Evaluating anchor variables and variation in meaningful score differences for PROMIS

purposeMeaningful score differences (MSDs), as defined by recent FDA guidance, can improve the interpretation of outcome measure scores and score changes. Well-accepted methods for estimating MSDs typically rely on external anchor variables, but the applications of these methods are limited in children and adolescents with rheumatic diseases. This project explored multiple candidate anchors for the PROMIS

methodsLongitudinal data were extracted from the Childhood Arthritis and Rheumatology Research Alliance (CARRA) Registry. Candidate anchors included patient-reported domain-specific global impressions of change (GIC) along with other parent- and clinician-reported variables. Prior to MSD estimation, the quality of the anchors was assessed using a priori criteria (correlation ≥0.30, n≥10, <10% missing). Anchors meeting criteria were used to calculate MSDs.

resultsAmong 289 children with JIA and 47 with SLE, the GIC did not meet criteria inhalf of the scenarios. Other candidate anchors performed slightly better. The calculated MSDs varied by external anchor across measures, diagnoses, and direction of change (better vs worse).

conclusionsMany of the candidate external anchoring variables did not meet pre-specified criteria for calculating MSDs. Even for those that did, the choice of anchoring variable had a strong impact on the estimated MSD value and were different from other published values. As in adults, establishing pediatric MSDs requires selection of high-quality anchors, as changes in the variables used as anchors can impact MSD values and any subsequent score interpretations.