Risk Factors and Prediction Model Construction in Elderly Population with Prediabetes Progressing to Diabetes

HU Pan; CAI Jialuo; ZHANG Yuejuan; LI Sha

doi:10.3969/j.issn.1002-3429.2025.08.010

Abstract：Objective To investigate the risk factors of the elderly population with prediabetes progressing to diabetes and to construct a risk prediction model.Methods A total of 480 elderly population with prediabetes admitted from February 2019 to February 2024 were followed up for 1 year,and were divided into model group and verification group according to the time of admission(February 1,2022).Finally,379 patients completed follow-up,including 284 patients in model group and 95 patients in verification group.In addition,the model group was divided into diabetes group(n=91)and pre-diabetes group(n=193)according to whether they developed diabetes within 1 year of follow-up.Clinical data of research subjects were collected,multivariate logistic regression were used to analyze the risk factors of the elderly pre-diabetic population pro-gressing into diabetes mellitus,and risk prediction models were constructed according to the risk factors.The predictive effica-cy of the models was tested by receiver operating characteristic(ROC)curve.Hosmer-Lemeshow test was used to evaluate the fit degree of the model,and calibration curve was used to evaluate the calibration degree of the model.Results The levels of age≥75 years,body mass index(BMI),hyperlipidemia,family history of diabetes,waist circumference,triglyceride,fast-ing blood glucose,2 h postprandial blood glucose and HBA1c in diabetic group were higher than those in prediabetic group(P＜0.01).Age≥75 years,family history of diabetes and elevated BMI were independent risk factors for progression to dia-betes in the elderly prediabetic population(P＜0.05,P＜0.01).Multivariate Logistic regression analysis predicted the prob-ability of the model P=[1/e(-47.704)+2.568×(age)+0.623×(BMI)+1.971×(family history of diabetes)].The area under the ROC curve(AUC,0.851)to test predictive efficacy of the model was0.851,with sensitivity of 94.51%and specificity of 78.24%.Hosmer-Lemeshow goodness of fit test showed χ2=6.605,P=0.580.The calibration curve showed that the model was well-calibrated.The AUC of the verification group was0.848,the sensitivity was85.21%and the specificity was77.89%.Con-clusion Age≥75 years,family history of diabetes and elevated BMI are risk factors for patients with prediabetes progressing to diabetes,and the prediction model constructed based on this can effectively predict the risk of elderly population with predi-abetes progressing to diabetes.