Antidepressants in the News | The Body of Evidence

Antidepressants in the News


There's been talk recently that antidepressants may be dangerous and may be killing people. Antidepressants, and SSRIs in particular, have serious side effects and their effectiveness has been questioned recently. But, to claim they increase mortality goes a bit far. 

My latest for the Montreal Gazette:


A friend drew my attention to a recent study showing that antidepressants increase your risk of dying. Unsurprisingly, many media outlets picked up the story and ran with it, with varying degrees of alarmism.

The study was a meta-analysis of 17 trials. Its final conclusion was that antidepressants appeared to increase mortality by 33 per cent and cardiac events by 14 per cent. It also drew some other rather surprising conclusions. While antidepressants were linked to increased mortality in the general population, they apparently decreased mortality in cardiac patients. There also appeared to be no difference between SSRI antidepressants and the older tricyclic antidepressants.

To better understand these contradictory findings, it is worth reviewing the sometimes complicated history of how SSRIs (like Prozac, Paxil and Zoloft) got to market.

Before SSRIs, older antidepressants like tricyclics were available, but they had multiple side effects and toxicities. Doctors were eager for a safer alternative, among other reasons, because it had long been known that cardiac patients with depression did worse than those without. The link between these two diseases was thought to be serotonin. Lack of serotonin in the brain can cause depression, but seratonin is also involved in platelet aggregation and clot formation. Thus a medication that could boost serotonin levels in the brain and prevent platelets from clogging the arteries would be most welcome. This is exactly what SSRIs do.

Unfortunately, as happens so often in science, a beautiful theory was slain by an ugly fact. The SADHART clinical trial with sertraline (Zoloft) found that while it did improve depression symptoms, it had no effect on cardiac outcomes. The ENRICHD trial later showed similar results. While the new class of medications had few side effects for the heart, they didn't improve survival or reduce heart attacks.

Thus the current meta-analysis is hard to reconcile with the existing facts. First off, the SADHART and ENRICHD trials were not part of the 17 studies in the review. This does introduce some doubt in the analysis. It is also somewhat hard to imagine how a medication could cause heart attacks in the general population, but prevent heart attacks in people who have already had one.

There are other problems. Double counting may have played a role here, as well. The problem is statistically subtle but, simply put, it results when reviewers include data from the same study (usually from different subgroups) multiple times in the same analysis. 

As well, the results seem to have been driven by a single observational study. It suggested that SSRIs increased the risk of death, but tricyclics did not, which is inconsistent with the existing data. More likely, older, sicker patients with cardiac problems were preferentially given SSRIs because of the known toxicities of tricyclics.

SSRIs certainly have their downsides. A 2008 study found that many of the studies showing SSRIs to be ineffective were never published. Another 2015 study in the British Medical Journal re-analyzed data from the now infamous Study 329 and found that SSRIs like Paxil seemed to be associated with increased suicidal attempts and thoughts in young patients. One of my earliest research projects looked at how SSRIs can potentially increase the risk of bleeding when combined with medications like Aspirin and Plavix. Nevertheless, they do have a role to play. Analyses suggest they have little benefit in people with mild and moderate depression, but do seem to help people with severe symptoms. Serious side effects were underreported and only came to light years later. But their toxicity potential does seem to be less than their predecessors.

You can make a good argument that SSRIs were, and still are, overprescribed to many people who may not need them. But for those who do need them, this new meta-analysis shouldn't be a source of worry.