# Press Q&A — Youth Mental Health Access Gap V1
**Companion to**: press release · working paper
**Audience**: journalists, policy researchers, advocates
**Format**: 12 anticipated questions with sourced answers
---
### Q1. The 39.4% headline is much higher than the adult number. Are these directly comparable?
The two measurements use different surveillance instruments and different reference periods. The adult number (16.80%) comes from BRFSS via CDC PLACES and measures "frequent mental distress" defined as ≥14 days of poor mental health in the past 30 days. The youth number (39.4%) comes from YRBSS and measures "persistent sadness or hopelessness for 2 weeks or more in a row" in the past 12 months. The youth measurement is a longer reference window (12 months vs 30 days), which inflates the number relative to a strict 30-day comparison.
But the *substantive* intent is the same: measuring sustained distress that disrupts function. Both are validated population-level measures. The 2.4× ratio is a useful headline; it is not a literal "youth experience 2.4 times more mental health problems than adults" claim. It is "the share of high-school students reporting two weeks of sustained distress is 2.4× the share of adults reporting frequent mental distress." That is what the data says.
### Q2. Why only 35 states? What about California, New York, Florida?
YRBSS state-level data is released only for states that participate in the survey AND have releasable response rates. In 2023, 39 states had releasable state-level data; after our population threshold of 50,000 under-18, 35 states are in the published analysis. California, New York, and Florida are *not* in the participating set for 2023 state-level YRBSS — they conduct their own state-level youth surveillance instruments instead, which are not directly comparable.
This is a real limitation. It is also the best the public federal surveillance infrastructure can support today.
### Q3. Why state geography and not tract?
YRBSS does not publish at sub-state resolution. The CDC's authorization is state and national only. No public federal youth mental-health surveillance instrument publishes at tract or county granularity. The adult Mental Health Access Gap V1 paper used CDC PLACES tract-level data, which exists for adults but not for under-18.
This is the central methodological constraint on youth access analysis: the geographic resolution available for the need metric forces the analysis to state level.
### Q4. Why is Puerto Rico the worst positive outlier?
Puerto Rico has the lowest youth-serving provider density in the analysis (5.8 per 100,000 under-18) despite low uninsured rates (Puerto Rico's ASES Medicaid-equivalent program covers approximately 98% of residents under 19). The structural driver is well-documented: the post-Hurricane Maria health-system disruption in 2017 caused substantial mental-health workforce migration to the U.S. mainland, compounding decades of federal Medicaid-payment cap policy that has constrained Puerto Rico's mental-health workforce build-out.
The residual flags this as +2.39σ — the strongest signal in the dataset. The data does not assign causation but the magnitude is consistent with the documented Maria-era workforce migration.
### Q5. Vermont has the lowest youth distress in the data. Is that because Vermont is unusual, or because Vermont's mental health system is working?
Both factors contribute. Vermont has a small, demographically homogeneous, high-insurance population, and that explains some of the lower baseline. But the framework's residual analysis controls for the most relevant SES proxy (uninsured rate), and Vermont still shows up as a negative outlier — meaning the gap is *smaller than the model predicts even after controlling for insurance*. So Vermont's structural advantage — the UVM Medical Center pediatric behavioral-health program, the state Designated Agency community-mental-health system, and the 2014 Medicaid expansion — accounts for the remaining gap closure beyond what demographic factors alone would predict.
### Q6. Alaska is striking — high distress AND high access?
Yes. Alaska has elevated youth distress (43.2%, among the higher in the dataset) and the *second-highest* youth-serving provider density (1,085 per 100,000 under-18). The driver is structural: the Indian Health Service plus Alaska Native tribal health organizations employ a federally-funded mental-health workforce that the state-level supply count picks up.
This is a structurally different pattern from Vermont's. Vermont's high supply is state-university-anchored + Medicaid-driven; Alaska's is federally-employed. Both close the gap below what insurance coverage alone would predict.
### Q7. What about North Carolina and New Jersey? These are large, relatively wealthy states. Why are they positive outliers?
