Antidepressants aren't much better than sugar pills

Computational network meta-analysis shows that SSRIs barely outperform placebo in treating depression.
by Tom Cleveland
February 9, 2018
6 minute read

Some quick history

In 1987 the FDA approved Prozac and started a pharmaceutical revolution. It was immediately hailed (read: marketed) as a wundercure—all of the benefits of older antidepressants without the gruesome side effects. Science had found a cure for sadness, and for 14 years Eli Lilly, its creator, printed money.

Delicious Prozac.

In 2001 Prozac’s patent was up, but the baton quickly passed to Pfizer, whose Zoloft (sertraline) was still patent-protected. Why didn’t people just start taking generic Prozac instead of the very-similar and very-expensive Zoloft? The cynic in me wants to tell you it was the billions of marketing dollars spent by Pfizer to sway the hearts and minds of consumers and doctors. But there’s actually compelling evidence that Zoloft is better than Prozac (more on that later). Anyway, big pharma continued its giant money party by releasing a series of slightly improved SSRIs in a suspiciously well-timed manner until the patent ran out on Lexapro (escitalopram) in 2012. In other words, big pharma milked this golden calf for 25 years, during which time the prescription of antidepressants became commonplace, often for conditions other than depression.

For a quarter-century we witnessed a deluge of direct-to-consumer marketing about SSRIs. And even if you’re not totally on-board with every man, woman, and child pouring Paxil into their corn flakes, you probably believe these drugs are very potent. Critics of overmedication usually paint pictures of a subdued population, their active minds crushed by a powerful psychoactive drug.

But the research paints a different picture.

Looking at the evidence

bar graph
Relative efficacies of SSRIs. Read more here.

GlacierMD is a new project that uses software to run large meta-analyses that are difficult for human researchers. You can read more about our mission here.

At the time of writing, GlacierMD has examined 53 SSRI studies on almost 10,000 participants, covering all 5 major SSRIs (fluoxetine, sertraline, paroxetine, citalopram, and escitalopram). We’re adding more studies every day, but at this point we feel pretty confident in our statistical analysis.

The first thing we noticed was that, with the notable exception of Celexa (citalopram), the newer drugs are always more effective. Presumably this is because big pharma wouldn’t bother going through FDA approval for a drug they know doesn’t work as well.

The next thing is that the newer drugs are also always safer. The risk difference for Prozac’s most common side effect, nausea, is a whopping 25%, while Lexapro’s top complaint (insomnia) only affects an additional 4% of participants. (Read more about side effect calculations here.)

So everyone should take Lexapro, right?

Not so fast. So far these comparisons are all relative. Among all SSRIs, Lexapro appears to be the safest and most effective. But exactly how effective is it?

To answer that question, let’s look at a totally different class of drugs: over-the-counter painkillers.

Tylenol—stronger than every SSRI.

GlacierMD has determined that Tylenol (acetaminophen) is the least-effective over-the-counter painkiller. At the time of writing, it beats placebo 62% of time. To put that in perspective, taking a sugar pill will cure your headache better than Tylenol 38% of the time. In other words, it’s not that effective.

Going back to the depression page, we can see that our favorite SSRI, Lexapro, beats placebo 58% of the time. In other words, the best SSRI that big pharma can muster has a probability of benefit 4 points worse than Tylenol—the least effective painkiller. Let me say that again: the worst headache cure you’ve tried is more powerful than every SSRI.

To clarify—Tylenol is more effective at treating pain than Lexapro is at treating depression. Don’t start taking Tylenol for your depression.

But have you taken Tylenol for a bad headache? It’s not great. Most of us are probably pretty familiar with the painkilling abilities of this class of drugs—good for mild headaches, bad for major surgery. But that level of mediocre efficacy is leaps and bounds higher than the efficacy of our most potent SSRIs.

This is where you say: “But depression is debilitating, and taking Lexapro is still better than doing nothing, right?”

The downsides

If SSRIs were free and safe, then they wouldn’t have to be that effective in order the justify taking them.

But—although these drugs are all off-patent—SSRIs are definitely not free. Neither are the psychiatrists who prescribe them.

And although SSRIs may have fewer side effects than their tricyclic forebears, they still have them. Looking at the side effects for Lexapro (the safest SSRI) we can see that the top side effect is insomnia, with a risk difference of 4%. That may not sound like a lot, but this means that for every 24 people that take Lexapro, one will have insomnia that otherwise wouldn’t have. And the top side effect for Prozacnausea/vomiting—will affect one more person for every three people.

Also, medical studies have a tough time looking at long-term safety. It’s expensive and impractical to follow 10,000 people for 30 years, so the side effects reported in most medical studies are those that surfaced during the very short duration of the trial. Hell, these drugs have only been around since 1987—how do we know they aren’t causing slow but serious long-term damage?

It sounds like fearmongering—but all I’m trying to say is that taking this kind of safety risk only makes sense if SSRIs really are wunderdrugs. But they aren’t, so why take the risk?

You’re wrong - SSRIs worked for me!

Good! Depression is awful, and if you’ve found something that works for you then by all means keep doing it.

The somewhat lackluster performance of SSRIs in medical studies doesn’t mean that individuals can’t see significant benefits. It just means that on average most people are helped only very slightly—if at all—by SSRIs.

It’s also worth noting that the control group in most depression studies does extremely well. A typical participant in a depression study can expect to shave 9 points off their HAM-D scorejust by taking a sugar pill. In contrast, the group taking the actual SSRI might improve by around 10 points, so the effect of the treatments is dwarfed by the placebo effect. This measley 1-point difference is the reason SSRIs don’t fare so well on GlacierMD, but a 10 point absolute difference is a meaningful boost to your happiness.

So if you’re reading this thinking that SSRIs worked for you, consider the possibility that eating a sugar pill might have also given you the same boost.

So what should I do?

SSRIs aren’t the only option for depressives. First, there are the older tricyclics and MAOIs. These have pretty gnarly side effects, but they also work a bit better. And there’s a newer class of drugs, SNRIs, which are prescribed to treat a variety of psychiatric disorders. Finally, there’s cognitive behavioral therapy, which has the big benefit of not being a synthetic drug.

Soon, GlacierMD will let you compare all of these treatment options. But at the time of writing we’re still in beta and rolling out our data slowly. Check back soon!