老年糖尿病病人衰弱风险预测模型的建立与验证

Establishment and validation of a frailty risk prediction model for elderly patients with diabetes mellitus

目的:探讨糖尿病病人的衰弱现状及影响因素,并构建老年糖尿病病人衰弱风险预测模型。方法:选取2020年8月—2021年6月于新疆某三级甲等医院内分泌科住院的糖尿病病人566例,依据入院顺序按6∶4分为训练集351例和验证集215例,收集研究对象的一般资料、实验室检查结果,运用衰弱表型(FP)将研究对象分为无衰弱组和衰弱组。应用LASSO回归筛选变量,通过多因素Logistic回归建立预测模型,应用受试者工作特征(ROC)曲线、校准曲线评价预测模型的区分度和校准度,临床决策曲线(DCA)评估预测模型的临床有效性。结果:LASSO回归筛选出年龄、病程、糖化血红蛋白(HbA1c)、低血糖次数、糖尿病足、婚姻状况6个预测变量;Logistic回归结果显示,年龄、HbA1c、病程、低血糖次数、糖尿病足及婚姻状况是糖尿病病人衰弱的影响因素(P<0.05);根据回归模型构建预测模型,ROC曲线分析显示,预测模型在训练集的ROC曲线下面积(AUC)为0.790[95%CI(0.742,0.838)],在验证集中的AUC为0.703[95%CI(0.628,0.778)];拟合优度检验结果显示有较好的拟合度,且预测模型的净收益率较高。结论:基于年龄、病程、HbA1c、低糖次数、糖尿病足、婚姻6个变量构建的预测模型可用于预测老年糖尿病病人衰弱的发生风险。

Objective:To explore the status and influencing factors of frailty in elderly patients with diabetes mellitus,and to construct a risk prediction model for frailty.Methods:A total of 566 patients with diabetes mellitus who were hospitalized in the department of endocrinology of a tertiary Grade A hospital in Xinjiang from August 2020 to June 2021 were selected and divided into a training set(n=351)and a validation set(n=215)according to the order of admission by 6∶4.The general data and laboratory examination results of the research objects were collected.The receiver operating characteristic(ROC)curve and calibration curve were used to evaluate the discrimination and calibration of the prediction model.The clinical decision curve(DCA)was used to evaluate the clinical effectiveness of the prediction model.Results:Six predictive variables including age,duration of disease,HbA1c,times of hypoglycemia,diabetic foot and marriage were screened out by LASSO regression.Logistic regression results showed that age,HbA1c,course of disease,frequency of hypoglycemia,diabetic foot and marriage were the influencing factors of frailty in diabetic patients(P<0.05).ROC curve analysis showed that the AUC of the prediction model was 0.790(95%CI 0.742,0.838)in the training set and 0.703[95%CI(0.628,0.778)]in the validation set.The goodness of fit test showed that the model had a good fit,and DCA showed that the net return rate of the prediction model was high.Conclusion:The prediction model based on age,duration of diabetes,HbA1c,frequency of hypoglycemia,diabetic foot,and marriage can be used to predict the risk of frailty in elderly diabetic patients.