Internal Validation of a Predictive Model for Complications After Total Hip Arthroplasty

Background: Total hip arthroplasty (THA) is projected to increase in prevalence and associated complications will impose significant cost on the US healthcare system. The purpose of this study is to validate a predictive model for postoperative complications utilizing a novel 11-component hip-specific questionnaire encompassing preoperatively available clinical and radiographic data. Methods: Consecutive primary THA patients between January 2014 and January 2016 were included. Exclusion criteria included patients without questionnaire scoring variables and less than 1-year follow-up. Patients were stratified into 4 tiers based on their questionnaire score: low risk (> 74), mild risk (5773), moderate risk (41-56), and high risk (< 40). A binary logistic regression was performed to determine if the questionnaire predicted complications. Receiver-operator curves were constructed to determine the threshold score below which there was a high likelihood of experiencing a complication. Results: Four hundred fifty patients were included in the final analysis with a mean (range) follow-up of 2.1 years (1.0-5.9), age of 63.1 years (25.7-9.17), and body mass index of 31.7 kg/m(2) (17.8-64.5). The complication rate was 13.6%. A hip questionnaire score of 73.8 conferred a 98.5% sensitivity and 98.5% negative predictive value for complications. The questionnaire score was the strongest predictor of a decreased complication likelihood (odds ratio 0.94, 95% confidence interval 0.90-0.97, P < .001). Risk tier was significantly associated with complications (low risk: 0; mild risk: 12; moderate risk: 25; and high risk: 24; P < .001). Conclusion: This novel hip questionnaire demonstrated a high sensitivity and negative predictive value to identify patients at risk for postoperative complications. Future studies should attempt to prospectively validate the use of this questionnaire. (C) 2018 Elsevier Inc. All rights reserved.