Public attention
Google Trends
Monthly relative search-interest indices for 12 mental-health topics from June 2016 through June 2026.
Every signal says something. No signal says everything.
This page keeps the definitions, source logic, comparability decisions, and caveats close to the research without asking them to carry the main story.
Clinical reality
CDC, SAMHSA and federal benchmarks
Condition-specific prevalence, diagnosis and treatment estimates are kept in their original populations and definitions rather than forced into a synthetic common scale.
Treatment signal
MEPS, NSDUH and NSCH
Public treatment estimates, including broad MEPS condition-to-prescription-event linkage, shown with explicit population and comparability limits.
Search Trends methodology
How the original exploration is calculated
What the index means
A value of 100 represents peak relative popularity within the underlying Google Trends series. It is not a count of searches.
What can be compared
Each topic is indexed independently. The analysis compares trajectory, direction, rate of change, and seasonal shape. It does not compare raw popularity across conditions.
Derived measures
Long-term change compares the first 12-month average with the latest 12-month average. Recent momentum compares the latest 12 months with the preceding 12. Seasonal values average observations for each calendar month.
Time span
The current workbook contains 121 monthly observations from June 2016 through June 2026.
Cross-source comparisons
Search, diagnosis, treatment and research are presented as distinct signals. Annual aggregation and within-condition percentage change are used only when the underlying definitions support them.
Monthly correlation field
Pearson correlations use month-to-month changes over the 38-month overlap from January 2020 through February 2023. The matrix compares 12 search topics with eight labor, sentiment, and mortality indicators. It is exploratory, includes 96 simultaneous comparisons, and does not test causation or lagged relationships.
Prescription guardrail
MEPS can support broad condition-linked treatment analysis through its prescribed-medicine event linkage. CMS Part D can describe drugs and prescribers but generally cannot identify the condition motivating a prescription. Mental Currents therefore does not label Part D drug volume as condition-specific treatment.
Definitions
A shared vocabulary for unequal sources
- Attention
- Observed information-seeking or public-interest behavior. It is not prevalence.
- Clinical signal
- A diagnosis, claims, encounter, or condition proxy. It may not represent every person with the condition.
- Treatment signal
- A prescription, dispensing, procedure, or care proxy. It may reflect multiple indications or access patterns.
- Divergence
- Two signals moving in different directions or at meaningfully different rates. It does not explain why.
Search interest is not diagnosis. Clinical activity is not prevalence. Treatment is not need.
Media coverage, awareness campaigns, coding practices, access to care, changing definitions, platform behavior, population coverage, and data availability can all change the observed series. Mental Currents makes those limits part of the finding.
Datasets and tools
Research products will be released with their provenance attached.
The interactive Search Trends dataset is available now through the live explorer. Downloadable enriched datasets will be added only after source mapping, comparability notes, and reuse terms are complete.
Open the live explorer →A reader’s guide
Three rules for reading the evidence.
Separate attention from prevalence. A topic can become more visible without becoming more common.
Check the population and year. National estimates often describe different ages, definitions, and periods.
Treat disagreement as information. When signals diverge, the gap is a question to investigate, not a conclusion to disguise.