Projects obesity prevalence in a state's Medicaid-eligible population forward, by anchoring to a current calibrated rate and borrowing the validated growth trajectory from Harvard CHOICES. A calibrated projection, not a prediction. Switch the lens for a pharma (market-opportunity) or payer (budget) read.
| Metric | 2026 | 2030 | Change |
|---|
The math (anchor + borrow slope). We take a current Medicaid-eligible obesity rate (the anchor) and scale the CHOICES low-income trajectory so it passes through that anchor at 2026: projected(year) = CHOICES_blend(year) × (anchor ÷ CHOICES_blend(2026)). This keeps the freshest level while borrowing Harvard's peer-reviewed slope — and avoids ingesting any licensed dataset.
CHOICES blend. The trajectory is the average of CHOICES's <$20k and $20k–$50k income bands (obesity, BMI ≥30) — the Medicaid-eligible income range straddles the $20k line, so a blend beats the poorest band alone.
Anchor default. New Jersey defaults to our independent v1.5 calibration (~40%). Other states default to the CHOICES-implied 2026 level (a starting estimate). Replace the anchor with your plan's measured rate for a real read.
Benchmarks. CHOICES (Ward 2019, NEJM, to 2030) is the borrowed trajectory and is shown on the chart. IHME / GBD (general-population, to 2050) is a cited external benchmark only — not ingested, because its data license bars commercial/derivative use.
Caveats. Public data is a proxy for actual enrollees; eligibility is computed, the burden is a calibrated estimate (not an individual predictive model — that needs claims). Enrollment "spikes" (recession, unwinding, expansion) are scenarios, not predictions. GLP-1, uptake, and net-cost figures are illustrative until run on a plan's real data.
Sources: Harvard CHOICES per-state income series (NEJM 2019); FPL/eligibility per the companion engine (HHS 2026, MACPAC 2025). Demo set: 7 states; extensible to all 51 via the same scrape.