METHODS: We included hospitalized patients with T2DM from the endocrinology department of Shenzhen Hospital, Southern Medical University, Shenzhen, China, from January 2019 to 2022. Factors associated with DBD in bivariate analysis with a p < 0.05 were included in a multivariate logistic regression analysis. Multivariate logistic regression analysis was used to determine independent risk factors and to construct a prediction model. The prediction model was presented as the model formula. The receiver operating characteristic (ROC) curve was used to evaluate the predictive value of the above risk factors and the prediction model for DBD. The model was internally verified by Boostrap resampling 1000 times.
RESULTS: Two hundred and eleven patients were included in this study, and they were divided into the DBD group (n = 101) and the non-DBD group (n = 110). Eight variables showed significant significance in the bivariate analysis, including age, diabetic peripheral neuropathy (DPN), glycated hemoglobin (HbA1c), urinary microalbumin (mALB), red blood cell count (RBC), white blood cell count (WBC), absolute neutrophil count (ANC), percentage of monocyte (Mono%). Furthermore, multivariate logistic regression analysis revealed that age (OR [95% CI]: 1.077 [1.042-1.112]), p < 0.001; DPN (OR [95% CI]: 2.373 [1.013-5.561]), p = 0.047; HbA1c (OR [95% CI]: 1.170 [1.029-1.330]), p = 0.017 and ANC (OR [95% CI]: 1.234 [1.059-1.438]), p = 0.007 were independent risk factors for the DBD. The prediction model formula was Logit (p) = -6.611 + 0.074 age + 0.864 DPN + 0.157 HbA 1 c + 0.078 ANC. The area under the ROC curve (AUC) for the four risk factors were 0.676, 0.582, 0.618, and 0.674, respectively. The prediction model predicted DBD with higher accuracy than the individual risk factors, AUC = 0.817 (95% CI: 0.757-0.877), and the sensitivity and specificity were 88.1% and 50.0%, respectively. The model internal validation results showed that the AUC = 0.804 (95% CI: 0.707-0.901), and the calibration curve is close to the ideal diagonal line.
CONCLUSIONS: Age, DPN, HbA1c, and ANC were risk factors for DBD. The prediction model constructed based on the four risk factors had a good predictive value for predicting the occurrence of DBD.
方法:我们纳入深圳医院内分泌科住院的T2DM患者,南方医科大学,深圳,中国,从2019年1月到2022年。双变量分析中与DBD相关的因素包括在多变量逻辑回归分析中,p<0.05。采用多因素logistic回归分析确定独立危险因素并构建预测模型。预测模型以模型公式表示。采用受试者工作特征(ROC)曲线评价上述危险因素的预测值和DBD的预测模型。通过Boostrap重采样1000次对该模型进行了内部验证。
结果:这项研究纳入了211例患者,分为DBD组(n=101)和非DBD组(n=110)。八个变量在双变量分析中显示出显着意义,包括年龄,糖尿病周围神经病变(DPN),糖化血红蛋白(HbA1c),尿微量白蛋白(mALB),红细胞计数(RBC),白细胞计数(WBC),中性粒细胞绝对计数(ANC),单核细胞百分比(Mono%)。此外,多因素logistic回归分析显示年龄(OR[95%CI]:1.077[1.042-1.112]),p<0.001;DPN(OR[95%CI]:2.373[1.013-5.561]),p=0.047;HbA1c(OR[95%CI]:1.170[1.029-1.330]),p=0.017和ANC(OR[95%CI]:1.234[1.059-1.438]),p=0.007是DBD的独立危险因素。预测模型公式为Logit(p)=-6.611+0.074年龄+0.864DPN+0.157HbA1c+0.078ANC。4个危险因素的ROC曲线下面积(AUC)分别为0.676、0.582、0.618和0.674。预测模型预测DBD的准确性高于单个风险因素,AUC=0.817(95%CI:0.757-0.877),敏感性和特异性分别为88.1%和50.0%,分别。模型内部验证结果显示,AUC=0.804(95%CI:0.707-0.901),校准曲线接近理想对角线。
结论:年龄,DPN,HbA1c,ANC是DBD的危险因素。基于4个危险因素构建的预测模型对DBD的发生具有较好的预测价值。