Background/Purpose: Patient-Reported Outcomes Measurement Information System (PROMIS) provides reliable, valid measures of health status to resolve many challenges with comparability and interpretability in OA. Minimally important differences (MID) allow for meaningful interpretation of patient-reported outcomes, however, MIDs of PROMIS scores are unknown for OA. Our purpose was to establish a range of MIDs for 4 PROMIS Short Forms: Physical Function, Pain Interference, Depression, and Anxiety for knee OA following the guidelines of the PROMIS Instrument Validation Standards.

Methods: We performed a longitudinal analysis using the pooled, similar treatment effects that resulted from a randomized trial comparing Tai Chi with physical therapy in adults with symptomatic knee OA (ACR criteria). Participants completed surveys before and after 12-week intervention (Table). We used legacy anchors as a reference to estimate the MIDs. We paired a set of legacy anchors with each PROMIS instrument based on similarity of their target constructs or clinical relevance (Table). We further defined important change in each anchor measure based on previously published MID (range = MID to 2 times MID). Among participants with important change for each anchor we calculated the absolute mean change in corresponding PROMIS T-scores to determine an individual MID estimate. We then used distribution-based methods to evaluate the quality of each MID, and only selected MIDs that met 3 a priori criteria: 1) Spearman correlations between the anchor and PROMIS change scores ≥0.3; 2) subset sample sizes ≥10; and 3) the absolute values of the effect sizes (Cohen’s d) between 0.2–0.8. The lowest and highest selected-MID estimates created the lower and upper bound of an initial MID range for each PROMIS instrument. When the lower bound estimate was smaller than the standard error of measurement (SEM, i.e. smallest change score exceeding measurement error), the lower bound would be set to the SEM.

Results: We had 165 participants (mean age 61 years, 70% female, 53% white, 92% KL Grade ≥ 2). Of 14 estimated MIDs, 8 met all criteria. Four MIDs did not meet criterion #3 (described above); 2 MIDs did not meet criterion #1; and 1 MID did not meet criterion #2 (Table). The SEMs for PROMIS Depression, Anxiety, Physical Function, and Pain Interference were 2.3, 2.3, 1.9, and 1.8, respectively. The final MID ranges were: PROMIS Depression = 2.8 to 3.0; PROMIS Anxiety = 2.3 to 3.4; PROMIS Physical Function = 1.9 to 2.2; and PROMIS Pain Interference = 2.4 to 3.1. These correspond to 20-33% of the normalized standard deviation units per scale.

Conclusion: We established the first estimates of MID for PROMIS Physical Function, Pain Interference, Depression, and Anxiety Short Forms among those with symptomatic knee OA. This information will be an important standard of reference to better apply or interpret PROMIS instruments in future studies.            

< td width="126" valign="top">

X