Construction of A Risk Prediction Model for Children with Community-Acquired Pneumonia from Mild to Severe During Hospitalization

GUO Huirong; JIANG Jilong; GE Shaofeng; WANG Yang; HUANG Qiumei

doi:10.3639/j.issn.2095-3097.2025.06.012

Abstract：Objective To analyze the risk factors for progression of children with community-acquired pneumonia(CAP)from mild to severe during hospitalization,and to construct its risk prediction model.Methods The clinical data of 475 chil-dren with CAP admitted from January 2022 to December 2024 Department of Pediatrics,Nanping First Hospital,Fujian Medical University were retrospectively analyzed.The distribution characteristics of pathogens in children with CAP were analyzed,and the children were divided into mild group(337 children)and severe group(138 children)according to their disease status after admission.General clinical data,including neutrophil/lymphocyte ratio(NLR),25-hydroxyvitamin D3[25-(OH)D3],and pe-diatric early warning score(PEWS)were compared between the two groups.Multivariate Logistic regression analysis was used to evaluate the influencing factors of severe CAP in children.The nomogram was used to construct a risk prediction model for severe CAP in children.The bootstrap method was used for internal validation,repeatedly sampling 100 times,to evaluate the predictive value of the model using a calibration curve.Results The positive rate of pathogenic examination in 475 children with CAP was 54.74%(260/475).Single pathogen infection predominated(76.54%,199/260)with Mycoplasma pneumoniae be-ing most common(52.26%,104/199).Mixed pathogen infection accounted for 23.46%(61/260).Of the 475 children with CAP,138(29.05%)progressed to severe disease after admission.In the severe group,the detection rate of bacteria+virus,detection rate of bacteria+Mycoplasma pneumoniae,neutrophil proportion,NLR,C-reactive protein(CRP)level and PEWS score were higher than those in the mild group(P<0.05),while lymphocyte proportion,albumin level,and 25-(OH)D3 level were lower than those in the mild group(P<0.05).Multivariate logistic forward stepwise regression analysis showed that NLR(OR=2.046,95%CI:1.200-3.490,P<0.05),CRP(OR=1.516,95%CI:1.260-1.823,P<0.05),and PEWS score(OR=235.077,95%CI:29.282-1 887.184,P<0.05)were independent risk factors for severe CAP in children,and albumin(OR=0.846,95%CI:0.763-0.937,P<0.05)and 25-(OH)D3(OR=0.757,95%CI:0.666-0.862,P<0.05)were independent protective factors.A risk nomo-gram early warning model for severe CAP in children was constructed using these five independent factors.Internal validation showed that the calibration curve approached the ideal curve.Conclusion Children with CAP are mostly infected with a single pathogen.High NLR,C-reactive protein levels,and PEWS score early in hospitalization may increase the risk of progression to severe CAP,while high albumin and high 25-(OH)D3 levels may reduce this risk.The five-factor nomogram model facilitates early prediction of severe CAP,aiding timely clinical intervention.