Prognostic significance of blood immune cells in children with sepsis and establishment of a predictive model for PICU mortality: a retrospective study

Abstract

Objectives: This article aimed to investigate the correlation between blood immune cells and the prognosis in the early phase of pediatric sepsis and construct a prediction model for pediatric intensive care unit (PICU) mortality.

Methods: A total of 348 children admitted with sepsis to our PICU were retrospectively collected between January 2020 and June 2024. Of these, 242 children admitted from January 2020 to October 2022 were designated as the modeling group, while 106 children admitted between November 2022 and June 2024 were designated as the prospective validation group. Peripheral blood immune-related parameters, measured from the day of PICU admission to day 7, were analyzed in the modeling group. Risk factors were identified through multivariate logistic regression and integrated into a predictive nomogram. The nomogram was then applied to the prospective validation group to assess its discrimination and calibration. The nomogram's performance was evaluated using the area under the receiver operating characteristic curves (AUC), calibration plots, and decision curve analysis for both groups.

Results: Complicated with underlying diseases, invasive mechanical ventilation, increased pediatric risk of mortality score or pediatric sequential organ failure assessment score, and lymphopenia (d1) were independent risk factors for PICU mortality. The 90-day survival of patients with lymphopenia on the first day after admission was low. In addition, patients with persistent lymphopenia had higher mortality. The nomogram showed an AUC of 0.861 (95% CI: 0.813 to 0.909) in the modeling group and 0.875 (95% CI: 0.797 to 0.953) in the prospective validation group. The nomogram also performed well based on the calibration curve and decision curve analysis.

Conclusion: Assessing lymphocytes within seven days of PICU admission may be conducive to identifying children with sepsis at increased mortality risk. The nomogram performed well in predicting PICU mortality among patients of interest.

Keywords: PICU mortality; children; lymphocytes; predictive model; sepsis.

Figure 1

Variations in parameters related to blood routine immune cells in the two groups [(A–F) The number of neutrophil, lymphocyte, monocyte, eosinophil, basophil, and platelet on d1, d3, and d7 after admission in the survival group and the non-survival group. (G–I) The neutrophil-to-lymphocyte ratio (NLR), lymphocyte-to-monocyte ratio (LMR) and platelet-to-lymphocyte ratio (PLR) on d1, d3, and d7. *P < 0.05, **P < 0.01, ***P < 0.001.]

Figure 2

Forest map of independent risk factors for PICU mortality in children with sepsis.

Figure 3

90-day survival curves of the lymphopenia group and non-lymphopenia group.

Figure 4

Nomogram of PICU mortality prediction model for children with sepsis. [Example illustrating the use of the model was marked in red: Taking a 1.90-year-old child with sepsis as an example, the pSOFA score is 3 (corresponding points = 48), no lymphopenia occurs (corresponding points = 34), PRISM is 7 (corresponding points = 55), invasive mechanical ventilation is required (corresponding points = 53), and no underlying diseases is present (corresponding points = 31), totaling 221 points, which corresponds to a mortality rate of about 25.6%.].

Figure 5

Validation and evaluation of the nomogram model. [(A–C): ROC curve, calibration chart and decision curve analysis of the modeling group; (D–F): ROC curve, calibration curve, and decision curve analysis for the prospective validation group.].