Development and validation of a nomogram to predict the risk of sepsis-associated encephalopathy for septic patients in PICU: a multicenter retrospective cohort study

Abstract

Abstract Background Patients with sepsis-associated encephalopathy (SAE) have higher mortality rates and longer ICU stays. Predictors of SAE are yet to be identified. We aimed to establish an effective and simple-to-use nomogram for the individual prediction of SAE in patients with sepsis admitted to pediatric intensive care unit (PICU) in order to prevent early onset of SAE. Methods In this retrospective multicenter study, we screened 790 patients with sepsis admitted to the PICU of three hospitals in Shandong, China. Least absolute shrinkage and selection operator regression was used for variable selection and regularization in the training cohort. The selected variables were used to construct a nomogram to predict the risk of SAE in patients with sepsis in the PICU. The nomogram performance was assessed using discrimination and calibration. Results From January 2017 to May 2022, 613 patients with sepsis from three centers were eligible for inclusion in the final study. The training cohort consisted of 251 patients, and the two independent validation cohorts consisted of 193 and 169 patients. Overall, 237 (38.7%) patients developed SAE. The morbidity of SAE in patients with sepsis is associated with the respiratory rate, blood urea nitrogen, activated partial thromboplastin time, arterial partial pressure of carbon dioxide, and pediatric critical illness score. We generated a nomogram for the early identification of SAE in the training cohort (area under curve [AUC] 0.82, 95% confidence interval [CI] 0.76–0.88, sensitivity 65.6%, specificity 88.8%) and validation cohort (validation cohort 1: AUC 0.80, 95% CI 0.74–0.86, sensitivity 75.0%, specificity 74.3%; validation cohort 2: AUC 0.81, 95% CI 0.73–0.88, sensitivity 69.1%, specificity 83.3%). Calibration plots for the nomogram showed excellent agreement between SAE probabilities of the observed and predicted values. Decision curve analysis indicated that the nomogram conferred a high net clinical benefit. Conclusions The novel nomogram and online calculator showed performance in predicting the morbidity of SAE in patients with sepsis admitted to the PICU, thereby potentially assisting clinicians in the early detection and intervention of SAE.

Keywords

• Sepsis-associated encephalopathy
• Nomogram
• Prediction
• PICU