Sex and the Multidimensional Prognostic Index in 3.5-year post-COVID-19 mortality among older adults: evidence of a time-varying effect

Purpose: The long-term prognostic impact of frailty in older adults recovering from COVID-19 remains underexplored. The Multidimensional Prognostic Index (MPI) has shown utility in predicting short-term outcomes, but its role over extended follow-up requires further investigation. The objective of this study is to evaluate the ability of an MPI-based model to predict 3.5-year mortality in older adults hospitalized for COVID-19. Methods: This single-center cohort study with prospective follow-up included 183 patients aged ≥ 65 years hospitalized with confirmed SARS-CoV-2 infection. MPI was calculated at admission and dichotomized into low (classes 1–2) and high (class 3). Multivariable Cox regression was used to estimate the hazard of mortality over a 3.5-year follow-up. Discriminative performance was assessed using time-dependent ROC analysis, with AUC values compared between the multivariable model and MPI alone. Results: During follow-up, 81/183 patients (44.3%) died. Kaplan–Meier curves showed lower survival in high-MPI patients (log-rank p = 0.0043). A Cox model with a time-varying effect for sex (sex × log(time)) confirmed that high MPI was associated with higher mortality (HR = 1.59, 95% CI 1.00–2.52), age was also associated (HR per year = 1.04, 95% CI 1.00–1.07), while vaccination was not. The female-to-male hazard ratio changed over time (HR at 180/365/730/1250 days: 0.90/1.31/1.87/2.48). AUCs for the full model at 180/365/730/1250 days were 0.704/0.654/0.680/0.659, derived from the sex-stratified Cox linear predictor, and exceeded the MPI-only model. Conclusions: The MPI demonstrated moderate prognostic ability for long-term mortality among older adults after COVID-19. Adding demographic and clinical variables modestly improved prediction, supporting the role of multidimensional assessment in geriatric prognosis, while highlighting the need for cautious interpretation over extended follow-up.