The most-cited antidepressant study, re-examined

Hey Micronaut,

No product pitch today. Just a story about how we come to believe what we believe, and why reading the primary sources for yourself is one of the most useful habits you can build.

Most of us grew up with one tidy sentence about mental health. That low mood comes down to a “chemical imbalance,” a serotonin shortage you top back up with a pill. It became so common that surveys suggest 80% or more of the public now take it as established fact.

Here is the part that is less well known. In 2022, researchers at University College London reviewed decades of studies and reported no consistent evidence for that specific low-serotonin explanation. Their lead author was careful about it, noting it is always hard to prove a negative.

And to be fair, that review was debated. A group of other scientists pushed back, arguing its methods were flawed. That back and forth is science working as it should. The honest takeaway is narrow but real. The simple serotonin story was presented to the public as settled long before the evidence had settled anything.

Important caveat, and we mean it. None of this says antidepressants do not work. Whether a medicine helps is a separate question from why it helps, and large reviews still find these medicines help a great many people. Nothing here is medical advice. If you take antidepressants, please do not change anything based on an email. That is a conversation for you and your doctor.

With that said, here is some of what we have found worth reading over the years.

What gets published, and what does not

In 2008, a team writing in the New England Journal of Medicine tracked down every antidepressant trial registered with the FDA. They found 74 of them, then checked which had actually been published.

Almost a third were never published. The trials with positive results tended to make it into journals, the disappointing ones tended not to. Put the full FDA dataset back together and the published literature had overstated the apparent benefit by around 32%.

This is not unique to antidepressants. It is a known issue across medicine called publication bias, and it matters for one simple reason. The picture you can see is shaped by which results made it into print, not only by what the research found.

The number everyone quotes

Publication bias is about which trials you get to see. There is a second, subtler issue. How a single trial gets written up in the first place.

The most influential antidepressant study ever run is probably STAR*D. It was large (more than 4,000 people), publicly funded, and designed to answer a practical question. If the first medication does not work, and you keep trying others, how many people eventually get better? The headline answer, published in 2006, was a cumulative remission rate of about 67%. That figure became something close to clinical common sense, repeated in textbooks, guidelines and news articles for years.

In 2023, an independent team went back to the original patient-level data and reran the numbers using the study’s own pre-registered protocol. Their result came out closer to 35%, roughly half the famous figure, because of choices made along the way about who counted and how. The original investigators dispute the reanalysis and stand by their numbers. That argument is still live, and reasonable people are on both sides.

Notice what made that disagreement possible. The raw data had been preserved and shared, so outsiders could check the working. That is the system functioning, not failing. The point is not that one number is right and the other wrong. It is that a headline figure and the data beneath it are not always the same thing, and you are allowed to ask to see it.

What gets the research budget

Here is a quieter point that took us a while to appreciate, and it is less of a conspiracy than it sounds.

A company will spend hundreds of millions running a trial when there is a patent at the end of it. A monopoly, a sales force, years of protected revenue. That investment makes sense when you can own the result.

Now ask who funds the large, glossy campaigns for the things nobody can own. Sleep. Sunlight. Exercise. A walk outside. There is no patent on any of it, so there is rarely a marketing budget behind it. That does not make those things magic, and it does not make them worthless. It simply means the volume of attention an idea gets is not always a measure of how true or useful it is. It is often a measure of who stood to gain from telling you about it.

Where we stand, plainly

We are not here to tell you what to take, and we are not making any claim about treating or curing anything. That is genuinely not our place.

What we believe in is much simpler. You deserve the full picture rather than the highlight reel, and you are perfectly capable of reading the source material yourself. Not a passive recipient of a tidy sentence. A curious adult who follows the footnotes.

Don’t take our word for any of it. The primary sources are right here:

We are genuinely grateful for modern medicine. It saves lives every single day, and nobody here is arguing otherwise.

But “the only story you have ever heard” and “the whole story” are rarely the same thing. The good news is the full record is mostly public. You just have to go and read it.

Flow strong,

Microdose Pro

P.S. The 67% figure is still quoted in plenty of places, years after the reanalysis questioned it. That gap is the whole point. The number you remember and the number the data supports can drift apart quietly, and the only way to know is to read the source. Make of that what you will.