摘要
目的探讨非小细胞肺癌(NSCLC)患者化学治疗(化疗)期间发生感染的相关因素,并构建预测模型。方法选取2020年5月—2023年1月某医院收治的387例NSCLC化疗患者为建模集,依据化疗期间是否发生感染分为感染组和对照组,收集患者一般资料、诊疗信息、生化检查结果行单因素分析,并将差异有统计学意义的项目纳入logistic回归分析,筛选出感染相关因素。应用R软件中RMS包构建NSCLC化疗患者感染风险预测模型,通过受试者工作特征(ROC)曲线下面积(AUC)及校准曲线验证模型预测患者感染的区分度及一致性,通过决策曲线分析(DCA)评估临床获益情况。另纳入2023年2月—2024年10月该院收治的165例NSCLC化疗患者为验证集,对模型开展外部验证。结果建模集387例患者中,93例化疗期间发生感染,感染发病率为24.03%;主要感染部位为呼吸(39.79%)、消化系统(24.73%)。建模集两组患者年龄、合并慢性呼吸疾病情况、肿瘤-淋巴结-转移分期系统(TNM分期)、化疗周期、联合放疗情况、侵入性操作次数及使用糖皮质激素情况差异均有统计学意义(均P<0.05)。logistic回归分析显示,NSCLC化疗患者合并感染的主要危险因素有年龄大、TNM分期为Ⅲ~Ⅳ期、化疗周期长、联合放射治疗(放疗)、侵入性操作次数>2次、使用糖皮质激素共6项。采用6项预测指标构建NSCLC化疗患者感染的风险预测模型,结果显示,建模集、验证集AUC分别为0.792、0.773,预测感染的概率与实际概率相近,建模集拟合优度HL检验χ^(2)=8.760,P=0.316,验证集拟合优度HL检验χ^(2)=9.013,P=0.287。DCA显示,模型有较高的临床获益度。结论基于年龄、TNM分期、化疗周期、联合放疗、侵入性操作次数、使用糖皮质激素6项指标构建的列线图预测模型可较好地预测NSCLC化疗患者合并感染。
Objective To explore the relevant factors for infection in patients with non-small cell lung cancer(NSCLC)during chemotherapy period,and construct a prediction model.Methods 387 NSCLC chemotherapy patients who admitted to a hospital from May 2020 to January 2023 were selected as the modeling set.They were divided into an infection group and a control group based on the occurrence of infection during chemotherapy period.General data,diagnosis and treatment information,and biochemical examination results of patients were collected for univariate analysis.Items with statistically significant differences were included in logistic regression analysis,and factors related to infection were screened out.A infection risk prediction model for NSCLC chemotherapy patients was constructed using the RMS package in R-based software.The discrimination and consistency of the model in infection prediction were validated through the area under the curve(AUC)of the receiver operating characteristic(ROC)curve and the calibration curve.Clinical benefits were evaluated through the decision curve analysis(DCA).A total of 165 NSCLC chemotherapy patients who admitted to the hospital from February 2023 to October 2024 were included as the validation set,and external validation of the model was conducted.Results Among the 387 patients in the modeling set,93 cases developed infection during chemotherapy period,with an infection rate of 24.03%.The main infection sites were respiratory system(39.79%)and digestive system(24.73%).There were statistically significant differences in age,combined chronic respiratory disease,tumor-node-metastasis-based(TNM)staging system,chemotherapy cycle,combined radiotherapy,episode number of invasive procedures,and glucocorticoid use between two groups of patients in the modeling set(all P<0.05).Logistic regression analysis showed 6 major risk factors for co-infection in NSCLC chemotherapy patients,namely age,TNM stageⅢ-Ⅳ,long chemotherapy cycle,combined radiotherapy,more than 2 episodes of invasive procedures,and glucocorticoid use.A risk prediction model for the infection in NSCLC chemotherapy patients was constructed using the 6 predictive indicators.The results showed that the AUCs of the modeling set and the validation set were 0.792 and 0.773,respectively.The predicted probability of infection was close to the actual probability.The HL test of goodness of fit of the modeling set showedχ2 value of 8.760 and P value of 0.316,and those of the modeling set were 9.013 and 0.287,respectively.DCA revealed a high clinical benefit of the model.Conclusion A nomogram prediction model based on age,TNM stage,chemotherapy cycle,combined radiotherapy,episodes of invasive procedures,and glucocorticoid use can effectively predict co-infection in NSCLC chemotherapy patients.
作者
徐敏洁
王洪
王雅杰
XU Minjie;WANG Hong;WANG Yajie(Department of General Medicine,The First Affiliated Hospital with Nanjing Medical University,Nanjing 210029,China;Department of Oncology,The First Affiliated Hospital with Nanjing Medical University,Nanjing 210029,China)
出处
《中国感染控制杂志》
北大核心
2025年第8期1120-1126,共7页
Chinese Journal of Infection Control
关键词
非小细胞肺癌
感染
危险因素
列线图
预测模型
non-small cell lung cancer
infection
risk factor
nomogram
prediction model
