that's what you can 17 get. 18 going to teach you that is if somebody figures that out. That's, I think, the only thing the Internet is 19 As for the Kaiser Permanente, could I see 20 slide 42? Slide 42, these are the comparisons that are 21 going to be made, and Dr. Kirsch already pointed out 22 that the Kaiser Permanente data is not a random sample PRECISE REPORTING, LLC jcp 283 1 from the population; it’s a very biased sample. 2 subscribers of this health insurance. 3 whether over a period of time the patients are changing, 4 the coverage is changing. 5 It’s It’s not clear It’s not clear. And then if I can go to slide 45, while you 6 address that. On slide 45, Dr. Perrin commented that 7 there are many different regions and different groups 8 that could be included, and if you want to get the point 9 0.05 significance in front of the FDA, I think you 10 usually have to specify your population before rather 11 than hunt around and find the ones that support your 12 hypothesis. 13 14 DR. PERRIN: first issue. Okay, so, we'll start with the So, which slide were we on? 15 DR. COPLAN: Slide 42. 16 DR. PERRIN: Forty-two. 17 DR. COPLAN: Can you go back to 42? 18 DR. PERRIN: So, yes, there are changes in 19 membership over time, and there are also changes in 20 prescribing patterns over time. 21 that using a time series design with a comparator is so 22 important. And that's why we feel And, also, looking at this in different PRECISE REPORTING, LLC jcp 284 1 subgroups where we know that there have been some 2 changes. 3 We can also add into our time series, and I 4 didn’t go into this in detail, but we can add time 5 varying covariates into the model that will adjust for 6 these different trends over time, if necessary. 7 And, so, for the second one, I promise you I 8 won't hunt around for which regions are the right 9 regions, but what we do plan to do is, based on the size 10 of the region, figure out how many more regions we need 11 to bring in. 12 the other regions, we’re only looking at the size of the 13 membership, and then we would begin to figure out which 14 ones to collaborate with and share our codes. 15 extracting records in exactly the same way. So, we are not looking at data from any of 16 DR. KIRSCH: 17 DR. MORRATO: So, we’re Dr. Morrato? Yes, my question relates to your 18 interest in demonstrating sustainability and how you’re 19 defining sustainability and duration of observation. 20 So, I understand and appreciate the rationale for the 21 time series, but it wasn’t clear to me always what was 22 the unit of time that's being done in some of the PRECISE REPORTING, LLC jcp 285 1 studies and how that might differ. 2 many points you’re actually looking at when you’re 3 assessing trends. 4 study, it looked like there’s annual rates based on what 5 I saw, so, that would give you two time points. 6 NSDUH survey, again, I think is annual. 7 two time points, and it’s a two-year lag for the data. 8 So, it’s actually four years out for that. 9 that's one question. 10 And, therefore, how So, for instance, in the Kaiser The That gives you and, so, And related to that is, on the other hand, you 11 have some opportunity to get very discreet data, let’s 12 say the Internet discussions or the abuser cohort in 13 Kentucky, and you only carry those studies out for six 14 to nine months. 15 to be an evolving market in which what you learn and 16 adapt to in the first few months is going to be very 17 different than how people might adapt a year later. 18 I don't know how that’s supporting sustainability. 19 20 One might anticipate that that's going DR. COPLAN: discussed a lot. So, Yes, that's something we’ve It’s obviously a key issue. 21 Could I have back up slide 35? 22 So, this is the Kaiser data. PRECISE REPORTING, LLC So, one of the jcp 286 1 questions was: What are the units? Is this annual? 2 This is actually in a six-month period. 3 years, we would have four periods. 4 we’ve been seeing on the Internet, there’s a real 5 aversion to the new formulation. 