New multivariable prediction model PEdiatric SEpsis recognition and stratification (PESERS score) shows excellent discriminatory capacity

Aim To develop a quantitative predictive scoring model for the early recognition and assessment of paediatric sepsis. Methods Prospective observational study including emergency department and in-hospital febrile patients under 18 years. Sepsis diagnose (Goldstein 2005 definitions) was the main outcome. Variables associated with the outcome were included in a multivariable analysis. Cut-off points, odds ratio and coefficients for the variables kept after the multivariable analysis were identified. The score was obtained from the coefficients, The AUC was obtained from ROC-analysis, and internal validation was performed using k-fold cross-validation. Results The analysis included 210 patients. 45 variables were evaluated and the bivariate analysis identified 24 variables associated with the outcome. After the multivariable regression, 11 variables were kept and the score was obtained. The model yielded an excellent AUC of 0.886 (95% CI 0.845-0.927), p < 0.001 for sepsis recognition. With a cut-off value of 5 for the score, we obtained a sensitivity of 98%, specificity of 76.7%, positive predictive value of 87.9% and negative predictive value of 93.3%. Conclusion The proposed scoring model for paediatric sepsis showed adequate discriminatory capacity and sufficient accuracy, which is of great clinical significance in detecting sepsis early and predicting its severity. Nevertheless external validation is needed before clinical use.

Automated summary

This study aimed to develop a quantitative predictive scoring model for early recognition and assessment of pediatric sepsis. The model showed excellent predictive performance, with an AUC of 0.886 and high sensitivity and specificity. This model has great clinical significance in detecting and predicting the severity of pediatric sepsis. However, external validation is still needed.