Skip to main content

Melatonin

Melatonin improves sleep, & sleep is valuable

I discuss melatonin’s effects on sleep & its safety with research up to 2015; I segue into the general benefits of sleep and the severely disrupted sleep of the modern Western world, the cost of melatonin use and the benefit (eg. enforcing regular bedtimes), followed by a basic cost-benefit analysis of melatonin concluding that the net profit is large enough to be worth giving it a try barring unusual conditions or very pessimistic safety estimates.

One of the problems with ‘productivity’ gimmicks is assessing their worth. Many of them claim immeasurable gains1, dubious gains2, or clear gains which mayn’t be worth the overhead3. The benefits of others are perfectly measurable, like adding keybindings for repetitive tasks; just time the manual way versus the shortcut, and multiply it by the usage. But then the cost is hard to measure. How to assess the price of creating, learning, and rendering habitual a custom shortcut? There are scads of useful bindings in Emacs that I have tried to learn, only to forget them or remember them after I needed them. My favorite ‘productivity tool’ is one that suffers from none of these problems, and has refreshingly clear-cut costs and benefits. Even though it eats up a third to a half of your life, sleep is neglected because people assume it is too hard to quantify or is too mysterious; but since it takes such a large chunk of time, even doubtful or unreliable improvements are well worth your while.

Use

Melatonin (Examine.com; FDA adverse events) is a hormone secreted by the pineal gland. Its primary purpose is regulating the sleep cycle; its abatement permits waking. Odds are your local greengrocer carries it. It’s often used by people suffering from insomnia or to combat conditions such as jet lag.4 It has other miscellaneous uses like combating other chronobiological issues and linked problems like depression (it’s been claimed melatonin use may cause or exacerbate depression, but the research doesn’t seem to bear this out).

But it’s also useful for adults5 with perfectly normal sleep-cycles6. I am unsure whether its effect is to put one to sleep faster, or to do that and also deepen sleep, but melatonin does it well. The 2005 meta-analysis Brzezinski et al concluded that, over all (mostly healthy) adult participants, melatonin improved on placebo to the extent that it

[statistically-]significantly reduced sleep onset latency by 4.0 min (95% CI 2.5-5.4) [The normal limits for latency to sleep are considered to be 15-20 min.]; increased sleep efficiency by 2.2% (95% CI 0.2-4.2) [The normal sleep efficiency is about 90-95%.], and increased total sleep duration by 12.8 min (95% CI 2.9-22.8). Since 15 of the 17 studies enrolled healthy subjects or people with no relevant medical condition other than insomnia, the analysis was also done including only these 15 studies. The sleep onset results were changed to 3.9 min (95% CI 2.5-5.4); sleep efficiency increased to 3.1% (95% CI 0.7-5.5); sleep duration increased to 13.7 min (95% CI 3.1-24.3).

The meta-analyses or reviews Buscemi et al 2006 & Braam et al 2009 & Keegan et al 2013 & Ferracioli-Oda et al 2013 turn up weaker or similar results in other populations, the last mentioning (importantly for dependency concerns) that the sleep quality benefits did not seem to be moderated by “trial duration and melatonin dose.” The effects may be stronger in the old or elderly7; perhaps due to their lowered secretion of melatonin, and Braam et al 2009 interestingly speculates that effects of melatonin are understated due to most data coming from subjects sleeping in an alien sleep laboratory environment:

Visual inspection of the standard forest plots, as presented in the three meta-analyses in individuals without intellectual disability, suggests that the change in sleep latency in studies in which measurements were performed in a sleep laboratory using polysomnography is smaller compared with studies that were performed under home conditions. Re-analyzing the data presented in these studies shows that the mean change in sleep latency in the studies using polysomnography is 10.1 minutes, whereas change in sleep latency in studies performed under home conditions is 16.8 minutes. The inclusion of a substantial number of studies using polysomnography in the meta-analyses in individuals without intellectual disability may have contributed to the smaller decrease in sleep latency compared with our meta-analysis, in which all studies were performed under home conditions.

Safety

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 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:

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

Benefits

Health & Performance

Luke: “Is the dark side stronger?”
Yoda: “No, no, no! Quicker, easier, more seductive…”14

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 history1516, 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 performance17, 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 comparisons18 19). It is striking that unemployed adults sleep a full hour longer than the employed20, 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.21 Additional sleep is linked with athletic22 and artistic success23, 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 20th 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 half24 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 light25. Light affects melatonin secretion even when your eyes are shut. Blue light is the main culprit26 (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 glasses27 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 idea28. (People who, for various reasons, abandon electrical lighting remark on the simultaneous shifts in their sleep - I noticed this when watching The 1900 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.)

