Pro

Melatonin is apparently safe⁸; from Wikipedia, a meta-analysis, and a review (respectively) of melatonin studies:

Melatonin appears to cause very few side-effects in the short term, up to three months, when healthy people take it at low doses. A systematic review^[63] in 2006_20ya looked specifically at efficacy and safety in two categories of melatonin usage: first, for sleep disturbances that are secondary to other diagnoses and, second, for sleep disorders such as jet lag and shift work that accompany sleep restriction.^[63] The study concluded that ‘There is no evidence that melatonin is effective in treating secondary sleep disorders or sleep disorders accompanying sleep restriction, such as jet lag and shift-work disorder. There is evidence that melatonin is safe with short term use.’^[63] A similar analysis^[64] by the same team a year earlier on the efficacy and safety of exogenous melatonin in the management of primary sleep disorders found that: ‘There is evidence to suggest that melatonin is safe with short-term use (3 months or less).’

The most commonly reported adverse effects of melatonin were nausea (incidence: ~1.5%), headache (incidence: ~7.8%), dizziness (incidence: 4.0%), and drowsiness (incidence: 20.33%); however, these effects were not [statistically-]significant compared to placebo. This result did not change by dose, the presence or absence of a sleep disorder, type of sleep disorder, duration of treatment, gender, age, formulation of melatonin, use of concurrent medication, study design, quality score, and allocation concealment score.⁹

A 2010 trial tested a delayed-release melatonin over 6 months and found minimal adverse effects and no tolerance or “addiction”, as well as benefits for both the young and old adults in the trial. A 2012 survey of 101 Australian doctors prescribing melatonin for children for as long as 4 years found little awareness of side-effects. Natural Standard 2011_15ya:

Based on available studies and clinical use, melatonin is generally regarded as safe in recommended doses (5 milligrams daily) for up to two years. Available trials report that overall adverse effects are not [statistically-]significantly more common with melatonin than placebo. However, case reports raise concerns about risks of blood clotting abnormalities (particularly in patients taking warfarin), increased risk of seizure, and disorientation with overdose.

Con

The FDA does not regulate melatonin, as a supplement, but it has declined to give it GRAS-status - the highest level of safety which allows it to be added to regular food; its warning letters are good sources if we are looking for evidence against melatonin’s safety. One such letter, sent to the makers of the soda Drank which includes melatonin, cites 3 pieces of evidence:

1. increased cancer in female lab mice (but an increased lifespan! And human studies have not found any increased cancer¹⁰)

2. retinal damage in 2 lab mice variants (human clinical study found a protective effect¹¹)

3. a summary of a Natural Standard review monograph mentioning human research finding weak potential for decreased blood pressure¹², hyperglycemia¹³, and increased atherosclerotic plaque, and its general concern about the use of a hormone

Health & Performance

One might object that they do not wish to tamper with their natural sleep, even if melatonin is a normally-secreted hormone.

Sad to say, I would point out to such readers that they are already profoundly tampering with their natural sleep cycle, and indeed, all of Western civilization is tampering with it; most of my readers do not even sleep multiple times during the day, as ‘Nature intends’ and as humans have usually slept through history¹⁴¹⁵, but rather in a single 7–9 hour long block. Sleep debt causes negative changes similar to aging, and sleep deprivation, both acute and chronic, damage mental performance¹⁶, and chronic does so cumulatively over the deprivation period (conversely, sleep improves performance); worse, one may not be able to simply sleep in on the weekend, both because it’s too short a time period to repay sleep debt (Dement2005 noting it may take a full month to repay the large sleep debt incurred by regular shortfalls; jet lag in well-rested mice can take up to 28 days to mentally recover from) and because in practice one doesn’t sleep in very much - the Zeo database of sleep records only an extra 23 minutes a day on the weekend. Subjects in chronic sleep deprivation studies subject to sleep debt were subjectively unable to notice their performance declines (NYT); and since “short sleepers” seem little different from everyone else, this suggests that many people who believe they are short sleepers are just hurting themselves. Finally, there are multiple lines of research suggesting chronic sleep deprivation is prevalent among young adults (including historical comparisons^{17 18}). It is striking that unemployed adults sleep a full hour longer than the employed¹⁹, and that when normal adults are placed in settings without artificial light like camping or without any time indicators, they sleep longer than before - exactly as if they were sleep deprived.²⁰ Additional sleep is linked with athletic²¹ and artistic success²², that even a hour’s tampering with melatonin secretion leads to measurable performance changes, melatonin secretion declines markedly with age, and so on and so on. One could well ask with Alexander Pushkin, “Why am I to feel this pain?”

