Measurement properties of the Dutch–Flemish patient-reported outcomes measurement information system (PROMIS) physical function item bank and instruments: a systematic review

Abstract

Background: Limitations in physical functioning are a big concern especially for patients with chronic or musculoskeletal diseases. Therefore, physical functioning is often used as a core outcome of treatments. The generic patient-reported outcomes information system (PROMIS) physical function (PF) item bank has shown potential to measure PF with better precision, interpretability and lower respondent burden compared with traditional patient-reported outcome measures. This study provides an overview of the current evidence on the quality of the measurement properties of the translated Dutch–Flemish PROMIS-PF item bank and its subdomains, and their derived short forms and computer adaptive tests (CATs). Methods: PubMed was searched up to June 17th 2020 for validation studies of Dutch–Flemish PROMIS-PF in Dutch and Flemish adults. Quality assessment of the included studies was conducted using the COSMIN Risk of bias checklist. The COSMIN criteria for good measurement properties were used to judge the results of the studies, which were adjusted and added to where needed for this review, in the context of IRT instruments and item banks. The quality of evidence was summarized for each measurement property based on the Grading of Recommendation Assessment, Development, and Evaluation (GRADE) approach. Results: Eleven studies were included, evaluating the PROMIS-PF item bank, the Upper Extremity (UE) subdomain, and/or their derived short forms and CATs in different clinical populations. There is evidence for sufficient structural validity, measurement precision, construct validity, and cross-cultural validity of the Dutch–Flemish PROMIS-PF item bank. The upper extremity subdomain item bank shows high quality evidence for structural validity and measurement precision. Content validity of these item banks has not been thoroughly demonstrated in a Dutch–Flemish population. Furthermore, the derived instruments have far less robust evidence: there are fewer validation studies available and none examined their performance as stand-alone administered instruments. Conclusions: The first studies into the Dutch–Flemish PROMIS-PF item bank and the UE subdomain show promising results, with especially high quality evidence for sufficient structural validity and measurement precision. However, more studies, and with higher methodological quality, are needed to study the instruments derived from these item banks. These studies should also evaluate content validity, reliability and responsiveness.