Construction and Evaluation of a Predictive Nomogram for Identifying Premature Failure of Arteriovenous Fistulas in Elderly Diabetic Patients

Abstract

Background: This research aimed to identify risk factors contributing to premature maturation of arteriovenous fistulas (AVF) in elderly diabetic patients and develop a clinical prediction model.

Methods: We conducted a retrospective review of 548 geriatric diabetic patients who underwent AVF creation for maintenance hemodialysis (MHD) at Baoding No 1 Central Hospital between January 2011 and December 2023. Patients were divided into mature (386) and immature (162) groups based on AVF maturation status. Univariate logistic regression analysis and the least absolute shrinkage and selection operator were used to identify independent risk factors, including D-dimer levels, low-density lipoprotein cholesterol levels, internal radial meridian, radial artery plaque presence, and cephalic vein indwelling needle use history. A predictive nomogram was developed specifically for immature AVF in elderly diabetic patients. Model performance was evaluated using the area under the receiver operating characteristic curve (AUROC), calibration curve, decision curve analysis (DCA), and clinical impact curve (CIC).

Results: Among elderly patients with diabetes mellitus, the incidence of immature AVF was 29.56%, affecting 162 of 548 individuals. The five-variable model demonstrated an AUROC value of 0.922, with a 95% confidence interval (CI) of 0.870 to 0.947 in the training dataset, and an AUROC of 0.912, accompanied by a 95% CI of 0.880 to 0.935 in the internal validation dataset. The calibration curve, derived from 1000 bootstrap samples, showed good agreement between predicted and observed outcomes. Additionally, both the DCA and CIC exhibited favorable clinical utility and net benefits.

Conclusions: The nomogram prediction model, based on independent risk factors, serves as a valuable tool for accurate prognosis and has potential to aid in establishing and preserving hemodialysis access in elderly diabetic patients, ultimately optimizing their healthcare outcomes.

Keywords: arteriovenous fistula; diabetes; hemodialysis; nomogram; vascular access.

Figure 1

AVF patients selection process.

Figure 2

Lasso regression coefficient path.

Figure 3

Lasso regression cross-validation plot.

Figure 4

The nomogram for predicting immature AVF based on the elderly diabetics (n = 548).

Figure 5

The individual nomogram (on the web) for predicting immature AVF.

Figure 6

(A): Diagnostic efficacy of predictive models in the training cohort; (B): Diagnostic efficacy of predictive models in the internal validation cohorts.

Figure 7

(A): The calibration curves for predicting immature AVF in the training cohort; (B): The calibration curves for predicting immature AVF in the internal validation cohorts.

Figure 8

(A): DCA analysis for the proposed nomogram model in the training cohort. The Y-axis shows the net benefit. The green line represents the assumption that all patients have the immature AVF and accept treatment. The black line represents the assumption that no patients have the immature AVF and not accept treatment. The red line represents the nomogram. The decision curve in indicated that if the threshold probability is between 0 and 1.0, then using the nomogram to predict the immature AVF adds more benefit than the treat-all patients scheme or treat-none scheme. (B). DCA analysis for the proposed nomogram model in the internal validation cohorts. The Y-axis shows the net benefit. The green line represents the assumption that all patients have the immature AVF and accept treatment. The black line represents the assumption that no patients have the immature AVF and not accept treatment. The blue line represents the nomogram. The decision curve in indicated that if the threshold probability is between 0 and 1.0, then using the nomogram to predict the immature AVF adds more benefit than the treat-all patients scheme or treat-none scheme.

Figure 9

(A): CIC of nomogram model in the training cohort. The red cure (number of high-risk individuals) indicates the number of people who are classified as positive (high risk) by the model at each threshold probability; the green curve (number of high-risk individuals with outcome) is the number of true positives at each threshold probability. CIC visually indicated that nomogram conferred high clinical net benefit and confirmed the clinical value of the nomogram model. (B): CIC of nomogram model in the internal validation cohorts. The blue cure (number of high-risk individuals) indicates the number of people who are classified as positive (high risk) by the model at each threshold probability; the green curve (number of high-risk individuals with outcome) is the number of true positives at each threshold probability. CIC visually indicated that nomogram conferred high clinical net benefit and confirmed the clinical value of the nomogram model.