Presentation (CTI)

OBJECTIVES: Austria-specific reference values and interpretable change thresholds for outcomes after knee arthroplasty rehabilitation are lacking. We derived normative reference data, estimated distribution-based MCID thresholds, and examined whether baseline-stratified thresholds alter responder classification in three-week phase II inpatient rehabilitation.
METHODS: Retrospective routine data from four Austrian rehabilitation centers (Dec 2023-Dec 2024) were analyzed. Adults after knee arthroplasty required valid admission and discharge data for EQ-5D-5L index, EQ-VAS, pain numeric rating scale (NRS), Health Assessment Questionnaire (HAQ), Patient Health Questionnaire 4 (PHQ-4), and WOMAC. Changes were assessed using paired t-tests and Cohen's d. Age- and sex-specific percentiles were generated. Exploratory models examined predictors of discharge status. Distribution-based MCIDs were calculated as 0.5×pooled SD globally and within five baseline strata. Responder classifications were compared using Holm-corrected McNemar tests.
RESULTS: The cohort included 1,200 patients (mean age 68.2±9.7 years; 60% female; mean stay 21.4±1.9 days). All core outcomes improved (p<0.001). Effect sizes were largest for WOMAC (d=0.57) and pain NRS (d=0.54), and smaller for EQ-VAS (d=0.22) and EQ-5D-5L index (d=0.27). The corresponding admission score was consistently the strongest predictor of discharge status; adding age, sex, and center explained little additional variance. Global MCID estimates were 0.068 (EQ-5D-5L index), 10.2 (EQ-VAS), 0.99 (NRS), 0.241 (HAQ), 1.17 (PHQ-4), and 19.9 (WOMAC). Baseline-stratified MCIDs differed across strata and classified more patients as responders overall: across the six outcomes, responder classifications increased from 40.4% to 50.8%, corresponding to +10.4 percentage points additional responders classified. Increases were significant for EQ-VAS, HAQ, PHQ-4, and WOMAC (all p<0.005), but not for EQ-5D-5L index or NRS.
CONCLUSIONS: Baseline-stratified MCIDs may provide a more individualized interpretation of rehabilitation outcomes after knee arthroplasty than global thresholds. By incorporating patients’ admission status, they may better reflect heterogeneous recovery trajectories and support personalized outcome interpretation. Future studies should validate these baseline-dependent thresholds against patient-anchored change ratings.