Abstract: FR-PO902
Novel Prediction Score for Early Death upon Transition to Dialysis: A Veterans Affairs (VA) and Kaiser Permanente Southern California (KPSC) Big Data Approach
Session Information
- Dialysis: Epidemiology, Outcomes, Clinical Trials - Non-Cardiovascular - II
November 03, 2017 | Location: Hall H, Morial Convention Center
Abstract Time: 10:00 AM - 10:00 AM
Category: Dialysis
- 607 Dialysis: Epidemiology, Outcomes, Clinical Trials - Non-Cardiovascular
Authors
- Obi, Yoshitsugu, UC Irvine, Orange, California, United States
- Nguyen, Danh V., UC Irvine, Orange, California, United States
- Zhou, Hui, Kaiser Permanente Southern California, Pasadena, California, United States
- Streja, Elani, UC Irvine, Orange, California, United States
- Soohoo, Melissa, UC Irvine, Orange, California, United States
- Zhang, Lishi, UC Irvine, Orange, California, United States
- Chen, Yanjun, UC Irvine, Orange, California, United States
- Molnar, Miklos Zsolt, University of Tennessee Health Science Center, Memphis, Tennessee, United States
- Sim, John J., Kaiser Permanente Southern California, Pasadena, California, United States
- Jacobsen, Steven J., Kaiser Permanente Southern California, Pasadena, California, United States
- Rhee, Connie, UC Irvine, Orange, California, United States
- Kovesdy, Csaba P., University of Tennessee Health Science Center, Memphis, Tennessee, United States
- Kalantar-Zadeh, Kamyar, UC Irvine, Orange, California, United States
Background
Mortality is exceptionally high during the first year among incident dialysis patients. We aimed to establish a risk prediction tool for early mortality based on pre-ESRD conditions.
Methods
Among 35,853 US veterans who transitioned to dialysis over 6.5 years (10/2007–03/2014), we developed new risk scores based on demographics, primary ESRD causes, comorbid conditions, and pre-ESRD laboratory data. We stratified patients by low and high eGFR (i.e., <15 and ≥15 mL/min/1.73m2, respectively), and then applied a Cox model with AIC-based backward variable selection to 14-month survival data. The final model was validated among 4,284 KPSC patients over 8.75 years (01/2007–09/2015).
Results
Observed 1-year mortality was 27% in the VA cohort, and was not significantly different from predicted mortality. C-index values in the VA cohort were 0.71 and 0.67 among patients with low vs. high eGFR, respectively. The external validation using the KPSC cohort showed C-index values of 0.77 and 0.74 among men vs. women with low eGFR, respectively, and 0.71 and 0.67 among men vs. women with high eGFR, respectively.
Conclusion
New risk scores for early mortality have been developed and externally validated among a cohort of racially, ethnically, and gender diverse ESRD patients transitioning to dialysis. It would help with the identification of a high risk population and provide information that may contribute to dialysis initiation planning.
Funding
- NIDDK Support