Summary

My basic argument is:

  • it works

If that’s not enough, an extended argument goes:

  1. electrical lighting and electrical devices in particular emit blue light

  2. blue light inhibits increase of melatonin levels

  3. melatonin supplementation increases melatonin levels

  4. increase in melatonin levels is necessary to sleep well

  5. either eliminating electricity or melatonin supplementation will cause melatonin levels to rise

  6. one will not eliminate electricity

  7. one can use melatonin supplementation

  8. it works

The existing phenomenon of alarm clocks etc is partially caused by this effect. People live normal modern lives with countless devices, and stay up too late or sleep poorly. To avoid one of the consequences, oversleeping, they use alarm clocks. Of course, either lack of melatonin or lack of sufficient scheduled time can torpedo one’s overall quality. Melatonin is necessary, but not sufficient, as is sufficient hours in bed.

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 & adolescents29), 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 1.5mg30; 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, 0.5mg31 32 33. 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!34 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 night35. 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.36

The federal minimum wage as of 2008 is $6.55 an hour37. 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% (). 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% (). 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% ().

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: ¢.

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 , 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% pessimistic38.

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: (), or to simplify: () = $2350.60. 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.39)

Self-Discipline

Speaking from personal experience, I know that one of the obstacles to sleeping well is going to sleep at all. Even though one knows that one ought to go to bed on time, and that not doing so will cause problems, it’s hard to actually do it. One wants to finish the book, chat with friends, play a game, etc. It is even more difficult when one doesn’t feel tired. For me, I had a chronic akrasia problem with going to sleep; in college, it was bad enough that I would on occasion stay up to 4 AM for no reason at all!

This dilemma is far from unique. It is called hyperbolic discounting; humans can make the rational decision when at a great distance from a choice, but the closer they come, the more warped their decisions are. Procrastination is often thanks to hyperbolic discounting - ‘hard work pays off tomorrow but procrastination pays off now’, and never mind that tomorrow always comes. Similarly, addicts want to be free of their addiction, but their want for a drink right now overwhelms their lifetime desire to not have drinks. (For more on the topic, see “Akrasia, hyperbolic discounting, and picoeconomics” and “Applied Picoeconomics”.)

How do we deal with this? The classic mechanism is avoiding the choice entirely. An addict can avoid bars or liquor stores, but no one can avoid sleep. Failing to avoid the choice entirely, one raises the cost of ‘procrastination’ - make the addict pay $100 for every drink they take. Even apparently trivial cost increases like someone watching our computer desktop through VNC, or working at a coffee shop40 can make a real difference with procrastination. We need to raise the cost, then, and somehow change the incentives to make us want to sleep.

We can do this simply by waiting until the need to sleep is so strong we can no longer resist; and in practice, many (especially college students) do just this. But few of us have the luxury of the bizarre schedule this entails. We could try some sort of monetary fine for not going to bed by midnight, but enforcement is difficult and if you’re a college student, you may not be able to afford a vow painful enough to deter you.

Melatonin allows us a different way of raising the cost, a physiological & self-enforcing way. Half an hour before we plan to go to sleep, we take a pill. The procrastinating effect will not work - half an hour is so far away that our decision-making process & willpower are undistorted and can make the right decision (viz. following the schedule). When the half-hour is up, the melatonin has begun to make us sleepy. Staying awake ceases to be free, to be the default option; now it is costly to fight the melatonin and remain awake. The choice of sleep may now prevail over the hyperbolically-distorted choice of video games. The long time-lag and the extreme ease of taking the melatonin makes it harder to succumb to a kind of ‘meta-akrasia41 where you come up with a good trick or solution to make yourself do whatever it is you need to do - and then you start procrastinating/suffering from akrasia about the trick!

And going to sleep when you need to go to sleep is, in the long-run, a very valuable thing in its own right.

Excuses, Excuses…

Melatonin is a clear-cut Good Thing. The gains I have laid out are large enough I consider it irrational for someone not to use it, unless:

  1. You’ve never heard of it, or seen an analysis of the possible benefits.

    • You, dear reader, are no longer allowed this excuse.

  2. You are so poor that 6 dollars every 150 or 300 days is a crippling expense.

    • But then how come you have the free time to read this?

