How researchers get heard
Abstract lines

Skunked by Correlation vs. Causation

Understanding correlation versus causation is one of the abiding problems in science journalism.

And, I’d argue, science communications takes advantage of this problem again and again, to everyone’s detriment.

Case in point: A new study in The Lancet Psychiatry links incidence of psychosis and daily use of cannabis.

It finds that

  • People who use cannabis daily are three times more likely to have a psychotic disorder than those who have never used the drug;
  • Daily users of high-potency cannabis (cannabis with THC levels above 10 percent) are four times more likely to report psychotic disorders than never-users; and
  • Rates of new diagnoses of psychosis were highest in three European cities in the study — London, Paris and Amsterdam — where high THC cannabis is readily available.

The results are correlative, not causative.

They’re also not precisely new, although they do extend the correlation.

The problem with correlative research on this issue, as Aaron Carroll wrote in the New York Times last year, is that “it’s hard to establish the arrow of causality. Are people who smoke pot more likely to develop mental health problems? Or are people with mental health problems more likely to smoke pot?”

To put it another way: The National Academies report on cannabis makes clear that people with predisposing risk factors for schizophrenia should refrain from using cannabis.

But that’s a different message than saying high-frequency use of cannabis causes psychosis.

Unfortunately, that was the message of all eight news articles covering the study I read this morning.

All but one treat the findings as locked-down conclusive, skirting the issue of correlation and causation.

Several of these articles then led into a discussion of whether cannabis should be legalized.

The NPR piece is the most honest on correlation and causation. While it has this quote high up declaring the link is definitive:

“This is more evidence that the link between cannabis and psychosis matters,” says Krista M. Lisdahl, a clinical neuropsychologist at the University of Wisconsin, Milwaukee, who wasn’t involved in the study.

It (much) later walks back a little on the issue of causality:

However, the study doesn’t prove causality, cautions Dr. Diana Martinez, a psychiatrist and addiction researcher at Columbia University. “You can’t say that cannabis causes psychosis,” she says. “It’s simply not supported by the data,” she says.

Lisdahl agrees. In order to show causality, one would have to follow people over time — before they started using weed to years later when they have their psychotic episodes, she says. “You need twins in the studies, you need genetic information,” among all other kinds of data, she says.

Psychotic disorders like schizophrenia and bipolar are complicated, “multi-faceted disorders,” notes Gage.

“In all psychotic disorders, there is this multiple hit hypothesis,” says Martinez. Many factors influence whether and how these disorders manifest.

There seems to be disagreement among the study authors on the policy implications of the findings. NBC quotes lead author Maria Di Forti as both advising caution when using high-potency cannabis, but seeming to advocate for the use of CBD:

“CBD, which we refer to as the ‘good guy’ component of cannabis, has no psycho-active component and has even been show to offset the psychoactive components of THC in experimental studies,” said Di Forti.

Di Forti cautions that until more research is done, it is in most people’s best interest to stay away from high potency cannabis.

“If you want to experiment with it, do it rarely,” she said.

Di Forti’s co-author Robin Murray, on the other hand, tells MSN that the study’s implications are clearly negative for legalizing pot — an issue of debate in England:

Prof Sir Robin Murray, another author of the study from King’s College London, said the study has implications for the debate on whether cannabis should be legalised.

“If you are going to legalise cannabis, unless you want to pay for more a lot more psychiatric beds and a lot more psychiatrists, then you need to devise a system where you would legalise in a way that wouldn’t increase the consumption and increase the potency,” he said.

Meanwhile, the BBC quotes a London-based drug and alcohol addiction center spokesperson that “it’s important not to over-react” about cannabis use:

Nick Hickmott from the drug and alcohol charity Addaction said: “We’ve got a problem with potency. People who regularly take lots of high strength cannabis are at risk of potentially serious harm. It can be particularly harmful for younger, developing brains.

“My advice is avoid using high-strength cannabis every day and pay attention to how it makes you feel. If you end up feeling anxious or just unsettled then it might be best to give it a miss. It’s also not a good idea to mix it with alcohol or other drugs.

“It’s also important not to over-react. Lots of people experiment with cannabis and then move on without any problems. For people who do need advice or help I’d recommend reaching out to a GP or a local drug service.”

German Lopez of Vox puts his finger on the problem:

I suppose this is a win for The Lancet’s PR machine.

It’s a loss for public understanding of the risks involved with cannabis use and legalization. I read eight articles as well as the journal article and still have numerous questions. Imagine the people who read one, or the headline of one, or heard a 15-second item on television or a biased discussion of the results on talk radio.

One solution: The journal should issue a fact sheet that deals not just with the findings of the study, but all of the potential implications of the study. The authors have an obligation to speak to those. That’s good and responsible science communications.

Another solution (as a consumer of medical news): Read researcher thought leaders who have read the literature and assessed from a 40,000-foot level both the risks and rational ways to view them.

Otherwise, findings tend to get weaponized, as they seem to be in this case. As, one could perhaps reasonably conclude, the authors and The Lancet wanted them to be.