The question a planner actually asks: "why is your number different from the published models, and which should I trust?" This puts our Medicaid-eligible estimate next to the two free state-level projections — Harvard CHOICES (low-income and general-population) and IHME/GBD — and explains, per state, exactly why they diverge. Phase 3 of the projection engine.
| Model / series | Population | 2026 | 2030 | Horizon | Base data |
|---|
Our estimate (re-based). We anchor a current Medicaid-eligible obesity rate and borrow CHOICES's validated low-income slope: projected(yr) = CHOICES_blend(yr) × (anchor ÷ CHOICES_blend(2026)). New Jersey's anchor is our independent v1.5 calibration (~40%); other states default to the CHOICES-implied 2026 level until you plug in a measured rate.
CHOICES low-income. Average of CHOICES's <$20k and $20k–$50k obesity bands — the Medicaid-eligible income range straddles the $20k line, so a blend beats the poorest band alone. CHOICES overall is its all-income state figure (general population).
IHME. Shown as a cited general-population benchmark only — its data license is non-commercial and bars derivative works, so its figures are referenced, never ingested. Published national adult obesity is forecast to reach ~57% by 2050 (men 55.3% / women 58.8%); Southern states (incl. Mississippi) stay highest, and Colorado is flagged for unusually rapid adult increase.
Honesty guardrail. Eligibility is computed (an FPL/MAGI rule), the burden is a calibrated extrapolation, and a true enrollee-level predictive model needs claims (the paid path). Public data is a proxy; the exact number needs the plan's own data.
Sources: CHOICES / Ward ZJ et al., NEJM 2019;381:2440-50 (per-state income tables, choicesproject.org). IHME / GBD 2021, Lancet 2024 (state + national obesity to 2050). KFF State Health Facts — total Medicaid/CHIP enrollment (~2024) and GLP-1 obesity coverage (Jan 2026). Demo set: 7 states; extensible to 51 via the same CHOICES scrape.