Trellison Institute · Working Paper · v1.0 draft · Companion to Mental Health Access Gap V1 (adults)

The Youth Mental Health Access Gap Is Structurally More Severe Than the Adult Gap and Wider Across States

Need-vs-Access Framework v1.1 (state geography), applied to U.S. youth data · May 2026
Trellison methodology rating — pending LedgerWell evidence chain — pending Dataset live · 35 states · 41.8M under-18

Of 41.8 million American minors in 35 participating states, 39.4% of high-school students report two weeks or more of persistent sadness or hopelessness in the past 12 months. That is 2.4× the adult companion paper's pop-weighted frequent-distress rate (16.8%). The state-to-state range in youth-serving mental-health provider density is approximately 85× — three times wider than the comparable adult range. The youth access conversation is a state policy choice.

Headline findings

MetricValue
Participating states35
Under-18 covered41.8M
Pop-weighted prevalence (sad/hopeless 2+ wks past 12mo)39.4%
Adult companion (frequent mental distress)16.8%
Youth-to-adult ratio~2.4×
Youth-serving provider density range (CT 500 vs PR 5.8 per 100K)~85×
Positive outliers (gap worse than uninsured predicts)PR (+2.39σ) · NC (+1.81) · NJ (+1.53)
Negative outliers (gap better than predicted)VT (-1.70σ) · AK (-1.69)

The viz layer

Three animated state chloropleths render the analysis end-to-end. Each is a single-take exploration of one dimension of the state-level dataset.

Need prevalence by state — 10s · 39.4% pop-weighted · 35 states · YRBSS 2023 open on YouTube ↗
Provider supply density — 8s · 85× state range · CT 500/100K to PR 5.8/100K open on YouTube ↗
Residual outliers — 12s · PR/NC/NJ positive · VT/AK negative · national OLS open on YouTube ↗

Two policy stories

What's working — Vermont, Alaska

Vermont: lowest youth-distress prevalence in the dataset (29.3%). University of Vermont Medical Center pediatric behavioral-health + state Medicaid expansion + Designated Agency system that mandates community-mental-health capacity. 1,059 youth-serving providers per 100,000 under-18.

Alaska: high distress (43.2%) but high access (1,085 per 100K). Indian Health Service + Alaska Native tribal health organizations + Title V MCH block grants for tribal communities. The federally-employed structural supply Alaska's negative-outlier status depends on.

Both are policy-driven, not market-driven.

What's not working — Puerto Rico, North Carolina, New Jersey

Puerto Rico: gap +2.39σ worse than uninsured rate predicts — the strongest signal in the dataset. Youth-serving provider density 5.8 per 100,000 under-18 — by far the lowest. Post-Hurricane Maria workforce migration + federal Medicaid payment cap policy.

North Carolina: gap +1.81σ. Strong adult workforce, thin youth-serving subset. CCBHC implementation is the most likely policy lever.

New Jersey: gap +1.53σ. Among the most insurance-saturated states but the youth-serving workforce has lagged behind the adult-serving capacity.

"The youth gap is a state-level supply problem, not an insurance problem alone."

Method, in one paragraph

State-level need from CDC YRBSS 2023 (high-school students grades 9-12 reporting persistent sadness/hopelessness 2+ weeks past 12 months, Total demographic). State-level provider supply from CMS NPPES (May 2026) filtered to 8 youth-serving mental-health taxonomies. Under-18 population from ACS 1-year 2023 (B09001). Uninsured rate under-19 from ACS 1-year 2023 (S2701). Gap ratio = (need × 1000) / state-supply per 100K. National OLS residual regression (state-level adaptation; within-state OLS at tract-level in adult companion). Z-score classification at ±1.5σ. Population-weighted dartboard sampling stratified by residual class. Framework atlas.need_vs_access_framework_v1 v1.1.0, reproducibility package below.

The complete content arsenal

Twenty-seven canonical parts. Same spec as the adult companion paper. Status as of this build:

foundation
Working paper✓ shipped
7 sections · ~3,800w
State-level adaptation + framework v1.1.0
Framework tool✓ shipped
atlas.need_vs_access_framework_v1 v1.1.0
Dataset CSV✓ shipped
35 states × 12 fields · sha256 stamped
Data dictionary✓ shipped
12 fields · type · unit · source
Authorship✓ shipped
Day-Zero standard
derivative text
Executive brief✓ shipped
2-page
Press release✓ shipped
~900w
Press Q&A✓ shipped
12 anticipated questions
9 case-study state profiles
σ ∈ {1.0, 1.5, 2.0} + taxonomy + threshold sweep
Bibliography✓ shipped
26 refs organized by provenance
End-to-end recipe
~1,400w
Slides outline✓ shipped
12-slide briefing deck
derivative visual
Static heatmaps⏳ pending
State chloropleth previews (forthcoming)
Animated viz clips⏳ pending
3 × mp4 state chloropleths (rendering)
Narration script✓ shipped
v1.0 · ~660w · 510s
Long-form video⏳ pending
~510s narrated data-story
Social clips⏳ pending
15-60s for X/LinkedIn
Interactive heatmap⏳ pending
Client-side web embed
distribution
Hub page✓ shipped
This page
YouTube uploads⏳ pending
After viz clips ready
DOI registration⏳ pending
Zenodo upload + DOI mint
Schema.org metadata✓ shipped
JSON-LD for academic indexability
audit gates
Trellison methodology rating⏳ review
Pending peer review
LedgerWell evidence-chain⏳ review
Pending cryptographic attestation

The reusable framework

The Need-vs-Access Framework v1.1.0 is the second-iteration version of the analytical pipeline that produced the adult companion paper. The state-geography adaptation required a single new parameter (regression_grouping) — switching from within-state OLS (default for tract/county) to single national OLS (default for state). All other code is unchanged.

This is the framework's second published application. Eleven other access domains are queued.

Status

StageStatus
Analysis complete✓ shipped
Working paper drafted✓ v1.0 draft
Framework v1.1.0✓ active
Dataset published to MinIO✓ live
14 text artifacts + 14 sub-pages✓ live
Animated visualizationsrendering
Trellison methodology ratingpending review
LedgerWell evidence-chain certificatepending
DOI registration for datasetpending
Public releaseafter Trellison + LedgerWell sign-off