Risk-Adjusted Hospital Outcomes in Medicare Total Joint Replacement Surgical Procedures

Background: Comparative measurement of hospital outcomes can define opportunities for care improvement and will assume great importance as alternative payment models for inpatient total joint replacement surgical procedures are introduced. The purpose of this study was to develop risk-adjusted models for Medicare inpatient and post-discharge adverse outcomes in elective lower-extremity total joint replacement and to apply these models for hospital comparison. Methods: Hospitals with ≥50 qualifying cases of elective total hip replacement and total knee replacement from the Medicare Limited Data Set database of 2010 to 2012 were studied. Logistic risk models were designed for adverse outcomes of inpatient mortality, prolonged length-of-stay outliers in the index hospitalization, 90-day post-discharge deaths without readmission, and 90-day readmissions after excluding non-related readmissions. For each hospital, models were used to predict total adverse outcomes, the number of standard deviations from the mean (z-scores) for hospital performance, and risk-adjusted adverse outcomes for each hospital. Results: A total of 253,978 patients who underwent total hip replacement and 672,515 patients who underwent total knee replacement were studied. The observed overall adverse outcome rates were 12.0% for total hip replacement and 11.6% for total knee replacement. The z-scores for 1,483 hospitals performing total hip replacements varied from −5.09 better than predicted to +5.62 poorer than predicted; 98 hospitals were ≥2 standard deviations better than predicted and 142 hospitals were ≥2 standard deviations poorer than predicted. The risk-adjusted adverse outcome rate for these hospitals was 6.6% for the best-decile hospitals and 19.8% for the poorest-decile hospitals. The z-scores for the 2,349 hospitals performing total knee replacements varied from −5.85 better than predicted to +11.75 poorer than predicted; 223 hospitals were ≥2 standard deviations better than predicted and 319 hospitals were ≥2 standard deviations poorer than predicted. The risk-adjusted adverse outcome rate for these hospitals was 6.4% for the best-decile hospitals and 19.3% for the poorest-decile hospitals. Conclusions: Risk-adjusted outcomes demonstrate wide variability and illustrate the need for improvement among poorer-performing hospitals for bundled payments of joint replacement surgical procedures. Clinical Relevance: Adverse outcomes are known to occur in the experience of all clinicians and hospitals. The risk-adjusted benchmarking of hospital performance permits the identification of adverse events that are potentially preventable.