INCORPORATING RISK PREFERENCES INTO VALUE ASSESSMENT FOR ALZHEIMER'S DISEASE: A DISCRETE CHOICE EXPERIMENT USING THE GRACE FRAMEWORK
Author(s)
Adam Atherly, PhD1, Eline Van Den Broek-Altenburg, BA, MA, MSc, PhD2.
1Professor, Virginia Commonwealth University, Richmond, VA, USA, 2University of Vermont, Larner College of Medicine, Burlington, VT, USA.
1Professor, Virginia Commonwealth University, Richmond, VA, USA, 2University of Vermont, Larner College of Medicine, Burlington, VT, USA.
OBJECTIVES: Standard cost-effectiveness analyses in Alzheimer’s Disease (AD) rely on QALYs and costs, often failing to capture how individuals value risk, uncertainty, and distributional outcomes. The Generalized Risk-Adjusted Cost-Effectiveness (GRACE) framework extends traditional CEA by incorporating risk preferences and non-linear utility, allowing valuation to reflect how individuals trade off uncertain benefits and harms. This study applies GRACE to estimate preference-based adjustments to value in AD treatments.
METHODS: We conducted a discrete choice experiment (DCE) informed by qualitative focus groups to identify non-QALY value attributes. Data were collected from 1,000 respondents via an online platform (Centiment) using quota sampling. Attributes and levels emphasized risk-based trade-offs consistent with Mulligan et al., including: probability of treatment benefit (20%, 50%, 80%), risk of serious side effects (1%, 5%, 10%), uncertainty in outcomes (vs probabilistic), delay to cognitive decline (6 months, 1 year, 5 years), caregiver burden (low, moderate, high), monthly out-of-pocket cost ($50, $100, $150), and equity framing (population average vs targeted benefit). We used NGene to create a D-efficient design generating the choice sets. Preferences were estimated using mixed logit and latent class models. GRACE was applied to transform estimated utilities into risk-adjusted value weights, capturing aversion to uncertainty and heterogeneity in preferences.
RESULTS: Respondents demonstrated strong sensitivity to risk and uncertainty, with willingness to accept higher treatment risk for larger or probabilistic benefits. GRACE-adjusted estimates revealed departures from linear QALY assumptions, with greater weight placed on treatments offering uncertain but meaningful gains. Substantial heterogeneity was observed across demographic groups, with disadvantaged populations exhibiting different risk-benefit trade-offs and higher valuation of equity-related outcomes.
CONCLUSIONS: Using GRACE to incorporate risk preferences into value assessment is an important approach improving upon standard CEA by capturing how individuals value uncertainty and trade-offs. This leads to more accurate and policy-relevant evaluations of AD/ADRD treatments.
METHODS: We conducted a discrete choice experiment (DCE) informed by qualitative focus groups to identify non-QALY value attributes. Data were collected from 1,000 respondents via an online platform (Centiment) using quota sampling. Attributes and levels emphasized risk-based trade-offs consistent with Mulligan et al., including: probability of treatment benefit (20%, 50%, 80%), risk of serious side effects (1%, 5%, 10%), uncertainty in outcomes (vs probabilistic), delay to cognitive decline (6 months, 1 year, 5 years), caregiver burden (low, moderate, high), monthly out-of-pocket cost ($50, $100, $150), and equity framing (population average vs targeted benefit). We used NGene to create a D-efficient design generating the choice sets. Preferences were estimated using mixed logit and latent class models. GRACE was applied to transform estimated utilities into risk-adjusted value weights, capturing aversion to uncertainty and heterogeneity in preferences.
RESULTS: Respondents demonstrated strong sensitivity to risk and uncertainty, with willingness to accept higher treatment risk for larger or probabilistic benefits. GRACE-adjusted estimates revealed departures from linear QALY assumptions, with greater weight placed on treatments offering uncertain but meaningful gains. Substantial heterogeneity was observed across demographic groups, with disadvantaged populations exhibiting different risk-benefit trade-offs and higher valuation of equity-related outcomes.
CONCLUSIONS: Using GRACE to incorporate risk preferences into value assessment is an important approach improving upon standard CEA by capturing how individuals value uncertainty and trade-offs. This leads to more accurate and policy-relevant evaluations of AD/ADRD treatments.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
EE661
Topic
Economic Evaluation
Topic Subcategory
Novel & Social Elements of Value
Disease
Alternative Medicine, Neurological Disorders