DEVELOPING DIRECT EQUITY WEIGHTS FOR COST-EFFECTIVENESS ANALYSIS: RESULTS FROM A US-BASED DISCRETE CHOICE EXPERIMENT
Author(s)
Christopher Cadham, PhD, MPH1, Shehzad Ali, MBBS, MPH, MSC, PhD2, Rafael Meza, PhD3, Lisa Prosser, MS, PhD4;
1Fred Hutchinson Cancer Center, Hutchinson Institute for Cancer Outcomes Research, Seattle, WA, USA, 2Western University, Department of Epidemiology and Biostatistics, London, ON, Canada, 3BC Cancer Research Institute, Department of Population Health Sciences, Vancouver, BC, Canada, 4University of Michigan, Susan B. Meister Child Health Evaluation and Research Center, Department of Pediatrics, Ann Arbor, MI, USA
1Fred Hutchinson Cancer Center, Hutchinson Institute for Cancer Outcomes Research, Seattle, WA, USA, 2Western University, Department of Epidemiology and Biostatistics, London, ON, Canada, 3BC Cancer Research Institute, Department of Population Health Sciences, Vancouver, BC, Canada, 4University of Michigan, Susan B. Meister Child Health Evaluation and Research Center, Department of Pediatrics, Ann Arbor, MI, USA
OBJECTIVES: Equity-informative cost-effectiveness methods require quantitative weights that differentially value health outcomes based on equity-relevant dimensions. We elicited public preferences for allocating scarce healthcare resources to develop a set of direct equity weights for equity-informative cost-effectiveness analyses.
METHODS: We fielded an online discrete choice experiment with 12 choice tasks using an efficient partial profile design. Attributes included: baseline health, health improvement, wait time, and two equity dimensions (income level, and prior experience with racial/ethnic discrimination). For equity dimensions, interventions could target an advantaged group, a disadvantaged group, or not target any group (untargeted interventions). Respondents selected between hypothetical healthcare interventions that differed by these attributes. Conditional logit models estimated preferences, and marginal rates of substitution were used to generate intersectional direct equity weights. Latent class analysis was conducted to classify subgroups of respondents with similar preferences.
RESULTS: A national sample of US adults completed the survey (n=1203). While respondents valued interventions that targeted less advantaged groups (those with limited income or likely to experience discrimination) relative to advantaged groups (those with higher income or unlikely to experience discrimination), the most preferred were untargeted interventions. Relative to the high-income/unlikely to experience discrimination reference group, weights ranged from 1.19 (95% CI: 1.11-1.27) for the high-income/likely to experience discrimination group to 2.30 (95% CI: 2.03-2.57) for the limited-income/unlikely to experience discrimination group. Latent class analysis identified six groups of respondents with diverging preferences towards equity dimensions, including limited value placed on the dimensions.
CONCLUSIONS: Preferences for resource allocation vary by equity dimensions; these preferences can be used to generate weights for economic evaluation. However, more research is needed to determine whether weighting accurately reflects public preferences given the strongest preference for untargeted interventions and latent class results that demonstrate diverging preferences for equity dimensions.
METHODS: We fielded an online discrete choice experiment with 12 choice tasks using an efficient partial profile design. Attributes included: baseline health, health improvement, wait time, and two equity dimensions (income level, and prior experience with racial/ethnic discrimination). For equity dimensions, interventions could target an advantaged group, a disadvantaged group, or not target any group (untargeted interventions). Respondents selected between hypothetical healthcare interventions that differed by these attributes. Conditional logit models estimated preferences, and marginal rates of substitution were used to generate intersectional direct equity weights. Latent class analysis was conducted to classify subgroups of respondents with similar preferences.
RESULTS: A national sample of US adults completed the survey (n=1203). While respondents valued interventions that targeted less advantaged groups (those with limited income or likely to experience discrimination) relative to advantaged groups (those with higher income or unlikely to experience discrimination), the most preferred were untargeted interventions. Relative to the high-income/unlikely to experience discrimination reference group, weights ranged from 1.19 (95% CI: 1.11-1.27) for the high-income/likely to experience discrimination group to 2.30 (95% CI: 2.03-2.57) for the limited-income/unlikely to experience discrimination group. Latent class analysis identified six groups of respondents with diverging preferences towards equity dimensions, including limited value placed on the dimensions.
CONCLUSIONS: Preferences for resource allocation vary by equity dimensions; these preferences can be used to generate weights for economic evaluation. However, more research is needed to determine whether weighting accurately reflects public preferences given the strongest preference for untargeted interventions and latent class results that demonstrate diverging preferences for equity dimensions.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
EE198
Topic
Economic Evaluation
Topic Subcategory
Novel & Social Elements of Value
Disease
No Additional Disease & Conditions/Specialized Treatment Areas