Abstract: PO1253
Challenging Assumptions of Outcomes and Costs Comparing Peritoneal and Hemodialysis
Session Information
- Peritoneal Dialysis - 1
October 22, 2020 | Location: On-Demand
Abstract Time: 10:00 AM - 12:00 PM
Category: Dialysis
- 703 Dialysis: Peritoneal Dialysis
Authors
- Lin, Eugene, University of Southern California Keck School of Medicine, Los Angeles, California, United States
- Lung, Khristina I., University of Southern California, Los Angeles, California, United States
- Chertow, Glenn Matthew, Stanford University School of Medicine, Stanford, California, United States
- Bhattacharya, Jay, Stanford University School of Medicine, Stanford, California, United States
- Lakdawalla, Darius, University of Southern California, Los Angeles, California, United States
Background
Policy makers have suggested that increasing peritoneal dialysis (PD) use would improve end-stage kidney disease (ESKD) outcomes and reduce Medicare spending.
Methods
Using Medicare claims, we exploited an idiosyncratic Medicare coverage rule to conduct an instrumental variable analysis comparing mortality, hospitalizations, and Medicare spending between PD and hemodialysis (HD) in uninsured adults with incident ESKD. Uninsured patients usually receive Medicare at dialysis month four; however, those starting with PD receive Medicare at dialysis start and retroactive pre-dialysis coverage for the entire calendar month of dialysis start. Because pre-dialysis coverage is essential for PD catheter placements, the rule encourages more PD use among patients starting at the end of the month by increasing pre-dialysis coverage. We used dialysis start day as an instrumental variable to mitigate selection bias when comparing outcomes and costs of the two modalities.
Results
Starting dialysis later in the calendar month was associated with an increased probability of using PD at day 1 (absolute increase of 1.0% for every 10 days later in the month, 95% CI: 0.8%, 1.3%) and at month 12 (absolute increase of 0.7% for every 10 days later in the month, 95% CI: 0.4%, 1.0%). We observed no significant absolute difference between PD and HD for all outcomes: 12-month mortality, –0.4% (–3.4%, 1.8%), hospitalizations during months 7-12, 0.01 (–0.16, 0.17) per patient, and Medicare spending during months 7-12, $2,803 (95% CI: –$6,355, $508) per patient. We assessed the potential role of selection bias in prior studies by repeating the same analyses using traditional regression methods. In contrast to the instrumental variable model, when using traditional regression methods, PD was associated with statistically significant decreases in mortality but significant increases in costs.
Conclusion
Using an instrumental variable analysis, PD did not result in improved outcomes or lower costs when compared to HD. We observed evidence of selection bias when using traditional study methods. Policy makers eager to promote home dialysis should temper expectations of improved outcomes and reduced spending.
Funding
- NIDDK Support