Evaluation of the PROMIS physical function item bank in orthopaedic patients

Abstract

The patient-reported outcomes measurement information system (PROMIS) physical function item bank v1 (PPFIB) contains 124 item response theory (IRT) calibrated items (Rose et al. 2008. J Clin Epidemiol 61:17–33).We report the psychometric properties of these items within an outpatient, orthopaedic patient population. In particular, we investigated whether a single unidimensional IRT scale can adequately define physical function of patients presenting with primarily upper or lower extremity orthopaedic complaints. We conducted a prospective study at an orthopaedic outpatient clinic to collect data from 865 adult patients with all 124 PROMIS physical function items and seven demographic items. Items were evaluated by a Rasch model. Total variance (60.6%) across the 124 items was explained by a single Rasch dimension. The variance explained by the second dimension was 7.7%, reflecting differential item functioning in the upper and lower extremity patients. The upper extremity physical function items had a pronounced ceiling effect. A single physical function dimension accounts for most of the item variance in the PPFIB, suggesting that the items are measuring predominantly one single construct. Separate subscales for lower versus upper extremities, especially with additional items at the upper trait level of the upper extremity subscale, may further enhance evaluation of physical function in orthopaedic patients.