Abstract: FR-PO0036
Eliciting Physician Preferences for Next-Generation Kidney Replacement Therapies: Development of a Discrete Choice Experiment
Session Information
- Artificial Intelligence and Digital Health at the Bedside
November 07, 2025 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Artificial Intelligence, Digital Health, and Data Science
- 300 Artificial Intelligence, Digital Health, and Data Science
Authors
- Quan, Amanda I., University of California San Francisco School of Medicine, San Francisco, California, United States
- Wilson, Leslie, University of California San Francisco School of Pharmacy, San Francisco, California, United States
- Frassetto, Lynda, University of California San Francisco School of Medicine, San Francisco, California, United States
- Fissell, William Henry, Vanderbilt University Medical Center, Nashville, Tennessee, United States
- Roy, Shuvo, University of California San Fransciso Dept of Bioengineering, San Francisco, California, United States
Group or Team Name
- The Kidney Project.
Background
Next generation kidney replacement therapies (KRTs) have potential to transform the management of end-stage renal disease (ESRD) by enabling continuous dialysis that approximates native kidney function. However, the clinical adoption and viability of these technologies require further understanding of therapeutic preferences among stakeholders. Building upon our prior research on patient preferences, this study describes the development of a choice-based conjoint (CBC) survey instrument to assess physician risk-benefit tradeoffs and preference utilities influencing treatment selection.
Methods
Our survey was developed with discrete choice methodology aligned with the International Society for Pharmacoeconomics and Outcomes Research. An initial list of therapeutic attributes found across KRTs was first compiled through a systematic literature review on emerging ESRD treatments and direct observation of decision-making in renal care. This list was then refined for relevance to physician treatment selection based on semi-structured interviews with 7 practicing nephrologists and transplant surgeons. Attribute definitions were revised to ensure precision and mutual exclusivity, and a range of descriptive outcome levels was established for each through iterative consultation with a multidisciplinary expert panel comprising a nephrologist, bioengineer, and health economist. Finally, full-profile treatment options were generated by combining variable attribute levels using a random, balanced overlap design and assembled into paired choice tasks using Sawtooth Software.
Results
The final CBC includes 10 device attributes, each with 2 to 4 levels reflecting risks and benefits associated with renal organ transplant, portable and implantable dialyzers, and implantable artificial kidneys. 16 full profile paired choice tasks were constructed, 10 random and 6 fixed, and introduced by a clinical vignette simulating real-world decision-making. The CBC was delivered as part of a larger instrument containing background on KRTs, instructional training, and demographic questionnaires assessing clinical experience and relevant patient (contra)indications.
Conclusion
The development of this CBC instrument marks a critical step toward understanding physician preferences in treatment selection to inform the design of next generation KRTs.
Funding
- Other NIH Support