Abstract:

Objective　To investigate the primary risk factors of pulmonary infection (PI) in patients during the recovery phase of cerebral hemorrhage and to construct and validate a nomogram predictive model for assessing the risk of PI. Methods　The clinical data of 761 patients in the recovery phase of cerebral hemorrhage admitted to Department of Hyperbaric Oxygen Medicine, Hefei Second People's Hospital from September 1, 2020 to December 31, 2022 were retrospectively collected. Based on their clinical manifestations and chest radiography, the patients were divided into a PI group (257 cases), including 116 females and 141 males who were 60 (53, 70) years old, and a non-PI group (504 cases), including 204 females and 300 males who were 58.5 (51, 69) years old. Mann-Whitney U and χ² tests were used. The univariate logistic regression was used to screen the risk factors initially, and the risk factors were optimized by the Least Absolute Shrinkage and Selection Operator (LASSO) regression. The model's discriminative ability, calibration, and clinical utility were evaluated by the area (AUC) under the receiver operating characteristic curve (ROC), calibration curve, and decision curve analysis (DCA). Results　Age, use of antimicrobial drugs, consciousness disorder, tracheotomy, dysphagia, bedridden duration, and nasogastric feeding were independent risk factors for pulmonary infection in the patients during the recovery phase of cerebral hemorrhage (OR=2.406,5.507,8.878,8.931,2.709,3.402, and 2.528; all P<0.05), and hyperbaric oxygenation was the protective factor (OR=0.202; P<0.05). The AUC of the ROC of the predictive model was 0.901 (95%CI 0.878-0.924). The bootstrap resampling (1 000 times) yielded an AUC of 0.900 (95%CI 0.877-0.923). The Hosmer-Lemeshow test showed a good fit (χ²=4.28, P=0.982). Conclusion　The nomogram predictive model developed in this study is highly accurate and facilitates the early identification and risk prediction of pulmonary infection in patients during the recovery phase of cerebral hemorrhage.

Key words:

Intracerebral hemorrhage, Convalescent period, Pulmonary infection, Clinical prediction model, Nomogram

摘要：

目的　探讨脑出血恢复期患者发生肺部感染（PI）的主要危险因素，构建并验证预测PI发生风险的列线图预测模型。方法　回顾性收集2020年9月1日至2022年12月31日住院于合肥市第二人民医院高压氧医学科的761例脑出血恢复期患者临床资料。根据临床表现与胸片确定发生肺部感染情况，将患者分为PI组［257例，其中女116例、男141例，年龄60（53，70）岁］与非PI组［504例，其中女204例、男300例，年龄58.5（51，69）岁］。统计学方法采用Mann-Whitney U检验、χ²检验。采用单因素logistic回归初筛出危险因素，随后采用最小绝对收缩和选择运算（LASSO）回归进行危险因素优化，最终将优化过的危险因素纳入多因素logistic回归中筛选出独立危险因素并构建列线图。采用受试者操作特征曲线（ROC）的曲线下面积（AUC）、校准曲线以及决策曲线分析（DCA）来确定模型的区分度、校准度和临床实用性。结果　年龄、抗菌药物使用、意识障碍、气管切开、吞咽障碍、卧床时间、鼻饲均是脑出血恢复期患者并发肺部感染的独立危险因素（OR=2.406、5.507、8.878、8.931、2.709、3.402、2.528，均P<0.05），高压氧治疗是减少肺部感染的保护因素（OR=0.202，P<0.05）。预测模型ROC的AUC为0.901（95%CI 0.878～0.924），采Bootstrap法从原始数据中重复抽样1 000次后AUC为0.900（95%CI 0.877～0.923）；Hosmer–Lemeshow检验：χ²=4.28，P=0.982。结论　本研究构建的列线图预测模型具有较高的准确性，方便识别脑出血恢复期患者并发肺部感染的早期识别和风险预测。

关键词:

脑出血, 恢复期, 肺部感染, 临床预测模型, 列线图