A Comparison of the Psychometric Properties of PROMIS Computer Adaptive Tests and Short Forms Vs Legacy Patient-Reported Outcome Measures in Total Knee Arthroplasty Patients

Background: Traditionally used PROMs evaluating total knee arthroplasty (TKA) have large measurement error and limited measurement range. Patient-Reported Outcomes Measurement Information System (PROMIS) has theoretical possibilities to overcome these limitations. This study evaluates the psychometric properties of PROMIS vs legacy PROMs in TKA patients. Methods: 210 patients were included from 3 orthopaedic departments. Patients completed a questionnaire twice in a 2-week interval, including 2 PROMIS Computer Adaptive Tests (CATs), assessing Physical Function (PF) and Pain Interference (PI), 5 PROMIS short forms (SF) (PF-SF8b, PF-SF10a, PF-SF-20a, PI-SF8a, PI-1a) and 6 legacy PROMs (KOOS, KOOS-PS, WOMAC, OKS, 2 numeric rating scales). Reliability, measurement precision, smallest detectable change (SDC), construct validity, burden and extreme scores were investigated. Results: All PROMIS CATs, SFs, and legacy PROMs showed adequate test-retest reliability. PROMIS CATs and SFs showed sufficient construct validity. Regarding PF, the SDC varied between 2.8-6.1 for PROMIS and 7.3-19.9 for legacy PROMs. PROMIS CAT and SFs showed no extreme scores using 5-20 items, legacy PROMs showed 1.2%-3% extreme scores using 7-17 items. Concerning pain, the SDC varied between 2.8-6.1 for PROMIS and 22.5-35.4 for legacy PROMs. PROMIS showed 0%-9.5% extreme scores using 5-8 items and legacy PROMs 8.7%-19.1% using 1-9 items. Conclusions: PROMIS CAT and SFs seem more efficient assessing PF in TKA compared to legacy PROMs by offering reduced burden and measurement error, and minimizing extreme scores. PROMIS PI-8a seems most suitable in measuring pain without extreme scores. This study supports a shift from legacy PROMs toward PROMIS in TKA patients.