America’s mental-health headlines can change when the measuring system changes. The 2025 National Survey on Drug Use and Health reports encouraging trends and substantial unmet need. It also updates the model used to estimate mental illness. Comparing its figures with an older press release can confuse a methodological revision with a change in people’s health.
Key findings
23.4% versus 21.5%: estimates of any mental illness for the same year, 2024, under the older and updated models. [1]
28.6%: estimated any mental illness among adults aged 18 to 25 in 2025, using the updated model. [2]
50.9%: adults with any mental illness who received mental-health treatment during 2025. Receipt is not a recovery measure. [3]
72.9%: respondents in a defined unmet-need subgroup who said they should manage on their own. This is not a finding about all Americans or wealthy executives. [4]
The same survey year, two different models
| Measure | Older model | Updated model |
|---|---|---|
| Any mental illness | 23.4% | 21.5% |
| Serious mental illness | 5.6% | 7.8% |
Both columns describe 2024. Serious mental illness is included within any mental illness; the rows must not be added. Source: SAMHSA methodological report, table 3.12. [1]
The 2025 products apply the new DSM-5-based model to the latest survey and retrospectively to 2021 through 2024. The revision moves the two categories in different directions.
An uncritical comparison could imply improvement in one measure and deterioration in another. The safer approach is to use the internally consistent historical series supplied with the new release.
The revised history still contains encouraging findings
The values all come from the 2025 release rather than earlier editions. Source: SAMHSA table 9.56B. [2]
The positive direction belongs in the story. A media-focused research program should not turn every release into deterioration. These are repeated population estimates, however, not evidence that the same young adults recovered over time.
Access remains a separate problem
Among adults with any mental illness, 50.9% reported mental-health treatment in 2025. Among those with a major depressive episode, the figure was 65.9%.
Within that depression group, treatment receipt was 59.1% for ages 18 to 25, 68.6% for ages 26 to 49, and 67.6% for ages 50 and older. These are treatment-use figures, not measures of adequacy or remission. [3]
The diagnostic groups overlap. They should not be added, and the age differences do not prove one generation is more motivated to recover. Need, access, preferences, and other circumstances require separate investigation.
The barriers are not exclusively financial
| Reason | Share reporting it |
|---|---|
| Should handle the problem alone | 72.9% |
| Expected treatment to cost too much | 61.4% |
| Insufficient time | 48.3% |
| Did not know how or where to obtain care | 44.9% |
Respondents could report several reasons. The percentages overlap and describe that unmet-need subgroup, not all Americans. Source: SAMHSA’s July 2026 release. [4]
For affluent-client research, the useful question is what remains difficult when affordability is less of a barrier. These national data do not answer it for high-net-worth households.
Self-reliance, professional responsibilities, and uncertainty about whom to approach are topics for a properly sampled study. They should not be assigned to every founder, executive, or family as assumed characteristics.
Overdose deaths fell, using clearly defined periods
CDC’s May 2026 provisional release estimated 69,973 drug-overdose deaths in 2025, compared with 81,313 in 2024. Opioid-involved estimates fell from 55,296 to 44,564. These figures were provisional and subject to revision. Drug categories overlap and must not be added as mutually exclusive totals. [5]
An August 2026 release reported 67,798 predicted deaths for the twelve months ending March 2026, down 12.3% against the same window a year earlier. That rolling window overlaps calendar 2025. It is not a completed 2026 annual total. [6]
The reduction matters without establishing how many people obtained treatment, sustained recovery, or avoided nonfatal harm. Mortality and recovery are related concerns, not interchangeable measures.
Nor do the totals establish which intervention caused the decline. A clinic should not claim credit for a national change without evidence of its contribution.
Why the source version changes the comparison
Historical estimates can change when records arrive or methods are revised. A chart assembled from old press releases may combine incompatible versions of the past.
For survey and mortality series, retain the release version and observation period. Use the consistent history supplied with a revision or label earlier as-published figures explicitly.
This matters for retrospective annual reports too. A report about 2024 written in 2026 can use later evidence about 2024, but should not imply those numbers were already available at the time.
What private treatment should add
National treatment receipt is not a benchmark for a clinic’s success. A private provider should describe its eligible patients, assessed needs, treatment content, and follow-up.
Symptoms, functioning, satisfaction, and completion should remain separate measures. For someone with co-occurring needs, a single diagnosis-specific percentage may leave an important part of the clinical picture unmeasured.
Money, convenience, and discretion can influence a choice. They do not establish a clinical advantage. Ask what evidence supports the proposed plan and how progress will be reviewed when the first approach is insufficient.
Cross-border care needs a receiving plan
A US resident considering care abroad should understand who assesses the needs, who holds clinical responsibility, and what receiving arrangements exist on return.
The question is not whether international treatment is inherently better or more fragmented. It is whether continuity is planned and appropriate under the relevant rules. A family office can organize logistics without replacing clinical responsibility.
No US-versus-European effectiveness comparison is reported here. A national survey rate and a foreign clinic’s selected outcomes would not constitute a matched study.
Five checks before quoting the numbers
| Check | Why it matters |
|---|---|
| Which model? | Old and updated classifications differ |
| Which population? | All adults and unmet-need subgroups are different denominators |
| Which period? | Annual and rolling windows may overlap |
| Provisional or revised? | Later data can change historical totals |
| Access or outcome? | Treatment receipt does not establish recovery |
The bottom line
The US evidence contains reasons for cautious optimism and a substantial need to understand who still does not receive care. The updated model makes careless historical comparisons especially risky. For families with more treatment options, the next question is how to choose appropriate care and verify what it changes.
For journalists
Key finding: the updated model changes estimates for the same survey year, while the consistent young-adult series shows a lower 2025 estimate.
Important caveat: national figures do not establish HNW-specific barriers or the effectiveness of private treatment.
Suggested attribution: THE BALANCE analysis of SAMHSA survey publications and CDC surveillance releases.
Methodology and sources
This narrative analysis separates revised survey estimates, treatment-use measures, and provisional mortality surveillance. The source periods and release versions remain explicit. No new survey or proprietary patient analysis was conducted.


