A population-based study from Scandinavia provides some real-world information.

In July of 2023, the European Medicines Agency – basically the FDA for the European Union – released a statement that threatened to put the brakes on what had been the runaway success of the new weight-loss drugs: GLP-1 receptor agonists like semaglutide (Ozempic) and Mounjaro.

The agency announced they were conducting a formal review into reports that the use of these drugs could increase the risk of suicide and suicidal thoughts.

The reason for the review? The Icelandic Medicines Agency had received reports of up to 150 people taking the drugs reporting suicidal thoughts or self-injury.

Now, the easiest way to prove that a drug might increase the risk of death from suicide is to look at randomized trial data – if individuals randomized to a GLP-1 receptor agonist die from suicide at a higher rate than those randomized to placebo, you’ve got your causality signal. But suicide is a rare thing – occurring in about 14 per 100,000 people in the US - and people who enroll in randomized trials are often heavily screened to ensure they don’t have any psychological issues at baseline. It’s no surprise, then, that the randomized trial data has not shown a psychiatric safety risk for the new weight loss drugs.

Which means, if you really want to figure out if this risk is real, you have to use real world data. And real-world data is like crude oil: hard to come by, difficult to extract and messy. Fortunately for us, we have Scandinavia.

They say that countries like Sweden and Denmark have some of the happiest populations in the world. I suspect this is driven largely by epidemiologists, who are thrilled to have access to medical datasets that basically encompass every person in the entire country for their whole life.

You can use those datasets for some pretty impressive things – like trying to figure out if Ozempic causes suicide.  Which is exactly what researchers did in an analysis appearing in this study in JAMA Internal Medicine.

But first things first. Is it even biologically plausible that Ozempic could lead to suicide? I think it is, actually. First off, we know that the GLP-1 receptor agonists act on receptors in the brain – any drug that can get across the blood brain barrier, in theory, can have adverse (or positive) psychiatric effects. Also, as I wrote about recently, these drugs can have pretty dramatic effects beyond their ability to curb appetite for food – decreasing compulsive gambling and shopping, reducing alcohol intake, and helping to quit smoking. These are all good things of course, but you wonder if they could be having some effect on psychological reward systems, and maybe disrupting that, in some people, could lead to despair?

To figure it out, the researchers identified everyone in Sweden and Denmark who started taking semaglutide or liraglutide for diabetes from 2013 to 2021. As a control group, they identified everyone with diabetes who started taking an SGLT2 inhibitor in the same time frame. Both drug classes are used for diabetes treatment, but it is worth noting that SGLT2-inhibitors don’t have much effect on weight.

In any case, we have 124,517 adults who started taking Ozempic or an Ozempic-like drug, and 174,036 who started taking one of the SGLT2-inhibitors.

This is not a randomized trial, of course. People who were started on Ozempic or related drugs were different than those started on SGLT2-inhibitors. Showing you the Swedish data here, you can see that those taking Ozempic or similar agents tended to be more often female, less likely to have cardiovascular disease, more likely to have a diagnosis of obesity, and more likely to be taking insulin.

From a psychiatric standpoint, prior to starting the drugs, those who were prescribed Ozempic or sister drugs were more likely to be using or to have previously used antidepressants, and to have had an outpatient visit for a psychiatric diagnosis.

The authors controlled for these important differences using propensity scores – which I’ve covered before if you want a little primer. After that process, the groups were much more balanced.

That done, they looked forward in time to see who would, well, die from suicide. Or suffer non-fatal self harm, or (among those without a prior psychiatric diagnosis), develop a new diagnosis of anxiety or depression.

The topline results are here. 77 new Ozempic or related drug users died from suicide (that’s 6 out of 10,000 people). 71 new SGLT2 inhibitor users died from suicide (that’s 4 out of 10,000 people).

After accounting for the differences between the two groups at baseline, the authors calculate an overall risk for suicide that is 25% higher for the GLP-1 receptor agonist users, though the margin of error is large here – ranging from a 17% reduction in risk to an 88% increase.

When death from suicide and non-fatal self-harm were combined, GLP1 receptor agonists appear to be protective.

Looking at new psychiatric diagnoses, it seems to be a total wash.

What have we learned here? At the top, I explained how randomized trials are really not ideal for detecting differences in rare events like suicide, and we need to depend on real-world data.

And yet here it is, more than 100,000 users of Ozempic and similar drugs with a reasonable control group, and the outcome is equivocal. The headline is certainly – “there is no significant association between Ozempic and suicide” – the results of this study are 100% consistent with that statement, because it’s clear you could see results like this due to chance alone – that’s why the 95% confidence interval of risk crosses 1.

Negative results are tricky though. You can conclude no relationship was evident from this data– which is the case here – but it’s harder to conclude that there is no risk whatsoever. To do that, you need to look at the range of the effect estimate – and this study – this high-quality, large, population-based study, has data that are consistent with Ozempic and sister drugs preventing suicide, or increasing the rate significantly.

I should point out that this is not the only study of this type. This one, in Nature Medicine, looked at more than a million patients and found that the use of Ozempic and similar drugs might substantially decrease the rate of death from suicide.

And of course, suicide is a product of multiple interacting factors – the individuals psychological state, the services available to them, the culture they are immersed in, and so on. Scandinavia – with its high national happiness and paradoxically high suicide rate – may not be comparable to the US.

For now, the weight of evidence does not suggest a major risk here. But it’s not a bad idea to remember, whether prescribing these drugs or taking them, that they do effect the brain – that is how they work – and to be on the lookout for any changes, positive or negative, that might result from that interaction.

A version of this commentary first appeared on Medscape.com.