article in Biological Psychiatry reviews evidence about the association between cannabis use and psychotic outcomes. Despite the consistent association, the article highlights how difficult it is to infer a causal link because of confounding and bias in the data.
As I said in my BMJ letter, the use of cannabis can cause emotional problems and people may use it to deal with their emotional problems. Cannabis use is likely to be a proxy measure for poor premorbid adjustment associated with psychosis. As the article says, "few studies have adjusted for measures of early life attachment, abuse, and trauma".
Bias may also be introduced as heavy users of cannabis may be rarely unintoxicated, leading to misdiagnosis of the induced psychotic-like experiences, which are usually transient in less heavy users. There is some evidence of a dose-response relationship between cannabis and psychotic diagnosis.
There is also an association between other drugs and psychosis and mixed data about whether the association with cannabis is more specific. Despite the increase in the use of cannabis since the 1960s there is no clear evidence of a corresponding increase in the incidence of psychosis. Cannabis exposure among adolescents and young people is common and psychosis remains rare.
Despite highlighting the methodological difficulties of making causal inferences from observational studies, the article suddenly jumps to the conclusion that, "There is no doubt that a public health message that cannabis use is harmful is appropriate". This leads to today's Guardian editorial saying that what it calls the "small risk of a dreadful outcome", ie. psychotic breakdowns that "smash up lives and can lead to full-blown schizophrenia", is something "well worth a proper public health campaign".
Of course cannabis can cause harm, as can alcohol. But, as the article points out, it is important to have the facts right for any public health campaign to be effective. The causal link between cannabis and psychosis has not been proven.