Leukocyte differential count to unveil inflammatory and immunosuppression response in critically ill patients

Background Critically ill disease remains a leading cause of morbidity and mortality worldwide. Haematological biomarkers are inexpensive and universally available, yet their prognostic value remains incompletely characterised. We aim to evaluate the predictive value of these biomarkers for 28-day mortality. Methods Retrospective observational cohort study of 5163 adult intensive care unit (ICU) admissions at Vila Franca de Xira Hospital (2015–2024). Of these, 1683 met the sepsis criteria. Leukocyte indices (white blood cell count [WBC], absolute neutrophil count [ANC], absolute lymphocyte count [ALC], neutrophil-to-lymphocyte ratio [NLR]) were collected at admission and during the first five days of ICU stay. The primary outcome was 28-day all-cause mortality; secondary outcomes included ICU and long-term (up to three years) mortality. A multivariate logistic regression (including age, Simplified Acute Physiology Score [SAPS]) II and comorbidities) was developed to address the independent predictive value of these haematological biomarkers. The Receiver Operating Characteristic curves and the Kaplan-Meier survival analysis were used to identify the optimal threshold for predicting mortality for each sub-group. A Gradient Boosting Machine (GBM) classifier was used to assess the predictive power of leucocytes differential series. Results Haematological parameters, namely the NLR, predict outcomes of the critically ill patients. Septic patients and non-survivors had significantly higher NLR at admission (p < 0.001). In the multivariate model, NLR was independently associated with 28-day mortality (adjusted odds ratio [aOR] 1.10, 95% CI 1.02–1.19). NLR showed good discriminatory performance for 28-day mortality and emerged as the most important predictor in the GBM analysis. A declining WBC trajectory was also noted in non-survivors, highlighting the prognostic relevance of biomarker kinetics. The optimal thresholds for predicting 28-day mortality differed between septic and non-septic patients. Conclusions NLR is a low-cost prognostic biomarker in critically ill patients. Dynamic leukocyte trends may provide prognostic information beyond static admission values when integrated into critical disease risk assessment stratification frameworks after prospective and external validation. Different biomarker prognostic thresholds between septic and non-septic patients were noted.