Page 58 - 南京医科大学学报自然科学版

南京医科大学学报（自然科学版） 第41卷第11期
·1620 · Journal of Nanjing Medical University（Natural Sciences） 2021年11月

·临床研究·

慢性阻塞性肺疾病院前筛查模型的初级构建

李瑞久 ，陈良玉 ，李媛媛 ，倪艾珺 ，孙培莉 1*
1
3
1
2
南京医科大学第一附属医院呼吸与危重症学科，江苏 南京 210029；南京市第二医院结核病科，江苏 南京 210003；南
1 2 3
京大学医学院附属金陵医院呼吸科，江苏 南京 210016

［摘 要］ 目的：通过观察研究慢性阻塞性肺疾病（chronic obstructive pulmonary disease，COPD）患者特异性相关因素，构建
COPD院前筛查模型，以提高COPD高危人群肺功能检查实施率，从而降低COPD的漏诊率及病死率。方法：入组稳定期COPD
患者68例，对照组40例，收集基本资料，同时进行COPD评估测试（COPD assessment test，CAT）、运动中脉率和指脉氧检测及肺
功能测试。采用Logistic回归分析法确定与COPD相关的自变量，构建COPD筛查模型，用Hosmer⁃Lemeshow检验进行校正，用
受试者工作特征曲线下面积（area under receiver operating characteristic curve，AUROC）进行判别。结果：与对照组相比，稳定期
COPD 患者的年龄、吸烟指数、体重指数（body mass index，BMI）、CAT 评分、最低指脉氧（SPO2L ）、最高脉率差异有统计学意义。
吸烟指数、CAT 评分、SPO2L构成筛查模型，用来估计COPD 的概率（probability of COPD，PCOPD ），PCOPD的最佳阈值为57.1%。筛查
模型中影响因素的最佳阈值为吸烟指数≥220支·年、CAT 评分≥7.5分、SPO2L≤93.5%。该模型的AUROC 为0.938，模型判别能
力强，Hosmer⁃Lemeshow拟合度检验P=0.789，模型校正良好。结论：吸烟指数，CAT评分，SpO2L构成了COPD院前筛查模型，该
模型具有良好的区分性和校正，且PCOPD≥57.1%可能为COPD高危患者，建议其进行肺功能检测以明确诊断，从而降低COPD的
漏诊率。吸烟指数≥220支·年、CAT评分≥7.5分、SPO2L≤93.5%时，对COPD筛查具有指导意义。
［关键词］ 慢性阻塞性肺疾病；CAT评分；指脉氧；筛查模型
［中图分类号］ R563 ［文献标志码］ A ［文章编号］ 1007⁃4368（2021）11⁃1620⁃06
doi：10.7655/NYDXBNS20211109

Primary construction of pre ⁃ hospital screening model for chronic obstructive pulmonary
disease

1
1
2
3
LI Ruijiu ，CHEN Liangyu ，LI Yuanyuan ，NI Aijun ，SUN Peili 1*
1 Department of Respiratory and Critical Care Medicine，the First Affiliated Hospital of Nanjing Medical University，
Nanjing 210029；Department of Tuberculosis，Nanjing Second Hospital，Nanjing 210003；Department of Respiratory
2
3
Medicine，Jinling Hospital Affiliated to Medical School of Nanjing University，Nanjing 210016，China
［Abstract］ Objective：By observing and studying the specific related factors of patients with chronic obstructive pulmonary disease
（COPD），a pre ⁃ hospital screening model of COPD was established to improve the implementation rate of pulmonary function
examination in high⁃risk groups of COPD，so as to reduce the rate of missed diagnosis and mortality of COPD. Methods：Total 68
patients with stable COPD and 40 patients in the control group were enrolled. Their basic data were collected，and COPD assessment
test（CAT），pulse rate and finger pulse oxygen measurement during exercise and pulmonary function test were performed at the same
time. Logistic regression analysis was used to determine the independent variables related to COPD，and COPD screening model was
constructed，which was corrected by Hosmer⁃Lemeshow test and discriminated by area under receiver operating characteristic curve
（AUROC）. Results：There were significant differences in age，smoking index，body mass index（BMI），CAT score，lowest finger pulse
oxygen（SPO2L ）and highest pulse rate between the control group and stable COPD patients. Smoking index，CAT score and SPO2L
constituted the screening model，and used to estimate the probability of COPD（PCOPD ）.The best threshold of P COPD was 57.1%. The best
thresholds of influencing factors in the screening model were smoking index ≥220 cigarette ⁃ years，CAT score ≥ 7.5 and SPO2L ≤
93.5% . The AUROC of the model was 0.938 with strong discriminant ability . The model correction was good with the Hosmer ⁃

［基金项目］ 国家重点研发计划专项（2018YFC1311900）；慢性阻塞性肺疾病大数据建设战略合作项目（KD18）
∗
通信作者（Corresponding author），E⁃mail：plisun9419@njmu.edu.cn