CHOOSING FOR THE WORSE-OFF: A STRUCTURAL SOCIAL WELFARE MODEL OF EQUITY-EFFICIENCY PREFERENCES IN TYPE 2 DIABETES TREATMENT CHOICES
Author(s)
Mesfin Genie, PhD1, Surachat Ngorsuraches, PhD2;
1The University of Newcastle, Senior Lecturer in Health Economics, Newcastle, Australia, 2Auburn University, Auburn, AL, USA
1The University of Newcastle, Senior Lecturer in Health Economics, Newcastle, Australia, 2Auburn University, Auburn, AL, USA
OBJECTIVES: To determine patient preferences for treatment outcomes and cost, both for themselves and others with poorer health, when choosing health plans that cover type 2 diabetes (T2D) treatments.
METHODS: An online discrete choice experiment recruited U.S. adults with T2D through a QualtricsXM panel (N=701). Patients completed a D-efficient design choosing between two unlabeled health plans, including an opt-out alternative. Each health plan included choice tasks with unequal health outcomes: different efficacy (the chance of reaching a target A1C in 6 months), side-effect risk, and out-of-pocket costs for the patient and other patients with poorer health. We estimated a structural multinomial logit model that separately identified preferences for self and others across efficacy, risk, and cost, and summarized distributional concern using an equity parameter defined as the proportional weight on others’ efficacy relative to own efficacy minus one. Preference heterogeneity was assessed using a latent class structural model.
RESULTS: Respondents preferred higher own efficacy (1.41, p<0.01) and valued others’ (0.47, p<0.01), while higher side-effect risk and higher cost reduced utility for both self and others (p<0.05), with stronger aversion to own risk and cost. The implied equity parameter was negative and statistically significant (-0.67, p<0.01), implying others’ A1C gains carried about one third of the marginal value of equivalent own gains. The latent class model identified a “treatment-seeking” group (72.2%) with a strong aversion to opting out and significant concern for others’ outcomes, and a “treatment-hesitant” group (27.8%) with a strong propensity to opt out and much greater sensitivity to own costs and side-effect risks; both classes nevertheless exhibited self-favoring equity parameters (-0.62 and -0.77, respectively).
CONCLUSIONS: Patients exhibited other-regarding preferences in an unequal-outcomes context, but average distributional weights were self-favoring rather than strongly prioritarian. Equity-efficiency trade-offs are present and heterogeneous, suggesting that policy designs should acknowledge both self-interest and concern for worse-off others.
METHODS: An online discrete choice experiment recruited U.S. adults with T2D through a QualtricsXM panel (N=701). Patients completed a D-efficient design choosing between two unlabeled health plans, including an opt-out alternative. Each health plan included choice tasks with unequal health outcomes: different efficacy (the chance of reaching a target A1C in 6 months), side-effect risk, and out-of-pocket costs for the patient and other patients with poorer health. We estimated a structural multinomial logit model that separately identified preferences for self and others across efficacy, risk, and cost, and summarized distributional concern using an equity parameter defined as the proportional weight on others’ efficacy relative to own efficacy minus one. Preference heterogeneity was assessed using a latent class structural model.
RESULTS: Respondents preferred higher own efficacy (1.41, p<0.01) and valued others’ (0.47, p<0.01), while higher side-effect risk and higher cost reduced utility for both self and others (p<0.05), with stronger aversion to own risk and cost. The implied equity parameter was negative and statistically significant (-0.67, p<0.01), implying others’ A1C gains carried about one third of the marginal value of equivalent own gains. The latent class model identified a “treatment-seeking” group (72.2%) with a strong aversion to opting out and significant concern for others’ outcomes, and a “treatment-hesitant” group (27.8%) with a strong propensity to opt out and much greater sensitivity to own costs and side-effect risks; both classes nevertheless exhibited self-favoring equity parameters (-0.62 and -0.77, respectively).
CONCLUSIONS: Patients exhibited other-regarding preferences in an unequal-outcomes context, but average distributional weights were self-favoring rather than strongly prioritarian. Equity-efficiency trade-offs are present and heterogeneous, suggesting that policy designs should acknowledge both self-interest and concern for worse-off others.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
PCR4
Topic
Patient-Centered Research
Topic Subcategory
Patient Behavior and Incentives
Disease
SDC: Diabetes/Endocrine/Metabolic Disorders (including obesity)