Standardizing physical function across assessment types: Three versions of the physical performance test were linked...

Physical function (PF) is a relevant outcome in aging and clinical populations, commonly measured through patient-reported outcome (PRO) or performance-based outcome (PerfO) assessments. While various PRO measures have been calibrated to the standardized PROMIS-PF T-score metric, PerfO scores yield scores on distinct metrics, limiting interpretability, comparability and integration of PF results across studies and settings. The aim of this study was to establish linking algorithms mapping PerfO assessments onto the PROMIS-PF metric.

This methodological work reports the main results of the "Standardizing-PF" project, a prospectively designed PRO-PerfO linking study in a heterogenous, multisite sample. Adults from clinical and community-based settings completed the 20-item PROMIS-PF short form (PROMIS-PF20a) and three physical performance test (PPT) versions (PPT-9, PPT-7, and modified PPT). Linking assumptions aligned with consensus-based standards for the selection of health measurement instruments-relevant criteria, including item- and test-level statistics and factor analyses. Item response theory-based and equipercentile linking methods were applied and systematically compared to derive crosswalk tables converting PPT scores into PROMIS-PF T-scores. Validity and stability of the linked scores were examined across multiple subsamples.

Data from N = 812 participants were analyzed (64.6% women, median age = 65.5), including geriatric (n = 226) and rheumatology (n = 204) inpatients, rehabilitants with psychocardiological (n = 124) or post-COVID conditions (n = 77), and non-clinical elderly (age ≥60; n = 181). Latent correlation between PROMIS-PF20a and PPT was high (r = 0.89). Confirmatory factor analysis supported essential unidimensionality of the pooled PRO and PerfO item set (comparative fit index = 0.98). In a bifactor model, the explained common variance of a general PF factor was 0.76. After linking the PPT to the PROMIS T-scale, we found satisfactory agreement between PROMIS-PF20a and PPT-based T-scores, with equipercentile linking yielding more stable agreement across validation subsamples. An exception was post-COVID patients, scoring systematically higher in PerfOs in both linking approaches.

Three PPT versions were successfully linked to the standardized PROMIS-PF metric across diverse populations, facilitating valid comparisons and consistent interpretation of performance-based and patient-reported assessments of PF. The resulting user-friendly crosswalk tables may engage in broader application of standardized PF measurement by supporting straightforward score translation across distinct assessment types. For some population, specific linking algorithms might be required; cautious use is advised particularly in post-COVID patients.

PF, reflecting an individual's ability to carry out daily activities independently, is a meaningful outcome that can be measured in multiple ways-most commonly with self-reported questionnaires and performance tests. Various self-reported questionnaires have been standardized by psychometrically linking them to the PROMIS-PF T-score metric (mean = 50, standard deviation = 10), allowing consistent interpretation across scores from different self-reported measures. However, linking performance tests to this metric has been challenging, as these assessments are different from self-reports. This study examined whether PPTs can be reliably linked to the PROMIS-PF scale. For this, data from 812 clinical and non-clinical participants were collected, which answered the 20-item PROMIS-PF short form and executed three versions of the PPT. Based on the linking analysis, which demonstrated largely strong relationship between both measures, we developed user-friendly crosswalk tables to simply translate PPT scores into standardized PROMIS-PF T-scores. Interpretation requires caution for the post-COVID patients, as they generally tended to underestimate their functional level. Nevertheless, the published crosswalk tables are a practical tool for researchers and clinicians. allowing efficient comparison and interpretation of PF scores-even across two different assessment types.