One sometimes sees people invoke a sort of evolutionary heuristic argument against melatonin; from the WSJ:

As soon as our own son started asking for the “magic” pill, my husband and I stopped giving it to him. Two years later, he still suffers from occasional insomnia. But whenever I get frustrated, I think back to what Dr. Ditchek told me. “For thousands of years our children have been falling asleep without the need for pills,” he said. “Giving your healthy child a pill to fall sleep is sending him the wrong message - that he needs a pill to do what should come naturally.”

However, we are not in a natural situation, but a profoundly unnatural one. For children in particular, Matricciani et al 2012 demonstrated a drastic fall in their sleep time over the past century, indicating that they are not in any state of nature. (Or consider the dramatic spike in myopia in many countries over the 20^th century, which is linked to lack of exposure to normal bright light causing insufficient ocular growth.) And aside from all the previous cited evidence that one’s sleep is already being heavily affected by modern lifestyles (deprivation, monophasic block sleep, etc.), there’s a specific argument to be made that the health of heavy electronics-users (such as myself and my readers) will improve through melatonin supplementation: modern artificial electrical lighting during the evening is systematically tampering with sleep and melatonin secretion in particular.

Computer use correlates with less sleep and negative events in teenagers, who are forced into being night-owls by puberty; melatonin helps them sleep more normally. Indoor lighting affects the timing of circadian shifts and ordinary room lighting at night, before bedtime, has been shown to improve alertness - but doing so by cutting melatonin secretion, by as much as half²³ with computer tablets such as the iPad cutting it by a fifth in one experiment & similarly with monitors, and afternoon/evening artificial light damages subjective alertness & dual n-back performance compared to natural light²⁴. Light affects melatonin secretion even when your eyes are shut. Blue light is the main culprit²⁵ (white is still bad, though). An additional list of relevant studies on light & melatonin & sleep is under f.lux’s summary page “NIH-Funded research”.

There are counter-measures to reduce one’s exposure to blue light like light-filtering glasses²⁶ or programs such as Redshift/f.lux which lower the color temperature (reducing the blueness & increasing redness) of one’s monitor. They are a good idea²⁷. (People who, for various reasons, abandon electrical lighting remark on the simultaneous shifts in their sleep - I noticed this when watching The 1900_126ya House & Frontier House, and occasionally run into anecdotal reports like J.D. Moyer.) But they aren’t perfect solutions and only reduce the damage; and if Robin Hanson is right that future bright LED lights will be cheaper and more popular, we can expect that much more problems with melatonin deficiency/suppression (Czeisler2013). And what are those problems?

Shift workers have been linked to a great many health problems (such as various cancers or mental delays; see shift work sleep disorder), and the proposed causal mechanism is usually a melatonin deficiency (and melatonin supplementation one of the treatments). (They have also been linked to many accidents and near-accidents.)

An additional advantage is the widely-reported increase in dreams or vividness of dreams; but some report this leads to increased nightmares as well, so this is not a major argument for using melatonin supplements. (Personally, I enjoy the greater number of vivid dreams, and even kept a dream diary for a while, until it became too much work.)

Tempus Fugit

My rule of thumb is melatonin subtracts an hour. (I originally guessed at this value, but my Zeo sleep recordings seem to suggest the value is more like 50 minutes.) That is: if one slept for 7 hours, one awakes as refreshed as if one had slept for 8 hours etc. From comparing with others, I think I benefit more than around two-thirds of people.

There are few to no side-effects to melatonin use in adults (there is uncertainty about the risks & benefits in children & adolescents²⁸), and it is not addictive or habit-forming like caffeine is. The usual dose for a night is 0.5-3 mg and I take 1.5mg²⁹; my dose is highly likely to be too high. High doses may well be responsible for why some people try melatonin and report that it does nothing or hurts them, since in one study, the best dose for old people was 10x smaller (0.1mg or 0.3mg) and for one blind person, 0.5mg^{30 31 32}. Zhdanova et al 1996 found 0.3mg & 1.0mg to affect sleep onset similarly. A study of delayed-release melatonin found with their high dose of 4mg (but not 0.4mg) elevated melatonin levels 10 hours after bedtime (Gooneratne et al 2011) - potentially interfering with waking time. Forsling et al 1999 used doses of 0.05, 0.5 & 5.0mg, and measured blood-levels of various hormones and chemicals; 0.5 & 5.0mg were similar except in inhibiting & stimulating (respectively) oxytocin and vasopressin. Tested supplements have the claimed amounts of melatonin, but their recommended doses can be terrible ideas and definitely should be ignored if they advise more than 5mg!³³ Online anecdotes frequently mention that doses of 0.5mg or less worked better than >1mg. (Experimentation may be necessary; people can be very different.)

