Accuracy and Validity of Computer Adaptive Testing for Outcome Assessment in Patients Undergoing Total Knee Arthroplasty

Computerized adaptive test (CAT) questionnaires may allow standardization of patient-reported outcome measures and reductions in questionnaire burden. We evaluated the validity, accuracy, and efficacy of a CAT system in patients with end-stage osteoarthritis undergoing total knee arthroplasty.

CAT Knee Osteoarthritis Outcome Scores (KOOS) and CAT KOOS-JR questionnaires were applied to 1871 standard form KOOS and 1493 KOOS-JR patient responses, respectively. Mean, standard deviations, Pearson’s correlation coefficients, interclass correlation coefficients (ICCs), frequency distribution plots, and Bland-Altman plots were used to compare the precision, validity, and accuracy between CAT scores and full-form scores.

There was a mean reduction of 14 questions (33%) in the CAT KOOS and 1.4 questions (20%) with the CAT KOOS-JR version, compared with the standard KOOS and KOOS-JR surveys, respectively. There were no significant differences between KOOS and CAT KOOS scores with respect to pain ( P = .66), symptoms ( P = .43), quality of life ( P = .99), activities of daily living ( P = .68), and sports ( P = .84). Similarly, there were no significant differences between the standard form KOOS-JR and CAT KOOS-JR scores ( P = .94). There were strong correlations with minimal variability between the CAT KOOS and standard KOOS questionnaires for pain ( r = 0.98, ICC: 0.98), symptoms ( r = 0.97, ICC: 0.97), quality of life scores ( r = 0.99, ICC: 0.99), activities of daily living scores ( r = 0.99, ICC: 0.99), and sports scores ( r = 0.99, ICC: 0.99). Similarly, there were strong correlations between the KOOS-JR and the CAT KOOS-JR scores ( r = 0.99, ICC: 0.99).

CAT KOOS and the CAT KOOS-JR versions are accurate and reduce questionnaire burden up to one-third compared with standard surveys. CAT versions may improve patient compliance and decrease fatigue.