Background/Objectives: Complex chronic patients represent a heterogeneous and high-risk population, for whom accurate prognostic tools are essential to guide clinical decision-making, optimize resource allocation, and support tailored interventions. The PROFUND index was developed for mortality prediction in polypathological patients, but its performance has not yet been evaluated in an ambulatory integrated care model. Methods: A retrospective observational study was conducted using two cohorts. Cohort H included complex chronic patients admitted to the Internal Medicine Department between March 2023 and February 2024. Cohort A comprised complex chronic patients followed by a multidisciplinary chronic care program between November 2016 and December 2023. PROFUND scores were derived from electronic health records. Discrimination for 12-month mortality was assessed using Kaplan–Meier curves, log-rank tests, and receiver operating characteristic curve analysis. Calibration was evaluated by comparing observed mortality with expected mortality based on the original PROFUND index and improved through intercept and slope recalibration. Results: A total of 660 patients were included in cohort H and 540 in cohort A. One-year mortality was 38.0% and 30.2%, respectively. Discriminatory performance was good in hospitalized patients (AUC 0.760; 95% CI 0.724–0.797) and moderate to good in ambulatory patients (AUC 0.705; 95% CI 0.656–0.754). Calibration analyses demonstrated systematic overestimation of mortality, particularly in the ambulatory cohort and intermediate–high risk strata, while recalibration improved agreement between predicted and observed risks. Conclusions: The PROFUND index provides useful risk stratification for 12-month mortality in CCP across care settings but overestimates absolute risk, particularly in ambulatory case management populations. Local recalibration may improve prognostic accuracy, support individualized care planning, and advance care planning discussions and allocation of multidisciplinary follow-up intensity.
Martins et al. (Sat,) studied this question.