Risk-adjusted outcomes in Medicare inpatient nephrectomy patients

Without risk-adjusted outcomes of surgical care across both the inpatient and postacute period of time, hospitals and surgeons cannot evaluate the effectiveness of current performance in nephrectomy and other operations, and will not have objective metrics to gauge improvements from care redesign efforts.We compared risk-adjusted hospital outcomes following elective total and partial nephrectomy to demonstrate differences that can be used to improve care. We used the Medicare Limited Dataset for 2010 to 2012 for total and partial nephrectomy for benign and malignant neoplasms to create prediction models for the adverse outcomes (AOs) of inpatient deaths, prolonged length-of-stay outliers, 90-day postdischarge deaths without readmission, and 90-day relevant readmissions. From the 4 prediction models, total predicted adverse outcomes were determined for each hospital in the dataset that met a minimum of 25 evaluable cases for the study period. Standard deviations (SDs) for each hospital were used to identify specific z-scores. Risk-adjusted adverse outcomes rates were computed to permit benchmarking each hospital's performance against the national standard. Differences between best and suboptimal performing hospitals defined the potential margin of preventable adverse outcomes for this operation.A total of 449 hospitals with 23,477 patients were evaluated. Overall AO rate was 20.8%; 17 hospitals had risk-adjusted AO rates that were 2 SDs poorer than predicted and 8 were 2 SDs better. The top performing decile of hospitals had a risk-adjusted AO rate of 10.2% while the lowest performing decile had 32.1%. With a minimum of 25 cases for each study hospital, no statistically valid improvement in outcomes was seen with increased case volume.Inpatient and 90-day postdischarge risk-adjusted adverse outcomes demonstrated marked variability among study hospitals and illustrate the opportunities for care improvement. This analytic design is applicable for comparing provider performance across a wide array of different inpatient episodes.