摘要
目的建立预测2型糖尿病(T2DM)外周血管疾病(PAD)发病风险的列线图模型。方法回顾性收集2017年12月至2020年8月于石家庄市第二医院就诊的1851例T2DM患者,采用简单随机抽样方法按3∶1比例分为训练集(1389例)和验证集(462例)。收集受试者的一般资料、体格检查指标、生化指标。收集的资料用于评估T2DM合并PAD的风险。通过运行RStudio软件,使用LASSO回归分析以及交叉验证,来筛选最佳预测因子,采用多因素logistic回归建立预测模型,引入从LASSO回归筛选出的预测因子,进而构建预测T2DM患者PAD风险列线图模型,采用受试者工作特征(ROC)曲线、校正及决策曲线评估模型的效能。结果在训练集中,LASSO结合logistic回归分析结果显示,年龄(OR=1.034,95%CI 1.020~1.047)、吸烟(OR=2.363,95%CI 1.720~2.804)、饮酒(OR=3.692,95%CI 2.618~5.230)、糖尿病病程(OR=1.040,95%CI 1.006~1.076)、收缩压(OR=1.026,95%CI 1.014~1.039)、高密度脂蛋白胆固醇(OR=0.467,95%CI 0.276~0.779)、低密度脂蛋白胆固醇(OR=1.459,95%CI 1.239~1.721)、白细胞计数(OR=1.137,95%CI 1.027~1.260)、血小板分布宽度(OR=1.285,95%CI 1.202~1.375)、大血小板比率(OR=0.911,95%CI 0.890~0.931)和降压药使用(OR=0.422,95%CI 0.322~0.551)是T2DM合并PAD的影响因素,差异均具有统计学意义(P<0.001)。训练集和验证集预测T2DM患者PAD发病风险的AUC、灵敏度、特异度分别为0.819(95%CI 0.796~0.842,P<0.001)、70.5%、82.9%和0.821(95%CI 0.779~0.862,P<0.001)、78.3%、75.4%。临床决策曲线显示,在训练集和验证集中,当PAD风险阈值概率分别在31%~81%和37%~94%之间时,预测T2DM患者PAD发病风险的净收益更高。结论本研究成功构建一种准确性高、性能优良的预测T2DM患者PAD发病风险的实用列线图模型。
Objective To establish a visualization tool for evaluating on risk prediction of peripheral vascular disease(PAD)in patients with type 2 diabetes mellitus(T2DM).Methods A retrospective analysis of 1851 T2DM patients treated in Shijiazhuang Second Hospital during December 2017 to August 2020 were randomly sampled into training set(1389 cases)and validation set(462 cases)by a ratio of 3∶1.General subject data,physical examination indicators,and biochemical indicators were collected.The collected data was used to assess the risk of T2DM combined with PAD.In order to construct a nomogram model to predict PAD risk in T2DM patients,LASSO regression analysis and cross-validation were used to screen the best predictors by RStudio software,and then multivariate logistic regression was used to introduce the predictors selected from LASSO regression.Receiver operating characteristic(ROC)curve,correction and decision curve were used to evaluate the efficacy of the model.Results In training set,the results of the LASSO combined with the logistic regression analysis showed that age(OR=1.034,95%CI 1.020-1.047),smoking(OR=2.363,95%CI 1.720-2.804),drinking(OR=3.692,95%CI 2.618-5.230),duration of diabetes(OR=1.040,95%CI 1.006-1.076),systolic blood pressure(OR=1.026,95%CI 1.014-1.039),high-density lipoprotein-cholesterol(OR=0.467,95%CI 0.276-0.779),low-density lipoprotein-cholesterol(OR=1.459,95%CI 1.239-1.721),white blood cell count(OR=1.137,95%CI 1.027-1.260),platelet distribution width(OR=1.285,95%CI 1.202-1.375),large-platelet ratio(OR=0.911,95%CI 0.890-0.931),and the use of antihypertensive drugs(OR=0.422,95%CI 0.322-0.551)were the influencing factors of T2DM with PAD,and all differences were statistically significant(P<0.001).The area under curve(AUC),sensitivity,and specificity of the training and validation sets for predicting the occurrence of PAD in patients with T2DM were 0.819(95%CI 0.796-0.842,P<0.001),70.5%,82.9%,and 0.821(95%CI 0.779-0.862,P<0.001),78.3%,and 75.4%,respectively.The clinical decision curve showed that the net gain for predicting the risk of PAD onset in T2DM was higher when the PAD risk threshold probability was between 31%and 81%,37%and 94%,respectively.Conclusions This study successfully constructed a practical nomogram model with high accuracy and excellent performance to predict the risk of PAD onset in T2DM patients.
作者
刘鑫
苏鹏
陈晋波
李素彦
李旭蕊
梁小华
马冬
Liu Xin;Su Peng;Chen Jinbo;Li Suyan;Li Xurui;Liang Xiaohua;Ma Dong(Department of General Medicine,Shijiazhuang Second Hospital,Shijiazhuang 050051,China;School of Public Health,North China University of Technology,Tangshan 063210,China;Department of General Medicine,Hebei Provincial People′s Hospital,Shijiazhuang 050057,China)
出处
《中华糖尿病杂志》
CAS
CSCD
北大核心
2022年第6期570-576,共7页
CHINESE JOURNAL OF DIABETES MELLITUS
基金
国家自然科学基金(81700416)
石家庄市科学技术研究与发展计划项目(191460933)。
关键词
糖尿病
2型
外周血管疾病
列线图
Diabetes mellitus,type 2
Peripheral arterial diseases
Nomograms
