[saltcured]

In a study of voluntary use like this, it seems like the null hypothesis is that people who were going to be diagnosed with mental illness chose to self-medicate ahead of time?

I don't understand how you can counter this potential selection bias, given that it would not be ethical to do a randomized trial instead of voluntary usage.

[JasonADrury]

The finding is:

> past-year cannabis use was associated with a significantly increased risk of incident psychotic, bipolar, depressive, and anxiety disorders by age 26 years

They claim correlation, not causation.

They also address this subject in the discussion:

>Existing research suggests that the relationship between adolescent cannabis use and psychiatric disorders is complex and bidirectional. Adolescents with mental health symptoms or diagnoses may use cannabis as a way to mitigate distress, cannabis may contribute to mental health symptoms and diagnoses through neurobiological changes, and there may be shared social and biological risk factors that contribute to both mental health symptoms and cannabis use.

>Despite potential use of cannabis to self-medicate mental health symptoms, ongoing use of cannabis is associated with worsening mood symptoms29 and poorer adherence to medication treatment and psychotherapy.

>While it is not possible to definitively determine causality, this study had a strong retrospective cohort design. The temporal order of cannabis use preceded incident psychiatric disorder diagnoses by a mean of 1.7 to 2.3 years, supporting the possibility of a contributory role.

>The findings remained significant even after adjusting for a history of psychiatric disorders and other time-varying substance use and excluding adolescents with any history of a psychiatric disorder in sensitivity analyses, indicating unique associations between adolescent cannabis use and psychiatric disorders that go beyond broader adolescent psychopathology or substance use.

>This was a conservative approach, as other psychiatric disorders might be mediators or confounders depending on the timing and the underlying pathways. Furthermore, E-values indicated that only a strong unmeasured confounder with HRs of 3.79 to 1.79 could explain the associations, suggesting that adolescent cannabis use may be an independent risk factor.

>However, reverse causation cannot be ruled out, as some individuals may begin to use cannabis to self-medicate prodromal symptoms of psychiatric disorders even before a diagnosis is made. Future research with more nuanced measurement of cannabis use, including frequency, mode of use, and product strength, alongside regular screening and assessment for psychiatric disorder symptoms and diagnosis would help to further elucidate the timing and mechanisms underlying these associations.