基于Nomogram预测模型分析住院患儿多重耐药菌感染的危险因素

Analysis of Risk Factors for Multidrug-Resistant Bacterial Infections in Hospitalized Children Based on Nomogram Prediction Model

摘要: 目的：分析住院患儿多重耐药菌(MDRO)感染的分布情况及医院感染的危险因素，方法：收集2020年3月~2023年3月中山市人民医院儿科住院病房共61例患儿的临床资料进行回顾性分析。收集MDRO感染患儿61例，同时匹配各病区非MDRO感染患儿61例。收集各病区感染患儿的基本感染临床资料，对所有患儿资料进行单因素及多因素Logistic回归分析。利用R4.3.1软件构建MDRO感染风险的列线图模型，并通过受试者工作特征曲线和校准曲线评价模型的预测拟合效果。结果：PICU和NICU是MDRO常发生的病区，分别占52.46%和29.51%，下呼吸道和血液系统是MDRO常发生的感染部位，分别占42.62%和29.50%。耐青耐碳青霉烯类大肠埃希菌(CRE)和耐甲氧西林金黄色葡萄球菌(MASA)是最常见的感染菌株，分别占24%和17%。长时间住院、长时间使用抗生素和侵入性操作是儿科各病区MDRO感染的独立危险因素(P < 0.05)。基于Logistic回归分析结果构建的列线图模型，受试者工作特征曲线下面积为0.834，95%置信区间：0.766~0.901，Hosmer-Lemeshow检验显示χ² = 139.69，P = 0.332，且校准曲线预测发生率与实测发生率之间偏差小，模型校准度好、预测拟合效果好。结论：长时间住院、长时间使用抗生素和侵入性操作是儿科各病区MDRO感染的独立危险因素感染的独立危险因素。基于危险因素构建的风险预测模型具有良好的预测能力，可以为及时采取预防性感染控制措施提供参考。

Abstract: Objective: To analyze the distribution of multidrug-resistant bacteria (MDRO) infection in hospitalized children and the risk factors for hospital acquired infections. Method: Clinical data of 61 hospitalized children in the pediatric ward of Zhongshan People’s Hospital from March 2020 to March 2023 were collected for retrospective analysis. Collect 61 cases of MDRO infected children and match 61 cases of non MDRO infected children in each ward. Collect basic clinical data of infected children in each ward, and conduct univariate and multivariate logistic regression analysis on all patient data. Construct a column chart model for MDRO infection risk using R4.3.1 software, and evaluate the predictive fitting effect of the model through subject working characteristic curves and calibration curves. As a result, PICU and NICU are the wards where MDRO often occurs, accounting for 52.46% and 29.51% respectively. The lower respiratory tract and blood system are the sites of infection where MDRO often occurs, accounting for 42.62% and 29.50% respectively. Penicillin resistant Escherichia coli (CRE) and methicillin-resistant Staphylococcus aureus (MASA) are the most common strains of infection, accounting for 24% and 17%, respectively. Long term hospitalization, prolonged use of antibiotics, and invasive procedures are independent risk factors for MDRO infection in various pediatric wards (P < 0.05). The column chart model constructed based on the results of logistic regression analysis showed that the area under the working characteristic curve of the subjects was 0.834, with a 95% confidence interval of 0.766~0.901. The Hosmer Lemeshow test showed χ² = 139.69, P = 0.332, and the deviation between the predicted and measured incidence rates of the calibration curve was small, indicating good model calibration and prediction fitting effect. Conclusion: Long term hospitalization, prolonged use of antibiotics, and invasive procedures are independent risk factors for MDRO infection in various pediatric wards. The risk prediction model based on risk factors has good predictive ability and can provide reference for timely adoption of preventive infection control measures.