A. N. Kothari1, T. Qu1, C. P. Fischer1, M. J. Anstadt2, P. P. Patel1, M. Singer1, G. J. Abood1 1Loyola University Medical Center,Surgery,Maywood, IL, USA 2Loyola University Chicago Stritch School Of Medicine,Maywood, IL, USA
Introduction: The development and use of novel measures of resident performance have gained the attention of national regulatory and accrediting bodies. With improvements in the availability of surgical quality improvement data and improved risk adjustment, a potential opportunity is to leverage resident-specific patient outcome data to evaluate resident performance. The objective of this study was to determine the feasibility of using surgical outcomes data to measure individual resident performance.
Methods: Institutional NSQIP data were obtained for all patients that underwent colectomy on a single surgical service from January, 2016 – December, 2017. A composite outcome of the following postoperative occurrences was used to develop risk-adjusted models: surgical site infection, wound disruption, unplanned intubation, pulmonary embolism, renal insufficiency, urinary tract infection, c. diff infection, readmission, death. These were chosen using a nominal group technique to identify occurrences potentially modified by resident involvement by program faculty. Resident-level outcomes were estimated from 2-level, random effects models.
Results: A total of 280 cases for 12 chief residents were analyzed. Resident case volume ranged from 15 to 32 with unadjusted composite occurrence rates ranging from 11.1% to 53.3%. No residents were classified as low outliers with better than expected outcomes, while one resident was classified as a high outlier with worse than expected outcomes (risk-adjusted rate: 41.8%). The proportion of variation in outcomes attributable to the resident was 0.03 (model intra-class correlation). Reliability of estimates was a median of 0.02 (0.01 – 0.06).
Conclusion: Resident-specific surgical outcomes cannot be reliably used to determine individual resident performance on an institutional level. Variation in measured outcomes can only minimally be attributed to the operating resident. Efforts to use resident-specific patient outcomes to measure performance should be avoided.