Abstract
BACKGROUND: Attention to the collection of patient-reported outcomes measures (PROMs) associated with total hip arthroplasty (THA) is growing. The aim of this study was to preoperatively predict failure to achieve the minimal clinically important difference (MCID) and substantial clinical benefit (SCB) between pre- and postoperative PROMs. In addition, we sought to identify factors predictive of failure to achieve MCID and SCB in patients undergoing a THA. METHODS: A retrospective query of the electronic health record data was performed at a single institution. Outcomes of interest were the anchor-based MCID, the distribution-based MCID, and the SCB for the Hip Disability and Osteoarthritis Outcome Score for Joint Replacement. Several machine learning models were built for each outcome and were evaluated by areas under the receiver operating characteristic curve and the precision-recall curve. Furthermore, logistic regression models were used to identify significant risk factors. RESULTS: Of the 857 patients who underwent THA, 350 patients completed both pre- and postoperative surveys. Of the final sample (i.e., 350 patients), 56 (16.0%), 29 (8.3%), and 71 (20.3%) failed to reach the anchor-based (i.e., 17.7 points) and distribution-based (10.6 points) MCIDs and the SCB (i.e., 22.0 points). The machine learning model performances were far beyond the baseline and comparable to the ones in existing studies, suggesting reliability in the prediction. Two shared factors associated with the failure in both the MCIDs and the SCB were highlighted: a patient's race and pre-existing mental illness. CONCLUSION: Understanding the risk factors of failing to meet MCID and SCB may provide a more objective opportunity to quantify patient and surgeon expectations associated with THA. Our findings call stakeholders' particular attention to patients with preoperative mental disorders, and raise further questions regarding the impact of race, in the care of patients with degenerative hip disease.