Self-Reported Versus Performance-Based Measures of Physical Function and Mortality Among Patients With Kidney Failure: A Cohort Analysis of the CONVINCE Trial

RATIONALE & OBJECTIVE: While physical function (PF) measures obtained through patient-reported outcomes (PRO) are simpler to obtain than performance-based outcomes (PerfO), their relative prognostic value remains unclear. This study compared the associations of these measures with all-cause mortality in patients with kidney failure enrolled in the CONVINCE trial. STUDY DESIGN: Observational multicenter cohort study. SETTING & PARTICIPANTS: Adult patients with kidney failure undergoing high-dose hemodiafiltration or high-flux hemodialysis enrolled in the CONVINCE trial between November 2018 and April 2021. PREDICTORS: Baseline PF scores derived from (1) PRO: the 4-item PROMIS-PF short form (PROMIS-PF4a), (2) PerfO: the Physical Performance Test (PPT), and (3) composite PF scores integrating both assessment types. The different PF metrics were standardized to permit comparisons on a common scale. OUTCOME: All-cause mortality over a median follow-up of 30 months. ANALYTICAL APPROACH: Unadjusted and multivariable Cox proportional hazards regression models fitted for each PF predictor. Model discrimination was evaluated using Harrell's C-index and time-dependent AUCs. Restricted cubic splines assessed risk across the PF continuum. Kaplan-Meier curves were stratified according to predefined PROMIS-PF score cut points. RESULTS: Of 1,360 included patients (37.1% women, mean age 62.4 years), 266 (19.6%) died during follow-up. All three PF measures demonstrated statistically significant and comparable associations with mortality in both unadjusted and adjusted analyses. Adjusted hazard ratios (HRs) per 1-point increase in PF score were 0.96 (95% CI, 0.95-0.98) for PROMIS-PF4a, 0.95 (95% CI, 0.94-0.97) for PPT, and 0.96 (95% CI, 0.94-0.97) for the composite score. Discriminative ability was modest but similar across measures (adjusted C-index ≈ 0.71). Across all assessment types, severe PF impairment (T-score<30), was associated with substantially higher mortality risk. LIMITATIONS: Observational cohort analysis of a trial population; generalizability is limited to trial participants and to the examined PF tools. CONCLUSIONS: Physical function was strongly and independently associated with mortality in kidney failure. The patient-reported PROMIS-PF4a and the performance-based PPT demonstrated comparable associations with mortality. These findings support use of the brief PROMIS-PF instrument as a practical alternative to performance-based assessments for routine monitoring of PF in dialysis care.