Background and purpose: New generic patient-reported outcomes like the Patient-Reported Outcomes Measurement Information System (PROMIS) are available to physical therapists to assess physical function. However, the interpretation of the PROMIS Physical Function (PF) T-score is abstract because it references the United States average and not specific tasks. The purposes of this study were to (1) determine convergent validity of the PROMIS PF scale with physical performance tests; (2) compare predicted performance test values to normative data; and (3) identify sets of PROMIS PF items similar to performance tests that also scale in increasing difficulty and align with normative data.
Methods: Community-dwelling older adults (n = 45; age = 77.1 ± 4.6 years) were recruited for this cross-sectional analysis of PROMIS PF and physical performance tests. The modified Physical Performance Test (mPPT), a multicomponent test of mostly timed items, was completed during the same session as the PROMIS PF scale. Regression analysis examined the relationship of mPPT total and component scores (walking velocity, stair ascent, and 5 times sit to stand) with the PROMIS PF scale T-scores. Normative data were compared with regression-predicted mPPT timed performance across PROMIS PF T-scores. The PROMIS PF items most similar to walking, stair ascent, or sit to stand were identified and then PROMIS PF model parameter-calibrated T-scores for these items were compared alongside normative data.
Results and discussion: There were statistically significant correlations (r = 0.32-0.64) between PROMIS PF T-score and mPPT total and component scores. Regression-predicted times for walking, stair ascent, and sit-to-stand tasks (based on T-scores) aligned with published normative values for older adults. Selected PF items for stair ascent and walking scaled well to discriminate increasing difficulty; however, sit-to-stand items discriminated only lower levels of functioning.
Conclusions: The PROMIS PF T-scores showed convergent validity with physical performance and aligned with published normative data. While the findings are not predictive of individual performance, they improve clinical interpretation by estimating a range of expected performance for walking, stair ascent, and sit to stand. These findings support application of T-scores in physical therapy testing, goal setting, and wellness plans of care for community-dwelling older adults.