B. T. Fry1,2, S. P. Shubeck2,3, J. R. Thumma2, J. B. Dimick2,3 1University Of Michigan,Medical School,Ann Arbor, MI, USA 2University of Michigan,Center For Healthcare Outcomes & Policy,Ann Arbor, MI, USA 3Michigan Medicine,Department Of Surgery,Ann Arbor, MI, USA
Introduction: Individual surgeon performance data has become increasingly available to patients and providers alike. Additionally, there is a push at many institutions for internal reporting of surgeon outcomes to promote quality improvement. However, major concerns exist over the reliability of using surgeon specific data to detect differences in performance. No study has comprehensively evaluated the effect of low surgeon volume on multiple surgical outcomes across a wide variety of procedures.
Methods: Using 2014 data from the State Inpatient Database, we calculated population level average mortality and complication rates across five surgical procedures: coronary artery bypass grafting, colectomy, pancreatectomy, total hip replacement, and bariatric procedures. We then calculated the minimum surgeon volume necessary to detect a doubling of each outcome rate at an alpha level of 0.05 and power level of 80%. Finally, we used annual individual surgeon caseloads to determine the proportion of surgeons who met or exceeded these minimum volumes. We then performed a sensitivity analysis to examine the proportion of surgeons who met the minimum volume threshold when aggregating caseloads over 3 years.
Results: Surgeon specific data was available for 13,708 surgeons who performed a total of 236,413 cases in 8 states. Average mortality rates ranged from 0.05% for bariatric procedures to 4.1% for colectomy. Average complication rates ranged from 2.2% for bariatric procedures to 31.3% for pancreatectomy. Virtually 0% (1 of 13,708) of all surgeons performed an adequate number of cases annually to detect a doubling of the average mortality rate, while 9% (1,280) of surgeons performed enough annual cases to detect a doubling of the average complication rate. When examining estimated 3-year aggregate caseloads, 0.3% (48) of surgeons would perform enough cases to detect a doubling in mortality, while 25% (3,414) of surgeons would perform enough cases to detect a doubling in complication rates.
Conclusion: The majority of surgeons do not perform an adequate number of procedures to detect differences in individual mortality and complication rates. These results suggest that surgeon level outcome data cannot reliably assess performance and quality.