6 of tampering being posted by abusers on the Web Site, 7 but, as Dr. Zelterman referred to, they are not 8 widespread. So, in two Some of the quotes We have seen methods 9 So, definitely, we’re going to see very 10 determined abusers finding ways to get around the 11 formulation. 12 abusers say wow, this works for me, this is great, we 13 found a way, and we haven't see that. 14 well, I tried it and I spent two hours heating and 15 freezing and I didn’t really get much of a high. 16 The question is whether 50 or 60 other Other people say So, based on that, we would expect to see a 17 relatively quick change. And then the sustainability 18 then becomes what do we see over the four halves of the 19 year? 20 that would be evidence of sustainability. 21 mean, as we mentioned, that we would no longer monitor. 22 We have been monitoring adverse event rates for As we mentioned, if it’s a very clear trend, then PRECISE REPORTING, LLC That does jcp 287 1 OxyContin since 2002, created a system to do that since 2 the database didn’t exist to do that, and we’re 3 continuing to evolve new systems of surveillance that 4 use more complex geographic information systems to do 5 that, which we haven't discussed today. 6 7 We also have some collaborating evidence. Could I have slide 573? 8 9 10 We have some collaborating evidence of changes in prescribing. That suggest we would see a relatively quick effect. 11 This is some data looking at change in 12 prescriptions of the new formulation and generic ER 13 oxycodone in the past eight weeks in health care 14 providers with questionable prescribing or medical 15 practices. 16 through various sources of information, primarily the 17 field sales force, who has a very good handle on which 18 prescribers are problematic. 19 those prescribers to make sure that we do not call on 20 them. 21 22 Purdue keeps a database of prescribers who So, we keep a list of The field sales force does not them. We’ve been tracking using the SDI data that has been referred to earlier to look at changes in PRECISE REPORTING, LLC jcp 288 1 prescriptions amongst these health care providers, and 2 we have some very preliminary data. 3 be further worked out, but for OxyContin, in the current 4 4 weeks, there were 10,700 prescriptions compared to the 5 previous 4 weeks of 16,000 prescriptions for a reduction 6 of 5,000 prescription or minus 34 percent. 7 I think it needs to For ER generic oxycodone, they were smaller in 8 absolute numbers, but there was a increase in 9 prescriptions in the 8-week period for a net increase of 10 24 percent. And for the total ER oxycodone brand and 11 generic, we’ve seen a reduction of 27 percent. 12 this, obviously, has many flaws in it. Now 13 We need to compare these with 1,300 14 prescribers on the do-not-call list, we need to look at 15 comparing prescribers who are not on the do-not-call 16 list, but this suggests that we’re likely to see a 17 relatively quick effect, and that four subsequent 18 measures would be reasonable. 19 perfectly willing to continue to survey this as long as 20 it’s needed. 21 DR. KIRSCH: 22 DR. OMOIGUI: If it’s not, we are Dr. Omoigui? I am wondering, bring back the PRECISE REPORTING, LLC jcp 289 1 slide 35 again. 2 opioids. 3 And ask if the others can do two Were those short-acting, long-acting, or both? DR. COPLAN: That's a mixture. The hierarchy 4 in determining this was first OxyContin or ER oxycodone, 5 then other oxycodone, which would be the immediate- 6 release oxycodone single and combination, and then other 7 Schedule IIs would be hydrocodone, methadone, fentanyl 8 patch. 9 PARTICIPANT: 10 DR. COPLAN: 11 12 you. Not hydrocodone. Sorry, not hydrocodone. Hydrocodone would be Schedule III. DR. OMOIGUI: Thank So, yes. Okay, if we look again at this 13 slide 53 and 83, it looks like in the last few years 14 there’s an increasing trend in the greater abuse and 15 diversion of the immediate-release oxycodone as compared 16 to the OxyContin, and I believe during those few years 17 was when we’ve had increased dose trends of some of the 18 immediate-release oxycodone. 19 out in the last few years. 