External Validity of a New Prediction Model for Patient Satisfaction After Total Knee Arthroplasty

The ability to identify patients at risk of dissatisfaction after total knee arthroplasty (TKA) remains elusive. This study's purpose was to determine the external validity of a recently published prediction model for patient satisfaction (PMPS) with the hypothesis that it would achieve similar predictive success in our study sample.

A 10-question PMPS statistically derived from 5 patient-reported outcome questionnaires was tested for external validity in this prospective cohort investigation. The PMPS incorporates gender, age, stiffness, noise, and pain catastrophizing, with a score of 20 or greater predictive of satisfaction. As in the original study, to determine satisfaction the 2011 Knee Society Score (KSS) satisfaction subscale was collected at 3 months postoperatively. Two hundred seventy-four patients were administered the PMPS preoperatively, and 145 patients completed the KSS at 3 months postoperatively (53.0% response rate; 59% female; age, 64.9; body mass index, 32.5). A Bland-Altman analysis to assess agreement was performed.

One hundred thirty-three patients (91.7%) were satisfied and 12 (8.3%) were dissatisfied based on their postoperative KSS. The mean difference between the PMPS and KSS was 3.6 ± 8, but with a 95% prediction interval of -15.3 to 22.1 signifying almost no correlation. The PMPS did not predict any of the 12 dissatisfied patients postoperatively, and falsely predicted 5 patients to be dissatisfied of which 4 actually had a maximum postoperative KSS of 40.

A previously published, internally validated 10-question PMPS was unable to predict satisfaction after TKA in our external study sample. This study emphasizes the difficulty of developing a simple, but robust questionnaire that consistently predicts patient satisfaction after TKA.