Here’s the second in our intermittent series of PhD students describing their research and why it’s useful / awesome. Suzi Gage, as well as doing a PhD in epidemiology, writes for her blog Sifting the Evidence which is hosted by the Guardian, runs marathons and is part of a band called Glis glis. There’s plenty more I could say but if you want more from her, I’d suggest her twitter @soozaphone.
Starting my PhD investigations into cannabis use and psychosis made me increasingly aware that everyone has their own explanation or interpretation of their own thoughts and experiences. As a psychologist I’m used to people having an opinion on what I do; who doesn’t have an opinion on the workings of their own mind? Perhaps it’s because of the illicit nature of researching an illegal substance, perhaps it’s due to the devastating effects that psychosis can have on a young person, but strangers’ responses to my research topic fall into two categories. Usually it’s ‘oh, can I be a participant?’, and occasionally ‘my friend smoked pot and had serious mental health problems afterwards’.
From these responses I think it’s possible to tell a lot about my research area, and its importance. Firstly, a lot of people have at least tried marijuana, it’s the most widely used illicit drug in the UK (although both alcohol and cigarettes are often used by people who are underage, so technically this is illicit too). If cannabis was indeed affecting mental health, it would mean a large number of young people are putting themselves in danger. The second response is also revealing. Although psychosis is a relatively rare disorder, it can be devastating to sufferers. Mental health illnesses are often stigmatised, but during teenage years when a person is discovering their identity, and is undertaking important schooling that can determine their future career path, the effects can be severely detrimental.
Understanding whether cannabis has a causal effect on the likelihood of developing psychosis is a hugely important question, and one which is fraught with difficulty. Given the prevalence of cannabis use but the rarity of psychosis, it’s not the case that everyone who smokes is likely to get ill; the vast majority of cannabis users will not develop psychosis. Something more nuanced is going on. Are a certain subgroup of people more vulnerable? Is it the heaviest users? Is it people who become dependent? Is it those with a family history of mental health problems who are more vulnerable? There are other possibilities as well. It may be that there is no causal relationship at all, but that something else affects both likelihood to smoke cannabis, and susceptibility to mental health problems. For example, adversity in childhood could affect both, so it’s not the cannabis causing the increased risk of psychosis, but rather acting as a marker for risk.
Teasing apart this question is tricky. What would be ideal, from a research perspective, would be to take a group of young people who’ve never used cannabis, and randomly assign half of them to use it, and half not to. This would mean the family background factors and other things that differ between people who decide to smoke and those who don’t should be roughly equivalent between groups, so any evidence that psychosis is more common in cannabis users would suggest a causal relationship. For obvious reasons this is both ethically and practically a non-starter. This means research looking in to the question relies on observing people. This comes with its own problems. As noted above, people are different from each other for all sorts of reasons. Those who choose to smoke cannabis will be different from those who don’t. Accounting for all these difficulties is really hard. There may be some differences we’re not aware of. We can’t take these in to account, but these could still be the drivers of any association we see between cannabis and psychosis.
There are some methods currently being developed that hope to get around these problems. A technique called Mendelian Randomisation uses genes as proxies for behaviours. Because the way we inherit our genes is unrelated to either the environment, or (for the most part) our other genes, the systematic differences between people who smoke and people who don’t will be unrelated to the gene we use as a proxy, even if we don’t know about these differences. At the moment, it’s known that cannabis use is partly predicted by genetics, but an individual gene associated with cannabis use has not been found. However, a risk score can be created from a number of genes that all seem to be slightly related to cannabis use. This risk score can then be used as a proxy in the same way. This technique effectively allows us to use observational data in a randomised design. I’m really hopeful that this method might allow researchers to really get a handle on whether cannabis use is causally related to risk of psychosis, or if this association is merely an indication of another predictor of psychosis. This will mean we can better warn teenagers of the risks of this kind of drug use, and potentially inform evidence-based drugs policy.