Mood Swings Predicted Cannabis Use in Teens With ADHD and Depression | Research

The takeaway

Among 13,025 adolescents with ADHD or depression, mood instability was associated with 61% higher odds of cannabis use in ADHD and 38% higher odds in depression, supporting the self-medication hypothesis.

Child psychiatrists, ADHD specialists, adolescent medicine clinicians, and mental health researchers.

What the researchers found

Mood instability was associated with increased cannabis use in both ADHD (aOR: 1.61, 95% CI: 1.41-1.84) and depression (aOR: 1.38, 95% CI: 1.21-1.57) after adjustment for covariates. The association was 25% stronger in ADHD than in depression, suggesting mood instability is a particularly important cannabis use predictor in ADHD populations.

Why it matters

Understanding why adolescents with ADHD and depression use cannabis could lead to better prevention strategies. If mood instability drives self-medication with cannabis, then treating mood instability directly might reduce cannabis use more effectively than anti-drug messaging.

The numbers in context

13,025 adolescents (7,985 ADHD, 5,738 depression). ADHD: aOR 1.61 (95% CI: 1.41-1.84). Depression: aOR 1.38 (95% CI: 1.21-1.57). Mood instability associated with 25% higher probability of cannabis use in ADHD compared to depression.

How the study worked

Natural language processing of electronic health records from 13,025 adolescents (ages 11-18) with primary diagnoses of ADHD (n=7,985) or depression (n=5,738). NLP identified references to mood instability and cannabis use in clinical notes. Logistic regression adjusted for sociodemographic and clinical covariates.

What this study cannot tell us

NLP identification of mood instability and cannabis use from clinical notes may miss mentions or introduce false positives. Cross-sectional analysis cannot establish whether mood instability causes cannabis use or vice versa. Electronic health records capture clinical encounters, not daily life. Cannot account for all confounders.

How to read the evidence

Large clinical cohort with innovative NLP methodology, but cross-sectional design and reliance on clinical documentation limit causal inference.

When this study was published

Published 2025.

The bigger picture

The self-medication hypothesis has been discussed for decades but rarely tested with this level of precision. Using NLP to mine clinical records from over 13,000 patients provides more ecologically valid data than surveys or lab studies, capturing real-world patterns in how mood problems and cannabis use co-occur.

Questions still open

Whether treating mood instability in ADHD reduces cannabis use
Whether the relationship is bidirectional, with cannabis use also worsening mood instability

Common questions

Are teens with ADHD self-medicating with cannabis?

This study provides indirect support for that idea. Mood instability, a common feature of ADHD that is often undertreated, was strongly associated with cannabis use. However, the study cannot prove that teens are intentionally using cannabis to manage mood swings.

What is natural language processing in medical research?

NLP uses algorithms to extract information from unstructured text, like doctors' clinical notes. Instead of relying on diagnosis codes alone, NLP can identify when clinicians wrote about mood instability or cannabis use in their notes, capturing information that might not appear in structured data fields.

Read the original research

Mood instability as a transdiagnostic predictor of cannabis use in attention-deficit/hyperactivity disorder and depression: A natural language processing analysis of electronic health records from 13,025 adolescents.

European psychiatry : the journal of the Association of European Psychiatrists, 68(1), e139

View on PubMed ↗Publisher’s paper ↗

Citation

Seker, Asilay; Bullock, Edward; Chandler, Susie; Patel, Rashmi; Quattrone, Diego; Colling, Craig; Sonuga-Barke, Edmund J S; Downs, Johnny. (2025). Mood instability as a transdiagnostic predictor of cannabis use in attention-deficit/hyperactivity disorder and depression: A natural language processing analysis of electronic health records from 13,025 adolescents.. European psychiatry : the journal of the Association of European Psychiatrists, 68(1), e139. https://doi.org/10.1192/j.eurpsy.2025.10095