Melatonin: the right dose is 0.3 mg, and it can drag your mood down

During my holidays I took melatonin every evening, at the dose on the box, to go to bed earlier. I am someone who goes to bed late, and who likes it. Keeping a schedule is the hardest thing I ask of my sleep.

Sleep came. My mood sank week after week. Less drive, including on my own projects, the ones I do because I want to. I stopped. Within days I was back up, and not halfway.

This is an observation, not proof. I did not alternate with and without, I measured nothing but how I felt, and holidays are a time when many things change at once. I tell it because it is clear-cut, and because the literature, which I read afterwards, does not contradict it. It even says something else, more surprising: the dose that works is far smaller than the one being sold.

A signal hormone, not a sleeping pill

Melatonin is a hormone the brain secretes at night, in darkness. It does not cause sleep the way a sleeping pill does: it tells the body that it is night. It is a clock signal. Sleep follows, when everything else is in place.

The consequence is that the direct effect on sleep is modest. The reference meta-analysis, 19 placebo-controlled trials and 1,683 participants, finds seven minutes faster sleep onset, eight minutes more sleep, and slightly better sleep quality. The authors themselves call it modest. It is real, it does not fade over time, and it is small.

The other consequence is that swallowing the hormone puts far more of it in the blood than a night produces. Anses points out that a dose as low as 1 mg leads to supraphysiological blood concentrations, above 100 pg/mL, that is, more than the night produces. A natural night is a whisper. A pharmacy capsule is a shout.

0.3 mg does the job, 3 mg spills into the day

Zhdanova's trial, in 2001, asks the dose question in the cleanest way: 30 people over 50, half with poor sleep confirmed by actigraphy, each going through placebo and three doses, 0.1, 0.3 and 3 mg, one week each, with polysomnography on the last three nights.

The result fits in two sentences. The physiological dose, 0.3 mg, restored sleep efficiency, mostly in the middle of the night, and brought blood melatonin back to a normal level. The pharmacological dose, 3 mg, also improved sleep, but it lowered body temperature and left melatonin elevated in the blood into the daytime.

Ten times more hormone added nothing to sleep. It added hypothermia and a morning with melatonin in the blood, that is, a night signal sent in broad daylight. In people who slept well, no dose changed anything.

What is in the box

In France, food supplements containing less than 2 mg of melatonin per dose are allowed by administrative decision. Above that it is a medicine, Circadin, 2 mg prolonged release, reserved for short-term treatment of insomnia in people over 55. The threshold varies by country: in Belgium and Germany, a product providing 0.3 mg or more per day is considered a medicine. In Spain and Italy, the limit is 1 mg. In Denmark, melatonin is banned from supplements.

The two health claims authorised in Europe set the same scale: 1 mg before bedtime to "contribute to the reduction of time taken to fall asleep", 0.5 mg to "alleviate subjective feelings of jet lag". These are shelf doses, not trial doses. The 0.3 mg that restored sleep under electrodes sits below every one of these references.

And what is written on the label is not necessarily what is inside. An analysis of 31 commercial products found a content ranging from 83% below to 478% above the stated dose, with up to 465% variation between two lots of the same product. More than 71% of products fell outside a 10% margin around the label, and 26% contained serotonin, which has no business being there.

Timing matters more than dose

If melatonin is a clock signal, then the moment you send it decides what it does. That is what phase response curves measure: how much the clock shifts depending on the time of intake. Burgess and colleagues built one for 0.5 mg and compared it with the 3 mg curve, in 34 adults aged 18 to 42, in the laboratory.

To advance the clock, going to bed and waking up earlier, the maximum shift with 0.5 mg is obtained when it is taken two to four hours before the natural rise of the hormone, that is nine to eleven hours before the middle of sleep. For someone who sleeps from 1 am to 9 am, that is between 6 pm and 8 pm. Not at bedtime. And at their respective optimal times, 0.5 mg and 3 mg produce shifts of the same size.

That is exactly the use I did not make of it. I took the capsule at bedtime, at the hour when it shifts almost nothing, hoping it would turn me into an early riser. It put me to sleep a little, and it left me, on waking, with a night hormone in my blood.

There is also what I was asking of my body. Late types who force an early-riser schedule on themselves live what Wittmann and colleagues call social jetlag: among 501 volunteers, late chronotypes accumulate sleep debt on work days and make it up on free days, and the link between this mismatch and wellbeing is strongest before 25. Melatonin does not create that conflict, but I was taking it to win it by force.

Mood, what the data say

This is the least known point about melatonin, and it is documented at three levels. The Circadin leaflet lists, among uncommon effects, irritability, nervousness, restlessness, abnormal dreams, nightmares and anxiety, and among rare effects, mood altered, aggression, crying, depressed mood and depression.

The Anses opinion counts 90 reports of adverse effects between 2009 and May 2017 for supplements containing melatonin, 19 of them documented enough to be analysed: headaches, dizziness, drowsiness, tremors, nausea, nightmares, irritability. Of the more than 200 cases collected by pharmacovigilance since 1985, all forms combined, 24% are psychiatric, anxiety and depressive disorders first. The agency draws a precise recommendation: without medical advice, no melatonin for people with mood, behavioural or personality disorders, and occasional use for everyone.

The third level is a trial. Riemersma-van der Lek and colleagues followed 189 care home residents, 86 years old on average, on 2.5 mg of melatonin or placebo every evening for fifteen months on average. Melatonin shortened sleep onset by eight minutes and lengthened sleep by 27 minutes. It also worsened mood, less positive affect, more negative affect, and increased withdrawn behaviour. The authors conclude that they only recommend it combined with light exposure, precisely to counter that effect.

Two mechanisms make the story plausible without proving it. The first is Zhdanova's: a dose above the physiological range leaves a night signal in the blood at waking, and a night signal in broad daylight makes nobody enterprising. The second is the conflict with my rhythm: forcing an early-riser schedule on a night owl costs something, capsule or not. I cannot tell the two apart. I know my mood came back when the capsule went.

What I would do if I had to take it again

In practice: the smallest dose I can find, never every evening, only for a precise shift, taken in the afternoon if the goal is to shift and not to fall asleep. And if my mood moves, I stop without waiting to understand why.

What this article does not say

• That melatonin makes you depressed. A rare effect in a leaflet, a quarter of pharmacovigilance reports, a trial in very old people with dementia, and a personal observation. It is a signal to watch, not a rule.
• That 0.3 mg is enough for everyone. The reference trial covers people over 50 with poor sleep. In those who slept well, no dose changed anything, and the meta-analysis sees slightly more effect with high doses from one trial to the next.
• That my observation is a measurement. I did not alternate, I did not count, and holidays blur everything. I report a clear before, during and after, nothing more.
• That melatonin has no use. For jet lag or a clock to advance, taken at the right time and at a small dose, it is the documented tool.
• That it is a treatment for insomnia. Established insomnia needs proper care, and Anses asks for medical advice before any intake in people on treatment or at risk.

« A hormone is dosed like a signal: at the right hour, in a low voice. Not like a drug, by turning up the volume. »

What Helix does

This case brings together three things Helix tries to make readable on every line of a stack. A useful dose ten times lower than the dose sold. An effect that depends on timing more than on quantity. And a documented mood signal shown at the same level as the effect you are after, not buried in a precautions paragraph nobody reads.

The engine separates the clock-signal effect from the sleep-onset effect, each with its level of evidence, and shows the Anses precaution next to the dose. When an observation rests only on what someone felt during their holidays, it says so too.