老年绝经女性冠心病发生风险的列线图预测模型的构建及验证研究

Construction and validation of a nomogram prediction model for the risk of coronary heart disease in elderly postmenopausal women

目的:构建老年绝经女性人群发生冠心病(CHD)的预测模型,并对列线图的预测效率进行验证。方法:回顾性分析2018年3月—2019年2月在安徽省某三级甲等医院心内科就诊并接受冠状动脉(冠脉)造影的639例老年绝经女性的临床资料,根据冠脉造影结果将其分为CHD组(439例)和对照组(200例),同时将研究对象依据7∶3的比例随机分为建模组(447例)和验证组(192例)。在建模组人群中采用多因素回归分析筛选出老年绝经女性人群发生CHD的独立影响因素、构建个性化的列线图模型,并在验证组人群中对模型的预测效率和适用性进行外部验证。结果:建模组人群的多因素logistic回归分析结果表明,每日睡眠时长(OR=0.674,95%CI:0.546~0.833)是老年绝经女性人群发生CHD的独立保护因素(P<0.05),而生育个数(OR=1.798,95%CI:1.391~2.325)、超敏C反应蛋白水平(OR=1.572,95%CI:1.381~1.789)及餐后2h血糖(OR=1.197,95%CI:1.097~1.305)为老年绝经女性人群发生CHD的独立危险因素(P<0.05)。利用上述指标构建列线图预测模型,列线图模型的ROC曲线下面积在建模组及验证中分别为0.815(95%CI:0.773~0.857)、0.823(95%CI:0.755~0.890),建模组及验证组人群中的校准曲线均提示列线图的预测概率与实际发生频率之间具有较好的一致性。临床决策分析曲线提示,当老年绝经女性人群发生CHD的阈值概率处于0.18~0.94时,采用列线图的临床净获益水平最高。结论:老年绝经女性人群发生CHD的列线图预测模型预测能力和适用能力较好,可用于筛查此类人群的CHD发生风险。

Objective To construct a prediction model for coronary heart disease(CHD)in elderly postmenopausal women,and to verify the prediction efficiency of nomogram.Methods There were 639 elderly postmenopausal women who received coronary angiography retrospectively in this study.According to the results of coronary angiography,they were divided into CHD group(439 cases)and control group(200 cases),and the subjects were randomly divided into modeling group(447 cases)and validation group(192 cases)according to the split ratio of 7∶3.Multivariate regression analysis was used in the modeling group to screen out the independent influencing factors of CHD,and a personalized nomogram model was constructed,and the prediction efficiency and applicability of the model were externally verified in the validation group.Results The multivariate logistic regression analysis of the modeling group showed that daily sleep duration(OR=0.674,95%CI:0.546-0.833)was an independent protective factor for the occurrence of CHD in elderly menopausal women(P<0.05),while Number of children born(OR=1.798,95%CI:1.391-2.325),level of high-sensitivity C-reactive protein(OR=1.572,95%CI:1.381-1.789)and 2-h postprandial blood glucose(OR=1.197,95%CI:1.097-1.305)were independent risk factors for CHD in elderly postmenopausal women(P<0.05).Using the above indicators to build a nomogram prediction model,the area under the ROC curve of the nomogram model was 0.815(95%CI:0.773-0.857)and 0.823(95%CI:0.755-0.890)in the modeling group and the validation group,respectively.The calibration curves in both the modeling group and the validation group indicated that the predicted probability of the nomogram was in good agreement with the actual frequency of occurrence.The clinical decision analysis curve suggests that when the threshold probability of coronary heart disease in elderly postmenopausal women is between 0.18 and 0.94,the level of net clinical benefit of using nomogram is the highest.Conclusion The nomogram prediction model of CHD in elderly postmenopausal women has good predictive ability and applicability,and can be used to screen the risk of CHD in this population.