女性特发性肺间质纤维化并发肺气肿的危险因素分析及相关预测模型构建  

Analysis of risk factors for emphysema in female patients with idiopathic pulmonary interstitial fibrosis and construction of related prediction models

作　　者：邓文静[1] 樊娜[1] 张娜[1] 王春亚[1] 方萍[1] 史红阳[1] 杨拴盈[1] DENG Wenjing;FAN Na;ZHANG Na;WANG Chunya;FANG Ping;SHI Hongyang;YANG Shuanying(Department of Respiratory and Critical Care Medicine,The Second Affiliated Hospital of Xi′an Jiaotong University,Xi′an 710004,China)

机构地区：[1]西安交通大学第二附属医院呼吸与危重症医学科,陕西西安710004

出　　处：《西部医学》2023年第11期1615-1619,1624,共6页Medical Journal of West China

基　　金：陕西省社会发展科技公关项目(2016SF-249)。

摘　　要：目的探讨女性特发性肺间质纤维化(IPF)并发肺气肿的危险因素,并构建列线图风险预测模型。方法回顾性选取2017年1月—2020年1月我院收治的132例女性IPF并发肺气肿患者纳为观察组,另选取同期我院收治的186例女性单纯IPF患者纳为对照组。收集两组患者的临床资料、肺功能指标、动脉血氧分压、胸部高分辨CT(HRCT)表现类型及血清指标。采用单因素及多因素Logistic回归分析筛选出女性IPF并发肺气肿的危险因素;应用(4.1.2)R建立列线图预测模型;采用ROC曲线评估模型区分度;采用Bootstrap法进行内部验证,并以校正曲线及H-L检验评估其有效性。前瞻性选取2020年2月—2022年1月在我院接受治疗的163例女性IPF患者进行外部验证。结果Logistic回归分析结果显示,吸烟史、反流性食管炎、二手烟暴露、工作中接触粉尘是女性IPF合并肺气肿的危险因素(P<0.05);基于上述各危险因素成功建立女性IPF并发肺气肿列线图模型;ROC曲线分析结果显示,AUC可达0.811(95%CI:0.764~0.859),灵敏度为0.780,特异度为0.710;采用Bootstrap内部验证显示,平均绝对误差(MAE)为0.028,校正曲线趋近于理想曲线;H-L检验提示模型不存在过度拟合现象(χ^(2)=5.764,P=0.674)。外部验证结果显示,ROC曲线AUC可达0.782(95%CI:0.711~0.854),灵敏度为0.770,特异度为0.696;校准曲线MAE为0.046,且校正曲线趋近于理想曲线;H-L检验不存在过度拟合现象(χ^(2)=7.367,P=0.497)。结论吸烟史、反流性食管炎、二手烟暴露、工作中接触粉是女性IPF并发肺气肿的危险因素,基于上述危险因素构建女性IPF并发肺气肿列线图风险预测模型具有良好的精准性与区分度。Objective To investigate the risk factors of female idiopathic pulmonary fibrosis(IPF)complicated with emphysema,and construct a nomogram risk prediction model.Methods 132 female patients with IPF complicated with emphysema treated in our hospital from January 2017 to January 2020 were retrospectively selected and included in the IPF complicated with emphysema group,and 186 female patients with pure IPF who were treated in our hospital during the same period were selected.The clinical data,pulmonary function indexes,arterial blood oxygen partial pressure,chest high-resolution CT(HRCT)manifestation types and serum indexes of the two groups of patients were collected.Univariate and multivariate Logistic regression analysis was used to screen out the risk factors of female IPF complicated with emphysema.(4.1.2)R was used to establish a nomogram prediction model.ROC curve was used to evaluate the model discrimination.Bootstrap method was used for internal verification.The validity was evaluated by calibration curve and H-L test.163 female IPF patients who were treated in our hospital from February 2020 to January 2022 were prospectively selected for external validation.Results Logistic regression analysis showed that smoking history,reflux esophagitis,exposure to secondhand smoke and exposure to dust at work were risk factors for IPF combined with emphysema in females(P<0.05).Based on the above risk factors,the female IPF with emphysema was successfully established.ROC curve analysis showed that the AUC was 0.811(95%CI:0.764~0.859),the sensitivity was 0.780,and the specificity was 0.710.Bootstrap internal verification shows that the average absolute error(MAE)is 0.028,and the correction curve is close to the ideal curve.H-L test indicated that there was no overfitting phenomenon in the model(χ^(2)=5.764,P=0.674).External verification results showed that the ROC curve AUC could reach 0.782(95%CI:0.711~0.854),the sensitivity was 0.770,and the specificity was 0.696.The calibration curve MAE is 0.046,and the calibration

关 键 词：特发性肺间质纤维化 肺气肿 危险因素 风险模型 

分 类 号：R563[医药卫生—呼吸系统]