  3. Melatonin supplements just doesn’t work on you, period.

    • Possible (I have read multiple anecdotes that melatonin did nothing to help), but it’s not that common. Melatonin isn’t some mental trick - it’s a fundamental fact of mammalian biology. I’ve run into more people who have tried melatonin and had it help than found it completely ineffective, and I wonder if some of the non-responders varied their doses to test the lower ranges like 0.1mg.

  4. They work, but not well enough.

    • This implies that melatonin saves you only a few minutes or seconds, else the gain would be smaller but still be positive. (This too strikes me as unlikely.)

  5. You value a simpler, less complicated life that much.

    • Taking a pill at night, and buying some pills once every year or two stresses you out?

  6. You value an hour at less than 11¢, so melatonin is not profitable.

    • Please contact me. I would like to hire you at the princely rate of a quarter an hour to do drudge-work on Wikipedia.42

  7. You are a child or adolescent (see the footnote previously)

    • Legitimate; at the very least, the cost-benefit analysis becomes uncertain enough that there’s no obvious right answer.

  8. Melatonin is prescription-only or banned.

    • You have my sympathies.

Competition

A parting thought: millions of Americans (estimated at around a third or more of adults) regularly take multivitamin supplements. These supplements are unlikely to help a balanced diet, there is little evidence they do, and there are studies which have indicated actual harm from the consumption of multivitamins43. Multivitamins are noticeably more expensive than melatonin pills, and are certainly harder to swallow. And this is to say nothing of supplements which are actually dangerous, like ephedra. Do you take a multivitamin, but not melatonin? If so, how can you justify this?

See Also

  • Modafinil - if you want to go even further down the sleep-modification rabbit hole of trading money for time

Appendix

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:

  • 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, Dalton 2000, Garzón et al 2009).


  1. mindmaps↩︎

  2. ‘image-streaming’↩︎

  3. schedulers, personal organizers, Getting Things Done↩︎

  4. For which it has a good scientific track record; eg. Rikkert & Rigaud2001, Zhdanova2001, Cardinali et al 2002, Herxheimer & Petrie2002 or Buscemi et al 2004; this is the only use of melatonin Natural Standard gave an A rating, ‘Strong scientific evidence for this use’.↩︎

  5. Abnormal children benefit, although this may be as much due to the fact that children usually sleep very well to begin with and who would bother to study them? The Natural Standard:

    There are multiple trials investigating melatonin use in children with various neuro-psychiatric disorders, including mental retardation, autism, psychiatric disorders, visual impairment, or epilepsy. Studies have demonstrated reduced time to fall asleep (sleep latency) and increased sleep duration. Well-designed controlled trials in select patient populations are needed before a stronger or more specific recommendation can be made…Based on human study, melatonin may be beneficial in children with insomnia. More well-designed studies are needed before a conclusion can be made.

    ↩︎
  6. Natural Standard 2011:

    Multiple human studies have measured the effects of melatonin supplements on sleep in healthy individuals. A wide range of doses has been used often taken by mouth 30 to 60 minutes prior to sleep time. Most trials have been small, brief in duration, and have not been rigorously designed or reported. However, the weight of scientific evidence does suggest that melatonin decreases the time it takes to fall asleep (“sleep latency”), increases the feeling of “sleepiness”, and may increase the duration of sleep. Better research is needed in this area.

    ↩︎
  7. “Treatment efficacy of exogenous melatonin for insomnia in older adults: a meta-analysis”, Brault et al 2012; but it seems to have only been published as a poster/abstract, with a few more details in media coverage.↩︎

  8. As one would expect if the null hypothesis really is true, there are occasional cases where melatonin has fewer side-effects than placebo. Lemoine et al 2012 found that slow-release melatonin improved sleep quality & latency, but also that “The rate of adverse events normalized per 100 patient-weeks was lower for PRM [melatonin] (3.66) than for placebo (8.53).”↩︎

  9. “Melatonin for Treatment of Sleep Disorders: Summary”; this meta-analysis also suggests that melatonin works on some metrics, but has little or no effect on other measurements, which is a contrast to my own generally sanguine belief.↩︎

  10. Natural Standard 2011:

    There are several early-phase and controlled human trials of melatonin in patients with various advanced stage malignancies, including brain, breast, colorectal, gastric, liver, lung, pancreatic, and testicular cancer, as well as lymphoma, melanoma, renal cell carcinoma, and soft-tissue sarcoma. Currently, no clear conclusion can be drawn in this area. There is not enough definitive scientific evidence to discern if melatonin is beneficial against any type of cancer, whether it increases (or decreases) the effectiveness of other cancer therapies, or if it safely reduces chemotherapy side effects.