20 I think the 30 mgs came So, the question is this then: Are we already 21 seeing a trend away from OxyContin into the immediate- 22 release oxycodone, and if we are doing that, are we PRECISE REPORTING, LLC jcp 290 1 going to be proactive in the fact that if this new 2 reformulated OxyContin is successful, you’re going to 3 see a shift, an even greater shift into the immediate- 4 release because the drug abuse problem is not going to 5 go away quietly into the night. 6 It’s going to try and shift, and we have to be 7 proactive in checking that out. 8 of your studies, you are assessing the impact of the new 9 formulation on the abuse of the immediate-release. 10 And I noticed that some Is there a way you can do that in all your 11 studies so that that way if you’re seeing a reduction in 12 the abuse of the OxyContin you can tie it into any 13 changes in abuse of the immediate-release formulations? 14 DR. COPLAN: Good, thank you for that 15 question. 16 will look at the data to estimate is there a shift 17 occurring, and, secondly, Dr. Landau will address the 18 issue of a potential shifting. 19 20 21 22 So, we’ll look at it in two ways. First, we Nelson, could I have backup slide 30? Thank you. So, as mentioned earlier, even before the new formulation of OxyContin came out. This is data going PRECISE REPORTING, LLC jcp 291 1 up to 2009, and this is data that was presented by the 2 FDA at the last adcom that this group had. 3 at the single ingredient oxycodone, it has increased 660 4 percent, whereas extended-release oxycodone has 5 increased by 40 percent in the last 10 years. 6 have been seeing a dramatic shift already occurring 7 independent of any new formulation of extended-release 8 oxycodone. 9 If you look So, we There was also a discussion at the last 10 advisory committee around what was the predominant drug 11 that was being dispensed in the Florida pill mills 12 which, as we all know, is one of the worst sources of 13 opioids for the purposes of abuse. 14 well, it was OxyContin that was the most prescribed in 15 the Florida pill mills. 16 board of directors of the Florida pill mills said no, 17 it’s shifted. 18 ingredients oxycodone, and we don’t actually have data 19 on that. 20 And some people said And then someone who’s on the It’s now immediate-release single We’re trying to look to see if we can actually 21 get data on that, but, based on that discussion, there 22 has been some evidence of the Florida pill mills PRECISE REPORTING, LLC jcp 292 1 reducing their prescription. 2 prescribing OxyContin, but there’s a reduction in the 3 overall patent, and that's largely because the DEA was 4 using certain prescribing metrics to determine who to 5 arrest. 6 Not that they stopped And so, if you look back in Google, you can 7 get the information of which providers were arrested by 8 DEA. 9 flag was that they were prescribing most 60 and 80 mg Initially, they were arrested because the major 10 OxyContin and for cash payments. 11 people got arrested for that, then people adapted. 12 So, once a couple of So, we’re already seeing a shift occurring, 13 and now we’ll talk about the second part of the problem 14 is how we do address the public health impact of this 15 formulation potentially adding to that shift? 16 DR. LANDAU: Thank you Paul. I would only add 17 that it’s an unfortunate reality, but it is our 18 expectation that if we’re successful with the 19 formulation, that abusers will shift either to other 20 routes of abuse that are more practical to them or to 21 other drugs. 22 already in close to real time through Internet We’ve seen evidence that this is occurring PRECISE REPORTING, LLC jcp 293 1 monitoring. We expect to see a significant reduction in 2 intravenous abuse and intranasal abuse. 3 speaks to the complexity and the limited role one 4 pharmaceutical company can play in a multi-factorial 5 problem. And I think it’s an excellent question. 6 Thank you. 7 DR. COPLAN: 8 9 10 I think it If I could add one thing, as we mentioned-DR. KIRSCH: Well, if it’s critical. We are out of time. 11 DR. COPLAN: Sorry. 12 DR. KIRSCH: So, I’m going to go through a 13 couple more of these. 