Abstract: SA-PO0461
Predictors of Central Venous Catheter Use at Dialysis Initiation in Patients with CKD
Session Information
- Dialysis: Vascular Access
November 08, 2025 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Dialysis
- 803 Dialysis: Vascular Access
Authors
- Epparla, Anurag, The University of Tennessee Health Science Center, Memphis, Tennessee, United States
- Donnelly, Lauren, The University of Tennessee Health Science Center, Memphis, Tennessee, United States
- Wall, Barry M., The University of Tennessee Health Science Center, Memphis, Tennessee, United States
- Kovesdy, Csaba P., The University of Tennessee Health Science Center, Memphis, Tennessee, United States
Background
Initiating hemodialysis with a central venous catheter (CVC) is associated with higher morbidity and mortality compared to arteriovenous fistulas (AVFs) or grafts (AVGs). Early identification of patients who are at risk for catheter use may improve outcomes. We aimed to identify significant clinical predictors of CVC use and develop a prediction score for early risk stratification in a real-world cohort of patients with CKD.
Methods
We examined a single center historic cohort of 926 patients with CKD treated in a nephrology clinic at a tertiary center. We used multivariable logistic regression to develop a prediction model of future CVC use at dialysis transition. We evaluated discrimination using Harrel’s C statistics and calibration by comparing predicted and observed values by deciles. A simplified prediction score for clinical application was constructed using variable importance analysis.
Results
The cohort had a mean (SD) age of 66.3 (11.5) years, 96% were males, 60% were African-American, 55% were diabetic, and the baseline eGFR was 40 (22) ml/min/1.73m2. 110 patients (12%) started dialysis with a CVC after a median of 2.4 years of follow-up. Our complete prediction model included 31 clinical variables, with a C statistic of 0.79 (95%CI: 0.74-0.83) and excellent calibration (p=0.7) (Figure). Our simplified model used the following variables: age, eGFR, proteinuria and hemoglobin level. The C statistic for the risk score developed from the simplified model was 0.75 (0.70-0.79) with excellent calibration (p=0.7).
Conclusion
In this cohort of patients with moderate-to-advanced CKD, clinical characteristics accurately and precisely predicted the risk of future initiation of hemodialysis using a CVC. Interventions targeting modifiable predictors could help achieve optimal vascular access. Our model will need to be validated in an independent cohort.
Receiver-operating curves and calibration plots for the complete prediction model (A) and the simplified prediction model (B)