    ↩︎
  11. From the 2011 Natural Standard ‘Bottom Line Monograph’, ‘Evidence’ section:

    Melatonin may exert antioxidant effects which may contribute to its beneficial effects on the eyes. According to clinical study, melatonin may play a role in protecting the retina to delay macular degeneration. Well-designed clinical trials are needed before a conclusion can be made….It has been theorized that high doses of melatonin may increase intraocular pressure and the risk of glaucoma, age-related maculopathy and myopia, or retinal damage. However, there is preliminary evidence that melatonin may actually decrease intraocular pressure in the eye and delay macular degeneration, and it has been suggested as a possible therapy for glaucoma. Additional study is necessary in this area. Patients with glaucoma taking melatonin should be monitored by a healthcare professional.

    ↩︎
  12. NS 2011:

    Several controlled studies in patients with high blood pressure report small reductions blood pressure when taking melatonin by mouth (orally) or inhaled through the nose (intranasally). Specifically, nocturnal high blood pressure may improve with melatonin use. Better-designed research is necessary before a firm conclusion can be reached.

    ↩︎
  13. NS 2011:

    Elevated blood sugar levels (hyperglycemia) have been reported in patients with type 1 diabetes (insulin-dependent diabetes), and low doses of melatonin have reduced glucose tolerance and insulin sensitivity. Caution is advised in patients with diabetes or hypoglycemia, and in those taking drugs, herbs, or supplements that affect blood sugar. Serum glucose levels may need to be monitored by a healthcare provider, and medication adjustments may be necessary.

    ↩︎
  14. The Empire Strikes Back↩︎

  15. Bimodal or biphasic sleep was universal until recently, especially in England; see A. Roger Ekirch’s “Sleep We Have Lost: Pre-industrial Slumber in the British Isles” and his 2005 At Day’s Close: Night in Times Past. Anthropology agrees; biphasic sleep is positively pedestrian↩︎

  16. from Claudio Stampi’s book on polyphasic sleep (Why We Nap pg 8–9):

    It is worth mentioning that anthropological studies conducted in tribes active at night show that human sleep can be highly polyphasic in certain cultures. Although they have different cultures and ways of life, both the Temiars of Indonesia and the Ibans of Sarawak have similar polyphasic sleep-wake behaviors (Petre-Quadens, 1983). Their average nocturnal sleep episode duration ranges between 4 and 6 hr, and nighttime activities (fishing, cooking, watching over the fire, rituals) at any one time involve approximately 25% of the adult members. Daytime napping is very common in both tribes: at almost any time of day, about 10% of the adult members are asleep. Whatever the cause of these polyphasic sleep patterns,whether the expression of an inborn ultradian rest-activity tendency or other factors, such populations exhibit extremely flexible and fragmentary sleep-wake cycles. The minimal contact with modern civilization could be one of the reasons for the preservation of this possibly ancestral sleep pattern.

    ↩︎
  17. With potentially fatal effect. From Coren1998:

    In fact, our societal sleep debt is so great that simply losing one additional hour of sleep due to the spring shift to daylight savings time can increase traffic accident rates by 7% (Coren, 1996b) and death rates due to all accidents by 6.5% (Coren, 1996c).

    ↩︎
  18. See Webb WB, Agnew HW (1975), “Are we chronically sleep deprived?” Bull. the Psychonomic Soc. 6:47-48.↩︎

  19. “Never Enough Sleep: A Brief History of Sleep Recommendations for Children” reviews literature from 2009 back to 1897, finding that sleep fell by around an hour on average (with very wide variation at both dates); they also find that sleep recommendations are regularly anchored towards recommending more sleep than children were actually getting - which the authors apparently take as evidence for the recommendations being false but I take as perfectly consistent with the picture I illustrate here of an ever more short-sleeping modern society and explicable by the fact that the average of the 1890s sleep recommendations—10.8 hours—would be hysterically laughed at by modern parents (hence the average recommendation in the 2000s of closer to 9.8 hours).↩︎

