Prognostic value of low-cost white blood cell indices and procalcitonin for mortality in Rwandan sepsis patients: a prospective intensive care unit study.

Abstract

Background: In resource-limited settings, early identification of sepsis and low-cost mortality predictors is critical for intensive care unit (ICU) triage. This study evaluated the prognostic value of baseline sociodemographic factors, routine hematological indices, and serum procalcitonin (PCT) levels for 40-day mortality among adult ICU patients meeting Sepsis-2 criteria in Rwanda.

Methods: A prospective cohort of 125 ICU patients was followed for 40 days. Baseline variables included sex, age, PCT, total white blood cell (WBC) count, differential counts (neutrophils, basophils, eosinophils, monocytes, lymphocytes), and neutrophil-to-lymphocyte ratio (NLR). Survival probabilities were estimated using Kaplan-Meier curves and log-rank tests. Cox proportional hazards models identified independent mortality predictors, with assumptions tested via Schoenfeld residuals and multicollinearity assessed using variance inflation factors. Time-dependent receiver operator curve (ROC) analysis evaluated model performance at days 6, 10, and 15 using the area under the curve (AUC) values.

Results: Of 125 patients, 56 (44.8%) were female. Median age was 41 years for survivors and 50 years for non-survivors (p = 0.097). In multivariable Cox regression, elevated neutrophil counts were independently associated with increased mortality [adjusted hazard ratio (aHR)] 1.99; 95% CI (confidence intervals) 1.37-2.88; p < 0.001), corresponding to a twofold higher hazard of death for approximately a threefold increase in neutrophil count. No significant associations were found for sex, age, or PCT. ROC analysis showed that models integrating neutrophils and total WBC (TotalWBC) achieved the highest predictive accuracy, with AUCs ranging from ~68% to 71% across all time points, outperforming simpler models.

Conclusions: Elevated neutrophil counts at ICU admission are independently associated with increased mortality. Integrating absolute neutrophil and WBC data into predictive models enhances early mortality risk stratification. These findings underscore the value of routine biomarkers and robust modeling to guide timely interventions in resource-constrained ICU settings.