[Vascular damage-related risk factors for erectile dysfunction in patients with type 2 diabetes mellitus: Analysis based on a nomogram model].

Objective: To analyze vascular damage-related risk factors for ED in patients with type 2 diabetes mellitus (DM) and develop a nomogram for the prediction of the factors.
Methods: A total of 181 patients with type 2 DM were included for sexual function assessment, and the clinical data on vascular damage were retrieved from the patients system. After preprocessing, the data were described by the number and percentage of different types of cases and subjected to statistical analysis with the R software. The Lasso regression model was used to optimize feature selection. On the premise of the sample size required for logistic regression analysis according to the number of events per variable, multivariable logistic regression analysis was performed on the selected variables and a nomogram was developed for diabetes-induced erectile dysfunction (DIED). Then, the performance of the nomogram was evaluated with respect to its calibration, discrimination and clinical utility using Harrell's concordance index (C-index), the calibration plot and decision curve analysis, as well as bootstrapping for internal validation.
Results: ED was diagnosed in 90 (49.7%) of the 181 patients. The risk factors subjected to logistic regression analysis included the duration of DM (OR = 4.440, 95% CI: 1.594－13.105; OR = 7.667, 95% CI: 1.444－48.733), status of carotid intima-media thickness (c-IMT) (OR = 3.767, 95% CI: 1.194－12.691), diabetic retinopathy (DR) (OR = 5.382, 95% CI: 1.373－28.301), diabetic kidney disease (DKD) (OR = 4.959, 95% CI: 1.156－27.728), low-density lipoprotein cholesterol (LDL-C) (OR = 8.210, 95% CI: 2.027－43.507), red blood cell distribution width (RDW) (OR = 2.418, 95% CI: 1.021－5.826), and plasma fibrinogen (Fbg) (OR = 4.649, 95% CI: 2.001－11.339). The C-index of the DIED model was 0.911 (95% CI: 0.869－0.954). The curve representing the performance of the nomogram fit in well with that representing a perfect prediction by the calibration plot. Decision curve analysis indicated that the nomogram was clinically useful for predicting DIED in the type 2 DM patients at the possibility threshold of 6% to 93%.
Conclusions: A nomogram was preliminarily developed for predicting the risk of DIED in type 2 DM patients with respect to the seven independent influencing factors, including the duration of DM, status of c-IMT, DR, DKD, LDL-C, RDW, and Fbg.