This is the question the residual analysis raises that the framework cannot fully answer. Both NC and NJ have strong adult mental-health workforces (NPPES counts are healthy). Both have low uninsured rates. Both should sit on or below the regression line. They do not — the youth-serving subset of their workforce is thin relative to what the adult-serving capacity would predict.
The most likely structural driver in both cases is that the youth-serving mental-health workforce has historically been a separate workforce line from the adult-serving one (different residencies, different practice patterns, different reimbursement). In states where adult-serving workforce has been the policy priority, the youth-serving lag is real even where adult capacity is strong. Both NC and NJ would benefit from explicit youth-serving CCBHC certification and BHWET workforce expansion.
### Q8. NPPES counts a provider regardless of whether they actually see under-18 patients. Doesn't that overstate the youth-serving supply?
Yes. The published analysis uses the *broad* taxonomy set (8 taxonomies including general clinical social workers and MFTs, totaling 79,868 providers). The *narrow* set (4 child-specific taxonomies, totaling 4,507 providers nationwide) is preserved for sensitivity testing. The broad set reflects the operational reality that a child accessing mental-health care in many states will see a clinical social worker or MFT rather than a board-certified child-and-adolescent psychiatrist.
The narrow set is more conservative; the broad set is more representative. Sensitivity analysis at both shows the residual pattern is preserved — PR, NC, NJ remain positive outliers; VT, AK remain negative outliers. The relative ranking is robust.
### Q9. How does the framework's residual analysis differ at state geography vs the adult tract-level analysis?
At tract geography, the framework runs within-state OLS for each of the 50 states (with hundreds of tracts per state, the regression is well-conditioned). At state geography, there's only one record per state, so within-state grouping is undefined. The framework adapts by running a single national OLS over all 35 participating states. The z-score classification at ±1.5σ is then applied to a single national residual distribution rather than 50 state-specific distributions.
This is a single-parameter change (`regression_grouping="national"`) in the framework's v1.1.0 release. No other code changes. The methodology supplement documents this.
### Q10. The dartboard returned only 9 hits instead of 12. Why?
The framework's dartboard samples 4 tracts per residual class. At state geography, there are only 2 negative-outlier states (VT, AK) and 3 positive-outlier states (PR, NC, NJ) in the published analysis. The framework handles small classes gracefully by sampling all available states in under-sampled classes rather than producing duplicate or synthetic noise. So negative_outlier yields 2 hits, positive_outlier yields 3 (with one duplicate due to the population-weighted sampling), and expected yields 4 — for a total of 9 hits.
This is a limitation of state-level geography (small total N) and is documented in the methodology supplement §3.
### Q11. Can the framework run on YRBSS questions other than "sad or hopeless 2+ weeks"?
Yes. YRBSS includes several mental-health questions that the framework can run on:
- Past-30-day mental health not good (the closer to PLACES adult measure)
- Considered suicide in past 12 months
- Made suicide plan in past 12 months
- Actually attempted suicide in past 12 months
Each is a different `need_metric` binding. The framework code does not change; only the data binding. The methodology supplement §1 lists all currently available bindings.
### Q12. Where is the dataset and how do I cite the work?
Dataset (35 states × 12 fields, CSV + sha256 manifest):
**https://api.daedarch.ai/api/v4/t/media/serve?bucket=daedarch-public-media&object=youth_mental_health_gap/dataset_v1/mh_gap_youth_state_v1.csv**
Working paper and content arsenal hub:
**https://trellison.com/research/youth-mental-health-supply-demand-gap**
Recommended citation (DOI pending Zenodo registration):
> Trellison Institute. *The Youth Mental Health Access Gap is Structurally More Severe Than the Adult Gap and Wider Across States*. Working paper, v1.0 (May 2026). DOI: [pending]. https://trellison.com/research/youth-mental-health-supply-demand-gap
The Need-vs-Access Framework code is documented in the methodology supplement and registered as `atlas.need_vs_access_framework_v1` v1.1.0 in the DaedArch tool registry.