[Construction and evaluation of a 28-day mortality nomogram prediction model for children with sepsis]

To analyze factors influencing 28-day mortality in children with sepsis, to construct a nomogram prediction model, and evaluate its predictive value.

A retrospective cohort study was conducted using clinical data from children with sepsis admitted to the pediatric intensive care unit (PICU) of the Second People's Hospital of Liaocheng from January 2017 to December 2023. Data collected included gender, age, pediatric Critical Illness Score (pCIS), pediatric Sequential Organ Failure Assessment (pSOFA), pediatric Sepsis-Induced Coagulopathy (pSIC) score, white blood cell count (WBC), neutrophil percentage (NEU%), lymphocyte percentage (LYM%), platelet count (PLT), mean platelet volume (MPV), C-reactive protein (CRP), procalcitonin (PCT), blood lactic acid, activated partial thromboplastin time (APTT), prothrombin time (PT), international normalized ratio (INR), fibrinogen (Fib), serum alanine aminotransferase (ALT), total bilirubin (TBil), serum creatinine, blood urea nitrogen (BUN), cardiac troponin I (cTnI), creatine kinase isoenzyme (CK-MB), and 28-day prognosis. According to the prognosis of children with sepsis within 28 days, the children were divided into two groups, namely the survival group and the death group, and the differences of each index between the two groups were compared. Lasso regression analysis was used for preliminary screening of risk factors affecting 28-day mortality in children with sepsis; multivariate Logistic regression analysis was employed to further identify independent risk factors; a nomogram prediction model was constructed based on the independent risk factors screened out, and the model was evaluated simultaneously.

A total of 165 children with sepsis were included, of whom 150 survived and 15 died within 28 days. Compared with the survival group, the death group had higher levels of pCIS score, pSOFA score, pSIC score, LYM%, PCT, blood lactic acid, APTT, PT, INR, serum creatinine, BUN, cTnI, and CK-MB, and lower levels of NEU%, PLT, and Fib (all P<0.05). Lasso regression and multivariate Logistic regression analyses showed that pSIC score [odds ratio (OR)=4.31, 95% confidence interval (95%CI) was 2.23-8.31, P<0.001], blood lactic acid (OR=1.51, 95%CI was 1.26-1.80, P<0.001) and CK-MB (OR=1.05, 95%CI was 1.03-1.07, P<0.001) were independent risk factors for 28-day mortality in children with sepsis. A nomogram prediction model for 28-day mortality in children with sepsis was constructed based on the above three indicators. Receiver operator characteristic curve (ROC curve) showed that the concordance index of the model was 0.95, and the area under the curve (AUC) was 0.976 (95%CI was 0.955-0.997), indicating that the model had good discrimination; calibration curve showed that the calibrated curve was close to the reference curve, indicating that the model had good calibration; decision curve analysis (DCA) showed that the model had significant application value in clinical practice.

pSIC score, blood lactic acid, and CK-MB are all independent risk factors for 28-day mortality in children with sepsis. The nomogram model constructed based on them has good predictive value for the 28-day mortality risk and can guide clinical decision-making.