  20. “A coming study by Mr. Krueger, using historical data on time use between 1991 and 2006, finds that unemployed Americans tend to sleep an hour longer than the employed.” “Leisure Trumps Learning in Time-Use Survey: Americans Opt for TV in Spare Hours, Not Workouts or Classes, Poll Finds”, 2011-06-23 Wall Street Journal↩︎

  21. “Sleep Deprivation, Psychosis and Mental Efficiency”, Coren1998, Psychiatric Times

    Confirmation of these natural sleep durations comes from Palinkas et al 1995. These researchers spent a summer above the arctic circle where there is continuous light 24 hours a day. All watches, clocks and other timekeeping devices were removed, and only the station’s computers tracked the times that the team went to sleep and awakened. Individual researchers did their work, and chose when to sleep or wake according to their “body time.” At the end of the experiment, they found that their overall average sleep daily time was 10.3 hours. Every member of the team showed an increase in sleep time, with the shortest logging in at 8.8 hours a day, and the longest almost 12 hours a day. This study, like many others, seems to suggest that our biological need for sleep might be closer to the 10 hours per day that is typical of monkeys and apes living in the wild, than the 7 to 7.5 hours typical of humans in today’s high-tech, clock-driven lifestyle.

    On the other hand, Antarctica “has a reputation for providing the worst quality of sleep on Planet Earth” during the winter (in this case, partially due to altitude).↩︎

  22. “Peak Performance: Why Records Fall”, Goleman1994, The New York Times:

    “When we train Olympic weight lifters, we find we often have to throttle back the total time they work out,” said Dr. Mahoney. “Otherwise you find a tremendous drop in mood, and a jump in irritability, fatigue and apathy.”

    Because their intense practice regimen puts them at risk for burnout or strain injuries, most elite competitors also make rest part of their training routine, sleeping a full eight hours and often napping a half-hour a day, Dr. Ericsson found.

    ↩︎
  23. “The Role of Deliberate Practice in the Acquisition of Expert Performance” 1993:

    The high relevance of sleep for improving violin performance must be indirect and related to the need to recover from effortful activities such as practice. Consistent with the ratings, sleep is the least effortful of the activities and thus constitutes the purest form of rest. The weekly amount of sleep during the diary week did not differ for the two best groups and averaged 60.0 hr. This average was reliably longer than that for the music teachers, which was 54.6 hr, jF(l, 27) = 5.02, p < .05. Hence the two best groups, who practice more, also sleep reliably longer.

    ↩︎
  24. See “Light Exposure May Cut Production of Melatonin: Study Shows Artificial Light Before Bedtime May Affect Quality of Sleep”, WebMD; “Melatonin production falls if the lights are on”, BBC News; and an older study, “The dark side of light at night: physiological, epidemiological, and ecological consequences”. More recently, the National Sleep Foundation released its 2011 survey finding ever more electronics use before bed; relevant quote:

    “Artificial light exposure between dusk and the time we go to bed at night suppresses release of the sleep-promoting hormone melatonin, enhances alertness and shifts circadian rhythms to a later hour—making it more difficult to fall asleep,” says Charles Czeisler, Ph.D., MD, Harvard Medical School and Brigham and Women’s Hospital. “This study reveals that light-emitting screens are in heavy use within the pivotal hour before sleep. Invasion of such alerting technologies into the bedroom may contribute to the high proportion of respondents who reported that they routinely get less sleep than they need.” Computer or laptop use is also common. Roughly six in ten (61%) say they use their laptops or computers at least a few nights a week within the hour before bed. More than half of generation Z’ers (55%) and slightly less of generation Y’ers (47%) say they surf the Internet every night or almost every night within the hour before sleep.

    ↩︎
  25. “Effects of prior light exposure on early evening performance, subjective sleepiness, and hormonal secretion”, Münch et al 2011 (media coverage):

    We tested the impact of two realistic office lighting conditions during the afternoon on subjective sleepiness, hormonal secretion, and cognitive performance in the early evening hours. Twenty-nine young subjects came twice and spent 8 h (12:00-20:00) in our laboratory, where they were exposed for 6 h to either artificial light (AL) or to mainly daylight (DL). In the early evening, we assessed their salivary cortisol and melatonin secretion, subjective sleepiness, and cognitive performance (n-back test) under dim light conditions. Subjects felt [statistically-]significantly more alert at the beginning of the evening after the DL condition, and they became sleepier at the end of the evening after the AL condition. For cognitive performance we found a [statistically-]significant interaction between light conditions, mental load (2- or 3-back task) and the order of light administration. On their first evening, subjects performed with similar accuracy after both light conditions, but on their second evening, subjects performed [statistically-]significantly more accurately after the DL in both n-back versions and committed fewer false alarms in the 2-back task compared to the AL group. Lower sleepiness in the evening was [statistically-]significantly correlated with better cognitive performance (p<.05).