So that’s what it does: it lets one cut 1 hour of sleep. (There are other benefits, such as enforcing a bedtime - invaluable for young people - but their values are indeterminate.)

Let’s work out the cost of regularly using melatonin. Conservatively, a bottle of melatonin pills will cost about 6 USD for 150 pills. Per milligram, the 3 mg pills are the least costly, and are easily convertible into lower dosages. Regardless, one will get a minimum of 150 doses per $6. 600⁄150 = 4¢ per pill. 1 night requires 1 dose, so each night costs 4¢. We need to consider how much we value the effort of remembering and taking the pill, though. Let us be conservative and price it as high as 10¢. So we spend 14¢ per night³⁴. The reward is that we gain an hour.

Now how do we value an hour of our time? One could make an intuitive guess, or consider one’s current salary and things like that, but in economics 101, it’s common to just set the value of an hour at minimum wage. This will hopefully not be too erroneous, since over our lifetime, the days where that extra hour was really valuable will balance out the days where the hour was worthless.

More concretely, the minimum wage is the floor for one’s opportunity cost; even if the hour’s opportunity you forgo by sleeping more wasn’t worth a million dollars, you could still have worked an unskilled minimum wage job and gotten the minimum wage.³⁵

The federal minimum wage as of 2008 is $6.55 an hour³⁶. So let’s say that is the reward.

ROI

See also value of information calculations for melatonin self-experimentation.

In other words, by investing 14¢, we realize a net of 655¢, for a profit of 641¢. This represents a return on our investment of approximately 4678% ($\frac{655}{14}\cdot 100$). Suppose we use only half a pill a night? Then our number of doses doubles, our pill price per night halves (though our 10¢ of effort remains constant), and our return becomes 5458% ($\frac{655}{12}\cdot 100$). My local store sometimes runs a 2-for-1 sale on melatonin. One could buy only at these sales - the low consumption rate (1 a night, or less) means you must buy less than annually. Thus, the pill price halves again to 1¢ per night, and now the return is 5959% ($\frac{655}{11}\cdot 100$).

This assumes melatonin delivers its benefit with certainty, or probability 1. Since, for obvious reasons, there have been few studies directly studying how melatonin improves sleep in healthy adults, we are forced to estimate based on anecdotes; as an example, we can look at >100 anecdotes about melatonin usage generated in response to this essay advocating melatonin: 64 positive/25 negative/12 unknown, or 60% positive. That the anecdotes aren’t 100% positive but only 60% suggests that the collection isn’t too badly skewed by various selection biases, since typically the worry is that only the deluded fans are posting about how substance X transformed their life. By expected value, our expected nightly profit is still positive: $641\cdot 0.6=384$¢.

Or one can reason in a “value of information” fashion: after using the first jar of melatonin, you will have a very good idea whether it makes you feel better and then you can either continue or stop. If you have a 40% chance of losing the $10 that jar cost you and a 60% chance of profiting, say, a dollar a night for a year (let’s be conservative), we can easily estimate the value of experimenting with that first jar: loss plus gain, or $0.4\cdot -10+0.6\cdot (1\cdot 365.25)$, or… $215. To put it another way, melatonin people benefit a lot from using it and non-melatonin people are out only the small cost of the experiment, so we would have to be very pessimistic before one quick experiment is not worth doing. And we have no reason to be >97% pessimistic³⁷.

Absolute Gains

Let’s look at another perspective. Those are impressive percentages, but they’re on a low base so perhaps it’s not worthwhile even with 6000% returns. What are the absolute gains one would realize over the course of a year? Well, to calculate: ($(365\cdot 6.55)-(365\cdot 0.11)$), or to simplify: ($365\cdot (6.55-0.11)$) = $2,350.60. We would have to assume our calculations are 3 or 4 orders of magnitude off before melatonin stops being a good deal; the cheapness of use dominates the calculation.

One year of melatonin amply repays the original costs of learning about and experimenting with it. (And the long-run benefits are substantial.³⁸)

Depression

Low melatonin levels and chronobiological disturbances have been frequently correlated with various forms of depression; eg.

• Mendlewicz et al 1979 (review of existing research linking depression with abnormal melatonin secretion & low levels)

• Claustrat et al 1984

• Nair et al 1984

• Brown et al 1985 & Frazer et al 1986 (2 studies: depressed subjects had reduced melatonin secretion. Frazer’s abstract notes that “These results are similar to those found recently by several other groups of investigators.” Indeed.)

