Predictors of ICU mortality in septic patients in a resource-limited African setting: a prospective multicenter cohort study in Lubumbashi, DR

BACKGROUND: Sepsis remains a leading cause of intensive care unit (ICU) mortality globally, with the highest burden observed in low- and middle-income countries where diagnostic capacity, timely referral, and access to organ support are limited. However, commonly used prognostic tools rely on laboratory and monitoring resources that are often unavailable in such settings. This study aimed to identify independent predictors of ICU mortality among adults with sepsis in Lubumbashi, Democratic Republic of the Congo, and to develop a context-adapted prognostic score suitable for resource-constrained environments. METHODS: We conducted a prospective multicenter cohort study across three ICUs between January 2021 and April 2023. Adults meeting Sepsis-3 criteria were consecutively enrolled. Clinical status, basic laboratory parameters, therapeutic timing, and socioeconomic characteristics were recorded within 24 h of ICU admission. Independent predictors of 28-day mortality were identified using multivariable logistic regression. A simplified point-based prognostic model (SPARS-Basique) was constructed and evaluated for discrimination (AUROC) and calibration, with internal validation using bootstrap resampling (1000 iterations). The outcome of interest was 28-day ICU mortality. RESULTS: < 90%, shock index ≥ 0.9, serum creatinine ≥ 3 mg/dL, and blood glucose ≥ 8 mmol/L. These variables formed the SPARS-Basique score (0-18 points). The model demonstrated strong discrimination (AUROC 0.89; bootstrap-corrected AUROC 0.87) and good calibration. Observed mortality increased across risk groups: 21% (score 0-5), 64% (6-9), and 91% (≥10). CONCLUSIONS: Sepsis mortality in Lubumbashi ICUs remains high and is influenced by both biological severity and structural barriers to timely care. The SPARS-Basique score demonstrated good internal performance for early risk stratification of ICU mortality in this cohort. However, as this was an exploratory derivation study in a modest sample, external validation in larger, independent cohorts is required before broader clinical application can be considered.