Kidney Week

Abstract: SA-PO1079

Development and Validation of a Prediction Model of Early Mortality for Conservative Management vs. Dialysis among US Veterans with Advanced CKD

Session Information

• CKD: Epidemiology, Risk Factors, and Prevention - 3
  October 26, 2024 | Location: Exhibit Hall, Convention Center
  Abstract Time: 10:00 AM - 12:00 PM

Category: CKD (Non-Dialysis)

• 2301 CKD (Non-Dialysis): Epidemiology, Risk Factors, and Prevention

Authors

• Rhee, Connie, University of California Los Angeles, Los Angeles, California, United States

Connie Rhee, MD, MS

• Narasaki, Yoko, University of California Los Angeles, Los Angeles, California, United States

Yoko Narasaki, MS, PhD

• You, Amy Seungsook, University of California Los Angeles, Los Angeles, California, United States

Amy Seungsook You, MS

• Kovesdy, Csaba P., The University of Tennessee Health Science Center, Memphis, Tennessee, United States

Csaba P. Kovesdy, MD, FASN

• Novoa Vargas, Alejandra, University of California Los Angeles, Los Angeles, California, United States

Alejandra Novoa Vargas, MPH

• Nguyen, Danh V., University of California Irvine, Irvine, California, United States

Danh V. Nguyen, PhD

• Kalantar-Zadeh, Kamyar, Harbor-UCLA Medical Center, Torrance, California, United States

Kamyar Kalantar-Zadeh, MD, DrMed, PhD, MPH, FASN

Background

Given that dialysis may result in greater healthcare utilization, loss of physical function/independence, and high early-mortality in certain subgroups (elderly, multi-morbid), there is growing interest in conservative management (CM) as an alternative patient-centered treatment strategy for advanced CKD. We developed and validated a prediction model to provide individualized predicted probability of survival with CM vs. dialysis transition among a national cohort of advanced stage CKD patients.

Methods

Using the national VA database linked to USRDS and Medicare data, we developed a risk prediction tool for mortality in US Veterans with advanced CKD (≥2 eGFRs <25 separated by ≥90 days) treated with CM vs. dialysis (defined as non-receipt vs. receipt of dialysis within 2-years of 1^st eGFR <25) over 2010-19. Patients were divided into a 2/3rds development set and a 1/3rd validation set. Prediction models for 1-year mortality were developed on the basis of survival data up to 2-years after the index eGFR date (1^st eGFR <25) using Cox models.

Results

Among a cohort of 61,118 Veterans (43,197 vs. 17,921 receiving CM vs. dialysis), selected characteristics associated with higher overall mortality included older age; higher index eGFR, UACR, and VA frailty index; faster eGFR decline; lower serum albumin and BMI; prior 1-year hospitalization; presence of heart disease, sepsis, and tobacco use; and dialysis transition. Model discrimination was good with similar C-statistics in the development and validation cohorts: 0.704 (95% CI 0.697-0.708) and 0.704 (95% CI 0.696-0.711), respectively. Model goodness-of-fit tests showed adequate fit in the validation cohort (P=0.65). Risk score and model calibration plots are shown in Fig 1.

Conclusion

In a national cohort of US Veterans with advanced CKD, we developed a clinical prediction tool for survival for CM vs. dialysis to inform individualized treatment and improve the shared decision-making process between clinicians and patients.

Funding

• NIDDK Support