Comparison of Surgeon Performance of Rotator Cuff Repair: Risk Adjustment Toward a More Accurate Performance Measure.

BACKGROUND

Movement toward a value-based health-care system necessitates the development of performance measures to compare physicians, hospitals, and health-care systems. Patient-reported outcomes (PROs) are a potential metric. However, valid use of PROs hinges on the ability to risk-adjust for baseline patient differences across a surgeon's panel of patients. The purpose of this study was to propose an approach for baseline risk adjustment and evaluate the importance of risk adjustment when comparing surgeons' performance of rotator cuff repair.

METHODS

Patients (n = 995) treated with arthroscopic rotator cuff repair by 34 surgeons from 2010 to 2017 were identified from a large sports medicine clinical data registry. A linear regression model was used to adjust for baseline PROs, patient demographics, and clinical characteristics to predict American Shoulder and Elbow Surgeons (ASES) change scores for each surgeon. A risk-adjusted performance measure was calculated as the difference between the average unadjusted ASES change scores and the risk-adjusted predicted ASES change scores across all patients treated by a surgeon.

RESULTS

The differences between unadjusted and risk-adjusted performance scores varied widely across surgeons (range, -13.8 to 10.3 ASES points). Use of the risk-adjusted performance scores resulted in a dramatic change in the relative ranking of surgeons, compared with the ranking based on the observed ASES change scores, with 31 of the 34 surgeons' rank changing following risk adjustment. On average, the observed ASES scores improved from 49.5 ± 17.5 at baseline to 78.0 ± 22.5 at 6 months across all surgeons. In the risk-adjustment model (R = 0.44), male sex, Workers' Compensation status, higher scores on the Veterans RAND 12-item Health Survey (VR-12), lower baseline ASES scores, fair and poor tendon quality, and night pain all had a significant effect on the predicted ASES change scores (p < 0.05).

CONCLUSIONS

Our results show wide variation of nearly 25 points in the risk-adjusted 6-month ASES performance difference from the highest to the lowest-performing surgeons. Additionally, 91% of surgeons' rank changed following risk adjustment. This suggests that performance measurement that does not account for baseline patient characteristics would likely result in incorrect conclusions about a surgeon's relative performance based on PROs.

LEVEL OF EVIDENCE

Therapeutic Level IV. See Instructions for Authors for a complete description of levels of evidence.