Kidney Week

Abstract: SA-PO0338

Optimal Timing for Dialysis Preparation in CKD

Session Information

• Dialysis: Epidemiology and Facility Management
  November 08, 2025 | Location: Exhibit Hall, Convention Center
  Abstract Time: 10:00 AM - 12:00 PM

Category: Dialysis

• 801 Dialysis: Hemodialysis and Frequent Dialysis

Author

• Choi, Yoonwon, Soonchunhyang University Hospital Bucheon, Bucheon-si, Gyeonggi-do, Korea (the Republic of)

Yoonwon Choi, MD

Background

Determining the optimal timing for dialysis preparation is a clinical challenge. As unplanned dialysis initiation increases morbidity and mortality, early planning is essential. This study aimed to develop a model to estimate the appropriate timing for dialysis preparation in patients with chronic kidney disease (CKD).

Methods

We retrospectively analyzed CKD patients followed in division of nephrology for ≥1 year who progressed to end-stage kidney disease (ESKD) and started hemodialysis (HD) between January 2011 and June 2024. Those with prior renal replacement therapy (RRT) or who discontinued HD within three months were excluded. A multiple linear regression (MLR) model was developed to predict eGFR six months prior to HD initiation (eGFR_6M), considered the optimal time for dialysis preparation.

Results

A total of 507 patients (299 males; mean age 61 ± 14 years) were included. The mean time from first nephrology visit to HD initiation was 69.0 ± 52.7 months. Significant predictors in the final MLR model included sex, impaired mobility, diabetes mellitus (DM), cardiovascular disease (CVD), dementia, ejection fraction (EF), blood urea nitrogen (BUN), and phosphorus (P). The equation was:
eGFR_6M = 13.029 − 1.28 × (female) + 1.66 × (impaired mobility) + 0.903 × (DM) + 1.15 × (CVD) + 1.951 × (dementia) − 0.068 × (EF) + 3.528 × (BUN ≦60) + 2.564 × (P ≦5.5)
The model demonstrated an adjusted R² of 0.325 and a Durbin-Watson statistic of 2.009.
A validation cohort of 50 patients under the same criteria showed a mean eGFR at 6 months before HD of 10.97 ± 3.60 mL/min/1.73 m². Using the model, 74% (n = 37) of patients had predicted eGFR within ±3.60 mL/min/1.73 m².

Conclusion

This model may assist nephrologists in identifying the optimal timing for dialysis preparation and vascular access planning, potentially improving outcomes in CKD care.

Table 1. Multivariate analysis of eGFR_6M

Digital Object Identifier (DOI)

• doi: 10.1681/ASN.2025apmdfygp