GUBMENTPlain talk · policy frontier
Filings / Mental health / Sources / §1 Findings: Baseline — prevalence, sp
GBMT-11 · Research record · No. 11

§1 Findings: Baseline — prevalence, spending, who pays

mental-health/research/ws01-baseline.md
This is a working research document from the mental health filing, published as written — including the parts later corrected. It is the underlying record for Whitepaper No. 11, not a summary of it.

Date: 2026-08-11. Extends Phase 0 anchors 1 (AMI/SMI lock) and the spending/payer half of the desk wager. Does not re-derive Phase 0 definitions.

PRISMA-lite search log

Field Logged
Strings NSDUH AMI SMI 2024; SAMHSA Behavioral Health Spending Use Accounts mental health Medicaid; NHEA behavioral OR mental health disease category; MEPS mental illness expenditure; MACPAC Medicaid behavioral health spending share; Commonwealth Fund Medicaid mental health spending 2019; NHIS any mental illness prevalence
Date range Prefer 2014–2025 vintages; older BHSUA retained where it is still the last official SAMHSA payer split
Inclusion Federal primary (SAMHSA NSDUH, SAMHSA BHSUA, CMS NHEA, AHRQ MEPS, MACPAC); peer/secondary only to locate primary roots. Exclude advocacy prevalence headlines without survey methods.
Searched 2026-08-11

1. Prevalence locked (carry Phase 0)

From SAMHSA 2024 NSDUH (Phase 0 / S1–S2):

Construct 2024 estimate Definition (locked)
Adult AMI 23.4% (61.5M) Past-year DSM-IV mental/behavioral/emotional disorder; excludes developmental + SUD
Adult SMI 5.6% (14.6M) Among AMI, substantial interference with ≥1 major life activities
Method Prediction models from 2008–2012 clinical interviews applied to 2021–2024 samples DSM-5 calibration in progress; published figures remain DSM-IV model

Band rule (protocol §1.3): SMI and AMI-minus-SMI are different markets. AMI/SMI under this instrument differ by ~4.2× (23.4 / 5.6). Circulating single-number “crisis” rates that treat them as interchangeable fail the definitional lock.

Adult AMI/SMI showed no statistically significant change overall from 2021–2024 (Phase 0). SMI among 18–25 fell 12.0% → 9.4%.

Treatment receipt (2024 NSDUH short report / Phase 0): among adults with AMI, 52.1% received mental-health treatment in the past year; among adults with SMI, 70.8%. 2024 treatment-receipt estimates are not comparable to 2022–2023 because of questionnaire changes (SAMHSA explicit break).

Alternate prevalence series for H1’s >1.5× cross-survey test: NHIS does not publish AMI/SMI. It publishes symptom indicators (regular feelings of worry/anxiety or depression) and cost-related unmet mental-health care — not a DSM-band crosswalk. Not located: a second authoritative adult AMI (or SMI) series that differs from NSDUH by >1.5× for a comparable year with a published definitional crosswalk.

2. Spending — what is findable

2a. SAMHSA Behavioral Health Spending and Use Accounts (last full official payer split)

2014 (SMA-16-4975 lineage, still the last complete SAMHSA MH payer table located this pass):

Implication: Medicaid is a plurality-scale public payer for MH, not a niche line item — but in 2014 it was not yet larger than private insurance’s MH share. The “Medicaid dominates serious illness” claim is about who covers people with SMI, not about Medicaid owning the majority of all MH dollars.

2b. Newer Medicaid MH dollar figures (secondary root; IHME/Commonwealth Fund)

Commonwealth Fund explainer (May 2025), citing IHME “Tracking Personal Health Care Spending” for 2019:

Treat as supporting, not headline: secondary compilation of IHME disease accounts, not a CMS NHEA table.

2c. NHEA / CMS

CMS NHEA highlights publish total Medicaid ($871.7B in 2023; 18% of NHE) and a catch-all “other health, residential, and personal care” that includes residential MH/SUD facilities — not a clean national MH disease line. Not located: a current CMS NHEA table that isolates mental-health treatment spending comparable to BHSUA’s $186B construct.

MACPAC states Medicaid pays ~24% of US mental health and SUD services (standing program fact; aligns directionally with BHSUA’s ~25% MH share).

2d. MEPS

AHRQ MEPS-HC remains the household expenditure survey; 2022 full-year consolidated file (HC-243) is available. BEA Health Care Satellite Account (MEPS Account basis) published mental-illness per-capita medical services at $639 (2021). Not located this pass: a clean, citable MEPS table of total US household-side MH expenditures with payer shares for a recent year without microdata re-tabulation. Record as gap, not invented total.

3. Who pays — by band (qualitative lock for scorecard)

Band Dominant financing (evidence) Confidence
Adult SMI / disability pathway Medicaid (+SSI/SSDI cash → GBMT-12); safety-net specialty Moderate — coverage share for AMI/SUD adults on Medicaid is material (Commonwealth Fund/NSDUH: Medicaid covers 26% of adults with AMI or SUD); inpatient MH+SUD half-Medicaid in IHME 2019
Common disorders / AMI-minus-SMI Private insurance + OOP; primary-care / therapy markets Moderate — BHSUA private 28% of all MH dollars; Bishop commercial acceptance higher than Medicaid (see §2)
Acute inpatient adult Mixed; Medicaid constrained by IMD default (see §3) Strong on the constraint; payer mix of realized stays still inventory-thin

4. H1 adjudication

H1 — Prevalence is not the scarce number. Circulating “crisis” prevalence figures are definition-sensitive and are not the binding constraint on access.

Criterion Result
≥2 authoritative series (different definitions) for adult AMI or SMI differ by >1.5× Partial. Within the federal instrument, locked AMI vs SMI differ by ~4.2× — definitional sensitivity is extreme and is exactly what protocol §1.3 forbids conflating. A second independent survey series for AMI (or SMI) at >1.5× vs NSDUH with a published crosswalk was not located (NHIS ≠ AMI).
or unmet-need measures move materially less than prevalence headlines imply across a decade Leaning support, survey-break limited. Adult AMI/SMI flat 2021–2024 while “crisis” rhetoric continued; ~half of AMI still untreated (52.1% treated). Pre-/post-2021 multimode and 2024 questionnaire breaks block a clean decade unmet-need series.
Refute path (single federal definition underpins public figures; unmet need tracks prevalence within 20%) Does not fire. Public figures routinely mix AMI, SMI, symptom screens, and youth MDE.

Adjudication: Supported — on definitional sensitivity (AMI≠SMI under the locked federal definitions) plus flat recent prevalence alongside persistent untreated share. The binding-constraint half of H1 is handed to §2/KC1 (no national realized-access instrument) rather than re-litigated here.

Deviation logged: literal two-survey AMI crosswalk at >1.5× not found; support rests on within-instrument band gap + flat recent prevalence.

Implications for scorecard

Confidence: Moderate–Strong

Strong on AMI/SMI lock and BHSUA 2014 payer shares (federal primary). Moderate on post-2014 Medicaid MH dollar levels (secondary disease accounts). Weak/absent on recent MEPS national MH totals and NHEA disease-line MH.

← All Mental health research documents Sources digest Read the whitepaper