Prediction model for radial-cephalic arteriovenous fistula failure to mature

Abstract

Background: Autologous radio-cephalic arteriovenous fistulas (RC-AVFs) represent the first option for hemodialysis in China. However, they exhibit a high rate of failure to mature. Methods: A total of 196 first-time RC-AVFs were included. We analyzed preoperative clinical and ultrasonic characteristics and bioelectrical impedance to screen risk factors using univariate and multivariate logistic regression. Subsequently, we constructed a nomogram and employed bootstrap resampling for internal validation. Additionally, we developed a risk score equation using a simplified Framingham heart study point system. Finally, we used a restricted cubic spline diagram to determine the clinical significance of the model variables. Results: Seventy-six (38.8%) RC-AVFs failed to mature within 6 months. We identified arterial diameter (AD), total cholesterol (CHO) levels, lean tissue index (LTI), and a history of coronary artery disease (CAD) (p < 0.005) as independent impact factors through univariate and multivariate logistic regression. The area under the receiver operating characteristic curve was 0.79 (95% confidence interval [CI]: 0.72–0.85), and the bootstrap-corrected concordance index was 0.73 (95 % CI: 0.713–0.763). Based on the risk scoring system (0–22 points), patients were categorized into low (0–10), medium (11–14), and high-risk (15–22) groups. Finally, a restricted cubic spline diagram illustrated a significant increase in adverse event risk with an AD ≤ 2 mm, CHO levels ≥ 3.8 mmol/L, and LTI ≤ 14 kg/m2. Conclusion: The risk prediction model incorporating LTI, CHO levels, AD, and a history of CAD showed good predictive performance for RC-AVF outcomes in patients with chronic kidney disease.