OK when a hypothesis is suggested, we first need to ask "is it biologically plausible?". This is the minimum barrier of entry before conducting a rigorous investigation. Is it plausible that drugs like Ozempic COULD cause suicide?

It is. For one, we know the drugs cross the blood-brain barrier, a pre-requisite (more or less) for any drug to have psychological effects (positive or negative). More than that, as I wrote about recently, the drugs do more than curb appetite for food. medscape.com/viewarticle/oze…

Studies have shown that drugs like Ozempic curb smoking, compulsive gambling and shopping, alcohol intake. All good things, of course, but you wonder if they are interacting with some reward pathway in the brain and, for some people at least, messing with that can lead to despair?

So it's possible. Fine. How do you figure it out? Randomized trial data is not great for this. For one, death from suicide is rare and so you need a lot of people in a study to see a signal. But also, the type of person who participates in an RCT is unlikely to have significant psychological problems.

You need real-world data, as imperfect as it is, to get to the bottom of this, and that's where this new study, appearing in @JAMAInternalMed comes in. jamanetwork.com/journals/jam…

Researchers identified every patient with diabetes in Sweden and Denmark who started a GLP-1 drug from 2013-2021 (N=124,517). As a control group, they chose individuals starting an SGLT2-inhibitor (N=174,036).

This is not an RCT, of course - these groups differ in some substantial ways at baseline. Broadly, GLP-1 group seems a bit healthier in terms of physical problems, a bit worse in terms of psychological problems.

Propensity-score matching balanced these baseline characteristics rather nicely. For a primer on that fun topic see: medscape.com/viewarticle/mou…

Groups statistically balanced, you can compare outcomes - in this case death from suicide. Overall HR 1.25 (95% CI 0.83-1.88). That is NOT a statistically significant result. Conclusion: no evidence for a link between GLP1 drugs like Ozempic and suicide. But...

That wide confidence interval can be interpreted as: Data like what we observed here is not unexpected in a reality where Ozempic DECREASES suicide risk by 17% OR increases it by 88%. The wide range, due to the low risk of suicide overall, means that we have not ruled out an effect that may matter.

Yes this is the old "more data is needed" thing. We actually do have a bit more data... This @NatureMedicine study found Ozempic significantly DECREASED the risk of death from suicide. But the lack of an incident-user design may select for better outcomes. nature.com/articles/s41591-0…

And it's worth noting that Sweden and Denmark are not the US. Scandinavia has a high national happiness rate and a paradoxically high suicide rate as well: medium.com/@dtg319/happier-c…

So the jury is still out, but I'm not terribly worried. As these drugs get more ubiquitous, we will have more clarity regarding risk. For now, worth remembering that they do affect the brain - and to take any concerns in that domain seriously.

More thoughts in my @Medscape column here: medscape.com/viewarticle/oze…