初诊2型糖尿病患者心脏舒张功能障碍预测模型的建立与验证

Establishment and verification of a prediction model for left ventricular diastolic dysfunction in patients newly diagnosed with type 2 diabetes mellitus

目的:探究新诊断的2型糖尿病(T2DM)无症状患者左心室舒张功能障碍(LVDD)的患病率及其相关因素,以便于对LVDD进行早期的识别和干预。方法:选取2014年1月至2021年10月在广州医科大学附属第二医院内分泌科住院的692例新诊断的T2DM患者。应用超声心动图评估心脏舒张功能,并收集患者临床资料及检查结果。探究LVDD的发生率,并采用logistic多因素逐步回归分析构建LVDD的预测模型。使用受试者工作特征曲线(ROC曲线)和校准图评价模型的区分度和准确性,并绘制列线图。结果:初诊T2DM患者的LVDD患病率较高(n=460,66.5%)。多因素logistic回归分析显示,年龄≥60岁、女性、合并高血压、eGFR、cIMT≥1.0 mm为初诊T2DM患者患LVDD的独立影响因素(均为P<0.05)。由此构建的预测模型的ROC曲线下面积(AUC)为0.752(95%CI:0.708~0.797),校准图显示出良好的一致性。结论:LVDD在新诊断的T2DM患者中很常见。本研究构建的预测模型有助于甄别心血管疾病高风险的初诊T2DM患者。

Objective:To investigate the prevalence and associated factors for left ventricular diastolic dysfunction(LVDD)in asymptomatic patients newly diagnosed with type 2 diabetes mellitus(T2DM),so as to facilitate early identification and intervention of LVDD.Methods:Enrolled were a total of 692 patients newly diagnosed with T2DM and treated in Department of Endocrinology,Second Affiliated Hospital of Guangzhou University between January 2014 and October 2021.Echocardiographic measurements were performed to evaluate the cardiac diastolic function.Demographic,clinical and laboratory data of these patients were retrieved.The incidence of LVDD was investigated.Stepwise Logistic regression analysis was used to construct a prediction model for LVDD.Receiver operating characteristic(ROC)curve and calibration plot were utilized to assess the discrimination and calibration power of the prediction model.A nomogram was thereby generated.Results:The prevalence of LVDD was obviously high in patients newly diagnosed with T2DM(n=460,66.5%).Logistic regression analysis showed that age≥60 years,female gender,concomitant hypertension,eGFR,and cIMT≥1.0 mm were independent risk factors for LVDD(all P<0.05).The area under the curve(AUC)of the constructed prediction model was 0.752(95%CI:0.708⁃0.797).The calibration plot showed good consistency.Conclusion:LVDD is common in patients newly diagnosed with T2DM.The prediction model constructed in this study can help to early identify patients newly diagnosed with T2DM who are at high risk of cardiovascular disease.