Presentation (CTI)

OBJECTIVES: GLP-1 receptor agonists offer substantial long-term health and economic benefits, yet treatment effect heterogeneity remains poorly understood. Amidst high drug costs and rising demand, a critical policy challenge is targeting GLP-1 therapy to those most likely to benefit. While current microsimulation models estimate population-level gains, they offer limited insight into the specific baseline characteristics driving individual responsiveness.
METHODS: We analyzed GLP-1 treatment effect heterogeneity using the Future Adult Model (FAM), a U.S. policy microsimulation. Three age cohorts (25-30, 30-40, 40-50) with BMI≥30 and no weight-related comorbidities were simulated for 30 years under GLP-1 and status-quo scenarios. Individual-level treatment effects were computed as differences in cardiometabolic outcomes, QALYs, and total medical spending. For each cohort, gradient-boosted decision tree models mapped baseline characteristics to these effects. We then utilized SHAP (Shapley Additive Explanations) values to quantify how each predictor shifted individual benefits relative to the average, providing a decomposition of how attributes amplify/attenuate GLP-1 benefits.
RESULTS: Across cohorts, GLP-1 therapy produced significant 30-year gains: hypertension prevalence fell by 7-12 percentage points, annual QALYs rose by 0.06-0.07, and medical spending (excluding drug costs) declined by $3,800-$5,100. Earlier initiation favored hypertension reduction, while later initiation generated greater cost savings. SHAP decompositions revealed substantial heterogeneity around these averages. Baseline BMI was the dominant nonlinear modifier; individuals in the 30-45 BMI range saw the largest health gains and cost offsets, whereas extremely high BMI attenuated benefits. Responsiveness also varied by demographics: women, lower-income adults (ages 30-40), and younger uninsured individuals derived above-average benefits. Black and Hispanic adults exhibited both larger gains and greater effect dispersion than non-Hispanic whites.
CONCLUSIONS: GLP-1 benefits vary sharply across individuals, with the greatest health and economic value concentrated among disadvantaged populations. These findings highlight opportunities to target GLP-1 coverage more efficiently while simultaneously advancing health equity.