摘要
Abstract
Objective An retrospective cohort study was conducted to analyze the risk factors of diabetes foot(DFU)recurrence and build a prediction model by time-dependent ROC combined survival analysis.Methods Collect the clinical data of DFU patients hospitalized in author's department from January 2018 to January 2023 and follow up the recurrence of DFU.Design nineteen research factors:age,gender,whether hypertension is combined,alanine aminotransferase(ALT),aspartate aminotransferase(AST),alkaline phosphatase(ALP),glutamyl transpeptidase(GGT),lactate dehydrogenase(LDH),albumin,blood glucose control,urea nitrogen,creatinine value,lower limb vascular disease,Wagner grade,whether there is bone destruction,whether there is intervention surgery,whether there is amputation,whether there is skin grafting,whether smoking.The continuous data is converted into categorical data by calculating the cutoff point by time-dependent ROC,and the recurrence rate is calculated by K-M method.The log-rank test is used for single-factor analysis of risk factors,and the COX risk regression model is used for multi-factor analysis.Results A total of 74 DFU patients were included.The recurrence rates of 1,2,3 and 5 years after healing of DFU were 12.1%,24.8%,33.8%and 43.3%.Smoking(P=0.002),Wagner grade(P=0.004),bone destruction(P=0.038),lower limb vascular disease(P＜0.001),age(P=0.034),creatinine value(P=0.045),amputation(P=0.022)are risk factors for DFU recurrence,among which smoking(P＜0.001,HR=19.805),age≥73 years old(P=0.009,HR=9.429),and blood creatinine value≥66 μ mol/L(P=0.024,HR=5.423)are independent risk factors for DFU recurrence.The prediction model of the DFU recurrence is:Y=2.986X1+2.244X2+1.691X3(smoking=X1,age level=X2,creatinine level=X3),C-index=0.888.The accuracy of the model is high.Conclusion The prediction model based on time-dependent ROC combined survival analysis has high accuracy in judging the recurrence rate of DFU.关键词
糖尿病足/复发/危险因素/生存分析/时间依赖ROCKey words
diabetes foot/recurrence/risk factors/survival analysis/time-dependent ROC分类
医药卫生
