Predictors of Short-Term Outcome to Exercise and Manual Therapy for People With Hip Osteoarthritis

Abstract

Background Physical therapy for hip osteoarthritis (OA) has shown short-term effects but limited long-term benefit. There has been limited research, with inconsistent results, in identifying prognostic factors associated with a positive response to physical therapy. Objectives The purpose of this study was to identify potential predictors of response to physical therapy (exercise therapy [ET] with or without adjunctive manual therapy [MT]) for hip OA based on baseline patient-specific and clinical characteristics. Design A prognostic study was conducted. Methods Secondary analysis of data from a multicenter randomized controlled trial (RCT) (N=131) that evaluated the effectiveness of ET and ET+MT for hip OA was undertaken. Treatment response was defined using OMERACT/OARSI responder criteria. Ten baseline measures were used as predictor variables. Regression analyses were undertaken to identify predictors of outcome. Discriminative ability (sensitivity, specificity, and likelihood ratios) of significant variables was calculated. Results The RCT results showed no significant difference in most outcomes between ET and ET+MT at 9 and 18 weeks posttreatment. Forty-six patients were classified as responders at 9 weeks, and 36 patients were classified as responders at 18 weeks. Four baseline variables were predictive of a positive outcome at 9 weeks: male sex, pain with activity (<6/10), Western Ontario and McMaster Universities Osteoarthritis Index physical function subscale score (<34/68), and psychological health (Hospital Anxiety and Depression Scale score <9/42). No predictor variables were identified at the 18-week follow-up. Prognostic accuracy was fair for all 4 variables (sensitivity=0.5–0.58, specificity=0.57–0.72, likelihood ratios=1.25–1.77), indicating fair discriminative ability at predicting treatment response. Limitations The short-term follow-up limits the interpretation of results, and the low number of identified responders may have resulted in possible overfitting of the predictor model. Conclusions The authors were unable to identify baseline variables in patients with hip OA that indicate those most likely to respond to treatment due to low discriminative ability. Further validation studies are needed to definitively define the best predictors of response to physical therapy in people with hip OA.