A state-level analysis of mental health and substance use disorder treatment gaps using SAMHSA NSDUH 2021–2022 survey estimates and HRSA mental health shortage area data. Built with Python 3, SQLite, NumPy, and Matplotlib — no scikit-learn, no pandas.
Author: Waleed Adawi | Year: 2026
| Metric | Value |
|---|---|
| WA adult AMI prevalence | 27.14% — 4th highest nationally |
| WA mental health treatment gap | +3.26 pp (rank #17 of 51) |
| National average MH gap | +2.20 pp |
| WA unmet AMI need | 51.2% — rank #31 nationally |
| WA counties with HPSA score ≥ 16 | Seven (federal high-priority threshold) |
| National SUD prevalence average | 18.64% |
Washington's AMI prevalence (27.14%) is among the highest in the country, yet the state's treatment gap (+3.26 percentage points) falls close to the national average (+2.20 pp), suggesting moderate treatment system capacity relative to need — though still leaving roughly one-quarter of adults with any mental illness without care.
Nevada leads the country with the largest MH treatment gap (+8.66 pp), while Massachusetts, Connecticut, and New Jersey have negative gaps — meaning estimated treatment rates exceed prevalence estimates, likely reflecting measurement uncertainty at the tails of the distribution. Washington ranks #17 with a gap of +3.26 pp.
Washington's AMI prevalence (27.14%) sits 4.16 points above the national average (22.98%), but its treatment rate (23.88%) is only 3.10 points above the national average (20.78%). The result is a gap that is slightly larger than the national mean but not an outlier.
Of Washington's 30 HPSA-designated counties, seven score at or above the federal high-priority threshold of 16: Ferry (20), Yakima (19), Pend Oreille (18), Stevens (17), Garfield (17), Lincoln (16), and Adams (16). Yakima County — the state's second-largest by HPSA score — serves a population of roughly 256,000 with severely limited access.
Higher prevalence states tend to have higher treatment rates, but the relationship is imperfect. Several high-prevalence states (Nevada, Texas, Idaho) show disproportionately low treatment rates, while DC and Massachusetts show high treatment rates relative to prevalence.
The distribution of MH treatment gaps across all 51 states is right-skewed, with a mean of +2.20 pp and a maximum of +8.66 pp (Nevada). A handful of states have negative gaps; these likely reflect limitations in small-area estimation methodology rather than genuine over-treatment.
Substance use disorder treatment gaps are consistently 2–5× larger than mental health treatment gaps across all states. The national SUD prevalence averages 18.64% while treatment rates are far lower, producing gaps that range from 6.08 pp (Alabama) to 20.21 pp (Oregon). Washington's SUD gap (15.55 pp) is close to the national median. Washington's MH gap is marked at rank #17.
Unmet need is calculated as the share of adults with AMI who received no mental health treatment in the past year. The national average is 52.0%. Washington's unmet need is 51.2%, placing it at rank #31 — just below the midpoint, meaning most states have a higher share of untreated AMI than Washington. Wyoming (61.8%) and Nevada (61.7%) have the highest unmet need nationally.
Each bubble represents a HPSA-designated WA county, sized by population and colored by HPSA score tier. Yakima County stands out as the highest-burden county combining a very high shortage score (19) with a large affected population (~256,000). Ferry County has the state's highest raw score (20) but a much smaller population (~8,500).
| Dataset | Description | Rows |
|---|---|---|
| SAMHSA NSDUH 2021–2022 | State-level AMI/SMI/SUD prevalence and treatment estimates | 51 |
| HRSA HPSA Designations | Mental health shortage area scores for WA counties | 30 |
NSDUH estimates are produced using Small Area Estimation (SAE) methodology and represent model-based statistical estimates, not direct survey counts. They carry confidence intervals that are not reflected in point-estimate comparisons. State-level estimates are available for all 50 states plus DC.
HRSA HPSA scores are composite indices (0–25) incorporating provider-to-population ratios, poverty rates, and travel distance to care. A score of ≥ 16 triggers federal high-priority designation for workforce development programs.
behavioral_health.db
├── nsduh_state (51 rows)
│ ├── state
│ ├── year
│ ├── ami_prevalence_pct
│ ├── ami_received_treatment_pct
│ ├── ami_unmet_need_pct
│ ├── smi_prevalence_pct
│ ├── sud_prevalence_pct
│ └── sud_received_treatment_pct
│
└── hrsa_shortage (30 rows)
├── county
├── hpsa_score
└── population_of_designation
Treatment gap is computed as ami_prevalence_pct − ami_received_treatment_pct.
Unmet need (ami_unmet_need_pct) uses a different denominator — it is the share
of adults with AMI who received no treatment — and is stored directly from NSDUH
rather than derived.
- All data is loaded into SQLite via named-column SQL queries (
SELECT col1, col2 …) to avoid ordering bugs fromSELECT *. - Charts are generated with Matplotlib only (no seaborn, no scikit-learn).
- K-means or clustering was not used; all groupings are threshold-based (e.g., HPSA ≥ 16).
behavioral-health-access-wa/
├── Code.py # All analysis and chart generation
├── behavioral_health.db # SQLite database
├── README.md
│
├── data/
│ ├── nsduh_state_estimates.csv # SAMHSA NSDUH 2021-2022, all 51 states
│ └── hrsa_hpsa_wa.csv # HRSA HPSA designations, WA counties
│
└── outputs/
├── fig1_treatment_gap_ranking.png
├── fig2_wa_vs_national.png
├── fig3_yakima_hpsa_rank.png
├── fig4_prevalence_vs_treatment.png
├── fig5_gap_distribution.png
├── fig6_mh_vs_sud_gaps.png
├── fig7_unmet_need_ranking.png
├── fig8_hpsa_bubble.png
├── treatment_gap_by_state.csv
├── wa_vs_national.csv
└── yakima_shortage_rank.csv
-
NSDUH estimates are modeled, not measured. State-level figures come from Small Area Estimation, which uses survey microdata combined with demographic covariates. Point estimates have confidence intervals that are not shown in the charts; small state-to-state differences (< 1 pp) are unlikely to be statistically meaningful.
-
Treatment gap ≠ unmet need. The gap metric (
prevalence − treatment rate) is computed from two separately estimated quantities, each with its own error. A "negative gap" (e.g., DC: −11.07 pp) does not mean more people are treated than have AMI; it reflects estimation uncertainty and should not be interpreted literally. -
HPSA scores reflect designated areas, not all counties. Only counties with an active HPSA designation appear in the HRSA dataset. Counties without a designation may still have provider shortages that fall below the federal threshold for formal recognition.
-
Single survey year. The analysis uses 2021–2022 NSDUH data. Behavioral health capacity and prevalence estimates shift year to year; these findings describe a snapshot, not a stable trend.
-
No causal claims. Correlations between HPSA scores, prevalence, and treatment rates in this analysis are descriptive. They do not establish that shortage area status causes lower treatment rates; unmeasured confounders (income, rurality, insurance coverage) likely contribute.
-
Population-level estimates only. NSDUH does not provide county-level data. The HRSA shortage data is county-level but limited to Washington State, so no county-to-county national comparison is possible.
# Install dependencies
pip install numpy matplotlib
# Run (generates all 8 charts + exports CSVs)
python Code.pyAll outputs are saved to outputs/. The SQLite database is created automatically on
first run and populated from the data in data/.
Data sources: SAMHSA NSDUH 2021–2022 State Estimates; HRSA Shortage Area Locator. NSDUH estimates use SAE methodology and should be interpreted with appropriate uncertainty.