中老年2型糖尿病合并骨质疏松症风险预测模型的构建与验证

Construction and Validation of Risk Prediction Model for Type 2 Diabetes Mellitus Complicated with Osteoporosis in the Middle-aged and Elderly

目的:构建中老年2型糖尿病(T2MD)合并骨质疏松症(OP)风险预测模型,并对模型的预测效能进行评价。方法:回顾性分析2019年2月-2021年8月在中国人民解放军联勤保障部队第九〇九医院住院治疗的中老年T2MD患者100例。对其进行分组,并收集所有患者临床资料,构建预测模型。统计分组情况,对中老年T2MD合并OP的影响因素进行单因素及多因素分析,构建中老年T2MD合并OP风险预测模型并对其进行评价。结果:100例患者中,OP组50例,非OP组50例。OP组年龄、糖尿病病程、糖尿病肾病占比均高于非OP组,男性占比、BMI均低于非OP组(P<0.05)。年龄、糖尿病病程、糖尿病肾病均为中老年T2MD合并OP的独立危险因素,男性、体重指数(BMI)均为中老年T2MD合并OP的保护因素(P<0.05)。基于二元Logistic回归分析原理,构建中老年T2MD合并OP风险预测模型,P=1/{1+exp[-(3.986+0.041×年龄-1.639×性别+0.115×糖尿病病程-0.345×BMI+0.843×糖尿病肾病)]}。绘制ROC曲线,AUC为0.798(0.713,0.884),说明模型有较高预测价值,与随机面积0.500比较差异有统计学意义(Z=3.412,P<0.05)。H-L检验χ^(2)=1.967,P=0.302,说明模型有较好的校准度。分别取预测概率0.544(Youden指数最大)和0.500两个切割点,当切割点为0.544,预测敏感度为76.00%,特异度为74.00%,正确率为75.00%,Youden指数为0.50;当预测概率切割点为0.500,模型预测敏感度为72.00%,特异度为68.00%,准确率为70.00%,Youden指数为0.40。结论:成功构建中老年T2MD合并OP风险预测模型,经验证该模型有较好预测效能和校准度。

Objective:To construct a risk prediction model for type 2 diabetes mellitus (T2MD) with osteoporosis (OP) in the middle-aged and elderly patients,and then to evaluate the prediction efficiency of the model.Method:A total of 100 middle-aged and elderly T2MD patients hospitalized in the 909 Hospital of the Joint Service Support Force of the Chinese People’s Liberation Army from February 2019 to August 2021 were retrospectively analyzed.They were grouped,the clinical data of all patients was collected,and a prediction model was builded.The grouping situation was counted,the influencing factors of middle-aged and elderly T2MD complicated with OP were analyzed by single factor and multi factor analysis,the risk prediction model of middle-aged and elderly T2MD complicated with OP was builded and evaluated.Result:Among the 100 patients,there were 50 cases in OP group and 50 cases in non OP group.The age,course of disease in diabetes mellitus and the proportion of diabetes mellitus nephropathy in OP group were higher than those in non OP group,and the proportion of men and body mass index (BMI) were lower than those in non OP group (P<0.05).The age,course of disease diabetes mellitus and diabetes mellitus nephropathy were all independent risk factors for middle-aged and elderly T2MD complicated with OP,while men and BMI were protective factors for middle-aged and elderly T2MD complicated with OP (P<0.05).Based on the principle of binary Logistic regression analysis,the risk prediction model of middle-aged and elderly T2MD complicated with OP was constructed,P=1/{1+exp[-(3.986+0.041)×age-1.639×gender+0.115×course of diabetes-0.345×BMI+0.843×diabetes nephropathy)]}.The ROC curve was drawn,and the AUC was 0.798 (0.713,0.884),indicating that the model had high predictive value,and the difference was statistically significant compared with the random area of 0.500 (Z=3.412,P<0.05).H-L test χ^(2)=1.967,P=0.302,indicating that the model had a good calibration degree.Taking the prediction probability of 0.544 (the largest Youden index) and 0.500 cut points respectively.When the cut point was 0.544,the prediction sensitivity was 76.00%,the specificity was 74.00%,the accuracy was 75.00%,and the Youden index was 0.50;when the cut point of prediction probability was 0.500,the prediction sensitivity of the model was 72.00%,the specificity was 68.00%,the accuracy was 70.00%,and the Youden index was 0.40.Conclusion:The risk prediction model of middle-aged and elderly T2MD complicated with OP is successfully constructed.It is verified that the model has good prediction efficiency and calibration degree.