14 Dr. Bickel? 15 DR. BICKEL: First, I want to commend the FDA 16 and the Sponsor for presenting very interesting and 17 important sets of presentations today. 18 For the Sponsor, and I guess Dr. Landau, I’d 19 like to understand what the company’s response would be 20 under two sets of circumstances. 21 studies produced results that are inconsistent, or, 22 alternatively, the set of eight studies suggests that One, the set of eight PRECISE REPORTING, LLC jcp 294 1 the problem is actually getting worse or staying the 2 same? 3 DR. LANDAU: Well, in either scenario, we’d be 4 interpreting and discussing the results of the studies, 5 as Dr. Coplan mentioned, on a periodic basis, reviewing 6 them internally with our expert panel, and sharing the 7 results with the Agency. 8 would respond, given the complex nature of the problem 9 and the complex interrelationship with some of the It’s hard to predict how we 10 behaviors and the outcomes these studies are intended to 11 measure. 12 What I can tell you is that we’re very 13 interested in measuring and monitoring, and we have been 14 and will continue to be very proactive in our actions, 15 and they're be appropriate and shared with the 16 regulators. 17 Thank you. 18 DR. COPLAN: We think that the cornerstone of 19 determining whether these studies are demonstrating an 20 effect will ultimately be expert judgment that will 21 involve clinical, statistical, and epidemiological 22 expertise. PRECISE REPORTING, LLC jcp 295 1 DR. KIRSCH: Dr. Morris-Kukoski? 2 DR. MORRIS-KUKOSKI: I have two questions. 3 One is since you said you’re already collecting data as 4 of August, but you realize that the market is still 5 going to continue to have the old formulation of the 6 OxyContin through the first of the year. 7 going to then shift your data that you’re actually 8 looking at so you actually have this slight overlap so 9 you can continue past that period of time when we know 10 the regular OxyContin or the old formula is still out 11 there? 12 Are you not That's my first question. DR. COPLAN: We thought a lot about how to do 13 that because that is a key issue. 14 determining a pre-post change between two incidence 15 rates, we would need to take that mixed time out, 16 perhaps that one quarter. 17 over time becomes important for us to include. 18 For trends, for For trends, that reduction DR. MORRIS-KUKOSKI: And my second question 19 is: Is you definition for more difficult to manipulate 20 only crushing with spoons and dissolving with water? 21 22 DR. LANDAU: moment. I'm going to call up a slide in a The short answer to the question is no. PRECISE REPORTING, LLC In jcp 296 1 March of 2009, we submitted along with our resubmission 2 to a complete response letter to the division the 3 results from seven separate and comprehensive in vitro 4 studies, and I mentioned earlier very briefly they were 5 designed with the assistance of experts and abuse in 6 tablet tampering and extraction techniques and even drug 7 enforcement. 8 things, and the interpretation goes well beyond more 9 difficult to crush and inject. The results tell us a great number of 10 May I have slide-- 11 DR. KIRSCH: 12 Is the slide different than what you’ve just said? 13 DR. LANDAU: Slide 481, please. Yes. 14 Okay, so, shown here, the experiments 15 replicated proceed replicated techniques of tablet 16 tampering that are relevant both to the abuse and the 17 patient error context, and I don't have a pointer here, 18 but under "Route," what’s common to each one of these 19 circumstances or settings is, in many cases, all, with 20 the exception, frankly, of swallowing intact tablets, is 21 some degree of physical or chemical manipulation. 22 the seven separate experiments were designed, as PRECISE REPORTING, LLC So, jcp 297 1 represented here, to inform a prediction for how 2 difficult or what incremental change would exist for 3 this formulation relative to the original formulation in 4 both the patient and the abuser setting. 