影响重症肺炎儿童院内死亡的相关因素及风险预测模型构建

Factors Affecting In-Hospital Mortality in Children With Severe Pneumonia and Construction of a Risk Prediction Model

目的 研究影响重症肺炎患儿院内死亡的相关因素,并构建Nomogram预测模型。方法 选择2019年1月1日至2022年12月31日于潍坊市妇幼保健院住院的130例重症肺炎患儿,按照是否于院内死亡分为生存组(95例)和院内死亡组(35例)。收集患儿的相关临床资料,构建多因素Logistic回归模型,分析重症肺炎患儿院内死亡的相关影响因素,并绘制Nomogram列线图,通过受试者工作特征(ROC)曲线进行内部验证和效能评估,对该模型的可行性进行评价。结果 纳入研究的130例患儿中有35例死亡,多因素Logistic回归分析显示,C反应蛋白(CRP)(OR=1.206)、降钙素原(PCT)(OR=12.975)、乳酸脱氢酶(LDH)(OR=1.010)、凝血酶原时间(PT)(OR=1.800)、D-二聚体(D-D)(OR=5.072)是重症肺炎患儿发生院内死亡的独立影响因素。内部验证结果示,重症肺炎患儿院内死亡Nomogram预测模型曲线下面积(AUC)为0.940 (95%CI:0.903~0.977),敏感度97.14%,特异度为78.95%,截断值为0.18,约登指数为0.76。结论 CRP、PCT、LDH、PT、D-D为重症肺炎儿童院内死亡的独立影响因素,基于此绘制的Nomogram预测模型能够简洁、直观地为重症肺炎儿童提供个体化的风险预测,具有较好的临床效能。

Objective To study the relevant factors affecting the in-hospital death of children with severe pneumonia and construct a Nomogram prediction model.Methods A total of 130 children with severe pneumonia hospitalized in Weifang Maternal and Child Health Hospital from January 1,2019,to December 31,2022,were selected and divided into a survival group(95 cases)and an in-hospital death group(35 cases)according to whether they died in the hospital.The relevant clinical data of the children were collected.A multifactor logistic regression model was constructed to analyze the risk factors of in-hospital death in children with severe pneumonia,and a Nomogram column chart was developed.The feasibility of the model was evaluated by internal validation and efficacy assessment through ROC curves.Results Of the 130 children included in the study,35 children died during hospitalization.Multivariate logistic regression analysis showed that CRP(OR=1.206),PCT(OR=12.975),LDH(OR=1.010),PT(OR=1.800),and D-D(OR=5.072)were independent risk factors for mortality.The model was internally validated,and the results showed that the area under the curve of the Nomogram prediction model for in-hospital mortality in children with severe pneumonia was 0.940(95%CI:0.903~0.977),with a sensitivity of 97.14%,a specificity of 78.95%,a cutoff value of 0.18,and a Yoden index of 0.76.Conclusion The results indicated that CRP,PCT,LDH,PT,and D-D were independent risk factors for death during hospitalization in children with severe pneumonia.The Nomogram prediction model based on these factors provides individualized risk prediction for children with severe pneumonia in a concise and intuitive manner,demonstrating good clinical efficacy.