Breast Recoding Audit Final Report Supplement A High Quality Abstractors Background Due to an increased backlog of cases, a decision was made by the CCR to “push through” cases that came from high quality abstractors. Abstractors who were considered to be high quality abstractors had to have an accuracy rate of 99% and above. For the purposes of this report, they will be referred to as HQA’s. This practice went into effect statewide in December 2005. Audit Results During the reconciliation process, the regions were asked to identify which of the cases were from an HQA and whether they agreed with the discrepancy or discrepancies made by that abstractor. The table below (Table 1) demonstrates the proportion of discrepancies from HQA’s compared to those from other abstractors. There were a total of 54 discrepancies (33.7%) from HQA’s. These discrepancies occurred in 33 cases. Discrepancies: High Quality Abstractors vs All Other Abstractors High Quality Abstractors 33.7% All Others 66.3% Table 1 Kyle Ziegler Page 1 12/30/2015 Macintosh HD:Users:peter:Desktop:PR LIGHT:CDPH to CPIC Et al:CDPH AUDIT 1\29\15:High Quality Abstractors Supplement 20808.doc Breast Recoding Audit Final Report Supplement A High Quality Abstractors Region 3 had the least number of discrepancies from HQA’s with 2 (3.7%). Region 8 had the highest number, with 20 (37%) discrepancies. Region 2 had no discrepancies from HQA’s. Table 2 illustrates the distribution of the discrepancies across the regional registries. Number of Discrepancies by Region - HQA's Only 20 20 18 16 14 12 10 8 8 6 5 5 6 4 2 8 2 0 0 Region 2 Region 3 Region 4 Region 5 Region 6 Region 8 Region 9 Region 10 Table 2 Kyle Ziegler Page 2 12/30/2015 Macintosh HD:Users:peter:Desktop:PR LIGHT:CDPH to CPIC Et al:CDPH AUDIT 1\29\15:High Quality Abstractors Supplement 20808.doc Breast Recoding Audit Final Report Supplement A High Quality Abstractors Table 3 illustrates the discrepancies by data item that were made by HQA’s Type of Discrpancies Made by High Quality Abstractors 6 5 4 3 2 1 0 Conclusion The number of discrepancies made by HQA’s was expected to be minimal since the audit was limited to only female breast cancer cases; however, with 33.7% of the discrepancies in this audit occurring in cases from HQA’s, this did not prove to be the case. These results indicate that the practice of pushing through high quality abstractors needed further review. Although 10% of the cases from HQA’s are visually edited, this may not be sufficient to ensure data quality. Recommendation 1) Perform recoding audit of high quality abstractors. Kyle Ziegler Page 3 12/30/2015 Macintosh HD:Users:peter:Desktop:PR LIGHT:CDPH to CPIC Et al:CDPH AUDIT 1\29\15:High Quality Abstractors Supplement 20808.doc Class of Case Of interest is that the discrepancies made by high quality abstractors and those made by other abstractors are very similar. Both groups had the highest number of discrepancies in Surgery of the Primary site and Scope Regional Lymph Node surgery. Sequence Number – Hospital The three data items with the most discrepancies are: • Scope Regional Lymph Node Surgery • Surgery of Primary Site • CS Site Specific Factor #3. RX Summ – Other RX Date – Other RX Summ – Transplt/Endocr Date_Transp_Endo RX Summ – BRM RX Date – BRM RX Summ – Hormone RX Date – Hormone RX Summ – Chemo RX Date – Chemo Rad – Boost RX Modality Rad – Regional RX – Modality RX Date – Radiation RX Summ – Surg Oth Reg/Dis RX Summ – Scope Reg LN Surgery RX Summ – Surg Prim Site RX Date – Surgery Regional Nodes Positive / Regional Nodes Examined CS Site Specific Factor 6 CS Site Specific Factor 5 CS Site Specific Factor 4 CS Site Specific Factor 3 CS Site Specific Factor 2 CS Site Specific Factor 1 CS Mets at DX CS Lymph Nodes CS Extension CS Tumor Size Diagnostic Confirmation Grade Behavior Code ICD-03 Histologic Type ICD-03 Laterality Primary Site including Subsite (CXX.X) Date of Diagnosis Spanish/Hispanic Origin Race 1 Sex County at DX Table 3