Sleep, Mood, and the Question I Get Asked Most

Of everything I have written on this site, the emails that arrive most often are about mental health. Some version of: I want to try this and I have a history of depression and my doctor looked uncomfortable when I brought it up.

So this is the month I take that on properly. It is also, conveniently, the month in which my own sleep changed enough to be worth reporting.

What the regulators actually found

Start with the thing everybody half-remembers.

In 2023 a safety signal emerged from spontaneous adverse-event reports — the databases where clinicians and patients file reports of things that happened while someone was on a drug. Some of those reports described suicidal thoughts in people taking GLP-1 receptor agonists. It made a great deal of news.

Spontaneous reports are a legitimate early-warning system and a terrible evidence base for causation. They have no denominator, no control group, and they are heavily influenced by media attention — once a possible association is reported, reporting of that association goes up regardless of whether it is real. This is a well-described phenomenon and it is why regulators investigate signals rather than acting on them directly.

Both major regulators investigated.

The FDA published a preliminary evaluation in January 2024 and stated that its review had not found evidence that these medicines cause suicidal thoughts or actions.

The European Medicines Agency’s safety committee concluded a fuller review in April 2024 and reached the same conclusion: the available evidence did not support a causal association.

I want to be careful about what that does and does not mean, because both the reassuring and the alarming versions get overstated.

It means the population-level signal was examined by people with access to far better data than any of us have, and was not confirmed. That is genuinely reassuring and it is the right basis for practice.

It does not mean nobody has ever had a mood change on one of these drugs. Regulatory reviews establish population-level causation. They do not describe every individual experience, and they cannot, because that is not what they are built to do.

“No causal association was found” and “this could never happen to you” are different sentences, and conflating them is how patients stop trusting doctors.

What I say in clinic

A history of depression is not a contraindication to GLP-1 treatment. I want to be unambiguous about that, because I keep meeting people who were effectively refused care over it.

What it is, is a reason to do three things properly:

Say it out loud, at the start. Not so it can be held against you. So there is a baseline, and so the person prescribing knows what to watch.

Agree in advance what you are both watching for, and what happens if it appears. A named threshold and a plan is worth a great deal more than a vague instruction to let someone know if you feel bad.

Do not stop anything else. A GLP-1 is not a substitute for psychiatric treatment and does not replace an antidepressant, and the two need coordinating rather than swapping.

There is also an honest limitation to acknowledge. People with significant psychiatric histories were frequently under-represented in the pivotal obesity trials. That is a real gap in the evidence. It is a reason for careful monitoring. It is not a reason to withhold an effective treatment from the group of people who — given how thoroughly obesity, stigma, and depression interact — often have the most to gain from it.

My own sleep, which I did not expect

Now the part I did not see coming.

I said at month six that I had stopped snoring and was waking differently. 8 months in, I have numbers rather than impressions: my average sleep has gone from about 6 hours and twelve minutes to about 7 hours and twenty-four minutes, and — more to the point — the wakings have largely stopped.

I did not think I had a sleep problem. I thought I had a schedule. It is now fairly clear to me in retrospect that I had mild sleep-disordered breathing, that I had normalised it over roughly a decade, and that it was quietly costing me an hour a night.

The evidence here is stronger than most people realise for one specific condition. SURMOUNT-OSA, published in the New England Journal of Medicine in 2024, was a randomised trial of tirzepatide in adults with obesity and moderate-to-severe obstructive sleep apnea, and it produced substantial reductions in the apnea-hypopnea index. That is a hard endpoint in a condition where the previous options were a machine most people abandon or surgery most people decline.

Outside diagnosed apnea the evidence is thinner and largely observational, and the plausible explanations are unglamorous: less weight on the airway, less reflux, less alcohol. I am not going to claim a direct hypnotic effect of semaglutide. I am going to say that an extra hour of sleep a night has probably done as much for my mood as anything else in this chronicle.

I have been tracking sleep alongside injection day in Zenday App since about month three, mostly out of curiosity, and the thing it surfaced was not what I expected either. It is not that sleep is worse after the shot. It is that on the two nights following an injection I sleep slightly more and wake slightly less, which I would never have picked up without the two data streams sitting next to each other. That is the difference between a tool that records and a tool that connects things — and it is why I have never gone back to keeping this in a notes app.

The thing I have not resolved

I promised at the start of this series that I would say where I am uncertain, so here it is.

Some people on these drugs describe a flatness. Not depression — they are usually careful to distinguish it — but a muting. Food stops being pleasurable, which is the point, and then occasionally something adjacent also gets quieter.

I have some of this. I wrote about the [grief of the third dinner party](/journey/the-first-4 weeks) in month one and I have written since about the wine I stopped wanting. 8 months on, I would say honestly: my emotional range is intact, my work is fine, my relationships are fine, and there is a small, specific dimming in the region where appetite and reward and celebration overlap.

Is that a side effect? Or is it what it feels like to no longer be using food the way I was using food, which is a change I chose? I do not know. I have thought about it a great deal and I cannot separate them, and I am suspicious of anyone who claims they can.

What I would say to you is: it is worth noticing, worth naming, and worth raising with your prescriber — particularly if it extends past food into things you used to enjoy. The failure mode is not having the feeling. The failure mode is deciding it is too vague to mention.

Where things stand

Down 17.6% at 8 months. Holding at 1.7 mg with no plan to escalate. Sleeping better than I have in a decade. Lifting three times a week and still adding weight to the bar.

Next month is nine, and it is the one where I stop describing what happened and start deciding what the rest of this looks like.

The research version of this, including what the regulators actually concluded about the suicidality reports, is in food noise, mood, and the quiet months. The mechanical causes of broken sleep worth ruling out first are in the sleep supplement rankings.