Abstract
BACKGROUND: In 2014, the Affordable Care Act Hospital Readmissions Reduction Program began penalizing hospitals for excessive readmission rates 30 days after total hip arthroplasty (THA) and total knee arthroplasty (TKA). Various data sets with nonstandardized validation processes report readmission data, which may provide conflicting outcome values for the same patient populations. METHODS: We queried 4 separate data sets: the American Joint Replacement Registry, Centers for Medicare and Medicaid Services billing data, the Vizient data set, and an advanced analytics integration (Cognos) report from our electronic medical record. We identified 2763 patients who underwent primary TKA and THA at a single academic medical center from June 2016 to June 2019. We then matched 613 surgery encounters in all 4 databases. Our primary outcome metric was 30-day readmissions. Fleiss' Kappa was used to measure agreement among the different data sets. RESULTS: Of the 613 THA and TKA patients, there were 45 (7.3%) readmissions noted. Data collected from the Centers for Medicare and Medicaid Services flagged 41 (6.7%) readmissions, Vizient flagged 11 (1.8%) readmissions, and the American Joint Replacement Registry and Cognos report both flagged 6 (0.98%) readmissions each. None of the readmissions were identified by all 4 data sets. There was significant disagreement among data sets using Fleiss' Kappa (kappa = -0.1318, P = .03). CONCLUSION: There is disagreement in readmission rates in databases receiving the same patient data after THA and TKA. Care must be taken to establish standard validation processes and reporting methods and scrutiny applied when interpreting readmission rates from various data sets.