• Kennedy et al 1989 (controls vs eating disorder victims; only eating disorder plus depression correlated with lower melatonin)

• Shafii et al 1996 (“Post hoc analysis showed a [statistically-]significantly higher melatonin profile in depressed subjects without psychosis (n = 15) than in depressed subjects with psychosis (n = 7) or in the controls.”)

• Voderholzer et al 1997 (a negative result - studying 9 young depressed former alcoholics, Voderholzer did not find any melatonin differences)

• Crasson et al 2004 (found delayed melatonin secretion eg. less in evening & more in morning, but still similar total secretions)

• Hasler et al 2010

• Dallaspezia & Benedetti2011:

  Alteration of the sleep-wake cycle and of the sleep structure are core symptoms of a major depressive episode, and occur both in course of bipolar disorder and of major depressive disorder. Many other circadian rhythms, such as the daily profiles of body temperature, cortisol, thyrotropin, prolactin, growth hormone, melatonin and excretion of various metabolites in the urine, are disrupted in depressed patients, both unipolar and bipolar individuals. These disrupted rhythms seem to return to normality with patient recovery. Research on circadian rhythms and sleep have led to the definition of nonpharmacological therapies of mood disorder that can be used in everyday practice…

• Bedrosian et al 2012 (Siberian hamsters acted depressed when sleep was disturbed by dim light)

Improvement or lack of harm:

• Fainstein et al 1997

• Jean-Louis et al 1998

• Lewy et al 1998 -(placebo-controlled trial of 5 patients)

• Dalton, et al 2000 (sleep improvements in the depressed; statistically-significant depression improvement also occurred (pg4), but authors don’t regard the effect size as large enough to be important. No mention of worsening.)

• Bellipanni et al 2001 (they note in passing, in a study of melatonin’s effects on hormones in women, that morning mood improved in the treated)

• Pacchierotti et al 2001 (review of previous trials; concluded the relationship between melatonin & mood is unclear and the evidence was not solid enough to outright recommend it)

• Danilenko & Putilov2005 (effect of total sleep deprivation treatment followed by melatonin supplementation on depressives; no change in mood due to the melatonin)

• Lewy et al 2006; 81 subjects; small improvements in depression, although not designed to test that. The results are very interesting inasmuch as it suggests SAD is made of two populations whose circadian rhythm are differently out of sync and who the theory predicts would benefit from melatonin supplementation at different times of day; the authors note that the subject whose depression got worse was also receiving melatonin at the ‘wrong’ time of day.

• Garzón et al 2009 (sleep improvements in the elderly using melatonin; abstract mentions improvements in depression questionnaire scores but not whether any had actually been diagnosed depressed)

• Bright light therapy, which affects melatonin, has been employed with various depressions:

  • SAD: Pail et al 2011

  • premenstrual dysphoric disorder: Parry et al 2011; see also Bellipanni et al 2005 (“At present we assert that the six-month treatment with MEL produced a remarkable and highly [statistically-]significant improvement of thyroid function, positive changes of gonadotropins towards more juvenile levels, and abrogation of menopause-related depression.”)

• Serfaty et al 2010; sleep improvement with 6mg doses, with non-statistically-significant improvement in the depressed subjects’ mood.

• Fava et al 2012: melatonin+buspirone improved depression scores more than buspirone or placebo

• Del Fabbro et al 2013

• Hansen et al 2014 meta-analyzes melatonin for depression, finding only weak evidence for benefits but also no evidence of harm.

Because of melatonin’s safety but weak anti-depressant effects, attempts have been made to improve on it. The best covered seems to be agomelatine which binds to melatonin receptors and in the clinical trials, has the desired anti-depressant effects. Hickie & Rogers2011 reviews “the associations between disrupted chronobiology and major depression” and regards agomelatine favorably, as does Quera Salva et al 2011; although Howland2011 reviewed 13 placebo-controlled trials of agomelatine, concluding that the benefit may not be large (but neither mentions agomelatine being pro-depressant)

The one study I was able to find showing any negative effects was Carman et al 1976, a double-blind study of 6 depressed patients; melatonin was administered at various times during the day at daily doses between 250–1600mg (>83x the typical doses for sleeping!) It does not seem to have been replicated. Lewy et al 2006 would seem to suggest that if there is anything to it, the results may be caused by the specific timing of melatonin supplementation; Serfaty et al 2010 simply calls it “a methodologically unsound trial”. If nothing else, melatonin may help with the depressed’s sleep quality (Dolberg et al 1998, Dalton2000, Garzón et al 2009).