5 On the right margin, you see are sort of the 6 results broadly characterized for a public setting like 7 this, and in each scenario of testing or each access of 8 testing, the new formulation, the reformulation was, 9 well, in most all, more robust or more difficult to 10 manipulate or convert to a dose form that was necessary 11 to abuse via one or more of these routes and never 12 worse. 13 DR. KIRSCH: 14 Dr. Mendelson? 15 DR. MENDELSON: 16 Thank you. Yes, hi. Just a couple of quick points. 17 First, the one thing that does seem to be 18 missing is the economic data again, and I think you 19 could collect this through Kaiser. 20 pay as a co-payment for their prescriptions? 21 that would be very useful information. 22 twice and much to get one and people prefer that, that's How much do people PRECISE REPORTING, LLC I think If it costs jcp 298 1 actual news, that it’s the Dr. Bickel’s behavioral 2 economics. 3 Second, I think it’s ironic that the FDA 4 requires suicide assessments for almost all new drug 5 evaluations right now, yet, suicide is not explicitly 6 parsed out or separated in this analysis. 7 guys have an inconsistency with your other drug programs 8 by not making suicide explicitly separated in the 9 overdose data. I think you And if a lot of these overdose deaths 10 are suicides either intentional with drug intoxication 11 or during withdrawal because they can’t obtain the drug, 12 that would be big news. 13 And finally, I would note that Purdue does 14 make immediate-release oxycodone. 15 of that. 16 so, you guys might be back here with more trouble in the 17 future. 18 ask what’s going on with your IR oxycodone products as 19 they move up the ladder of acceptability and 20 preferability in addicts. 21 22 It sells two brands We just looked it up on Epocrates here. And, You may want to address that now and actually DR. LANDAU: Sure. I’d like to address I guess the latter part of your series of questions, the PRECISE REPORTING, LLC jcp 299 1 last one. We no longer manufacturer and market 2 immediate release oxycodone, just for a point of 3 clarity. 4 DR. KIRSCH: Thank you. 5 Could you pull up slide 119, please? 6 DR. COPLAN: We had some data on prospect. 7 DR. KIRSCH: The question that I have about 8 119 is it was unclear to me how the data is going to be 9 delivered to the FDA. Will it be delivered from these 10 individual studies after being filtered through the 11 company or will they investigate individual groups doing 12 the investigation, report directly to the FDA? 13 DR. COPLAN: The nature of FDA’s reporting 14 mechanism is that the sponsors were responsible for 15 providing an annual report on whatever timeframe the FDA 16 deems is appropriate. 17 Generally, it’s annually. What we have stated here is that we would 18 include a PDF that's obtained from the investigators of 19 the study, Dr. Perrin, Dr. Dart, Dr. Cassidy, and submit 20 those as part of the report with an overall integration 21 by the expert panel and by Purdue. 22 DR. KIRSCH: It’s two separate. That is-- PRECISE REPORTING, LLC jcp 300 1 DR. RAPPAPORT: 2 DR. KIRSCH: 3 DR. RAPPAPORT: Jeff? Yes. Can I clarify something about 4 that? I mean, everything that comes into the agency 5 comes in through the sponsors. 6 citizen’s petitions and things like that, but in regard 7 to an application, it comes from a sponsor. 8 a long history of investigating data integrity, and we 9 get all of the data and the raw data, as well. Occasionally, we do get But we have So, we 10 look at that very carefully to make sure that what 11 they’ve synopsized for us is consistent with the raw 12 data and we go out and investigate their sites and all 13 of that. 14 DR. KIRSCH: Thank you. 15 Dr. Denisco? 16 MR. DENISCO: 17 DR. KIRSCH: 18 Dr. Flick? 19 DR. FLICK: 20 DR. KIRSCH: 21 DR. KERNS: 22 DR. KIRSCH: Asked and answered. Asked and answered. That was my question. Dr. Kerns? Last question. I’ll wait until tomorrow. Okay. With that, we’ll adjourn PRECISE REPORTING, LLC jcp 301 1 the meeting. Thanks, everybody for their attention. 2 DR. COPLAN: 3 (Whereupon, at 4:07 p.m., the meeting was 4 Thank you. adjourned.) 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 PRECISE REPORTING, LLC