    ↩︎
  26. To quote extensively from “In Eyes, a Clock Calibrated by Wavelengths of Light”, New York Times:

    So scientists at the University of Basel in Switzerland tried a simple experiment: They asked 13 men to sit before a computer each evening for two weeks before going to bed. During one week, for five hours every night, the volunteers sat before an old-style fluorescent monitor emitting light composed of several colors from the visible spectrum, though very little blue. Another week, the men sat at screens backlighted by light-emitting diodes, or LEDs. This screen was twice as blue.

    “To our surprise, we saw huge differences,” said Christian Cajochen, who heads the Center for Chronobiology at the University of Basel. Melatonin levels in volunteers watching the LED screens took longer to rise at night, compared with when the participants were watching the fluorescent screens, and the deficit persisted throughout the evening. The subjects also scored higher on tests of memory and cognition after exposure to blue light, Dr. Cajochen and his team reported in the May issue of The Journal of Applied Physiology. While men were able to recall pairs of words flashed across the fluorescent screen about half the time, some scores rose to almost 70% when they stared at the LED monitors.

    The finding adds to a series of others suggesting, though certainly not proving, that exposure to blue light may keep us more awake and alert, partly by suppressing production of melatonin. An LED screen bright enough and big enough “could be giving you an alert stimulus at a time that will frustrate your body’s ability to go to sleep later,” said Dr. Brainard. “When you turn it off, it doesn’t mean that instantly the alerting effects go away. There’s an underlying biology that’s stimulated.”

    …Artificial light has been around for more than 120 years. But the light emitted by older sources, like incandescent bulbs, contains more red wavelengths. The problem now, Dr. Brainard and other researchers fear, is that our world is increasingly illuminated in blue. By one estimate, 1.6 billion new computers, televisions and cellphones were sold last year alone, and incandescent lights are being replaced by more energy-efficient, and often bluer, bulbs.

    In January in the journal PLoS One, the University of Basel team also compared the effects of incandescent bulbs to fluorescents modified to emit more blue light. Men exposed to the fluorescent lights produced 40% less melatonin than when they were exposed to incandescent bulbs, and they reported feeling more awake an hour after the lights went off. In addition, the quantity of light necessary to affect melatonin may be much smaller than once thought. In research published in March in The Journal of Clinical Endocrinology and Metabolism, a team at the Harvard Medical School reported that ordinary indoor lighting before bedtime suppressed melatonin in the brain, even delaying production of the hormone for 90 minutes after the lights were off, compared with people exposed to only dim light.

    What do these findings mean to everyday life? Some experts believe that any kind of light too late into the evening could have broad health effects, independent of any effect on sleep. For example, a report published last year in the journal PNAS found that mice exposed to light at night gained more weight than those housed in normal light, even though both groups consumed the same number of calories. Light at night has been examined as a contributor to breast cancer for two decades. While there is still no consensus, enough laboratory and epidemiological studies have supported the idea that in 2007, the World Health Organization declared shift work a probable carcinogen. Body clock disruptions “can alter sleep-activity patterns, suppress melatonin production and disregulate genes involved in tumor development,” the agency concluded.

    See also a Navy-funded RPI sleep study, which found “Only the higher level of blue light resulted in a reduction in melatonin levels relative to the other lighting conditions.”, or an animal study finding that blue light at night induced more depressive symptoms than did red light (Bedrosian et al 2013).↩︎

  27. A fellow Zeo user, Sanjiv Shah has reportedly found large benefits for himself from the orange glasses, although he seems to not have a webpage on it but given talks; from Technology Review, “The Measured Life”:

    Sanjiv Shah, a longtime insomniac who participates in the Boston group, believes that wearing orange-tinted glasses for several hours before bed makes it easier for him to fall asleep. (The theory is that the orange tint blocks blue light, which has been shown in both human and animal studies to influence circadian rhythms.) To quantify the effects, he used not only the Zeo but also a thumb-size device called the Fitbit, which incorporates an accelerometer that measures movement, and a camera trained on his bed to record his sleep for a month. His results: without the glasses, he took an average of 28 minutes to fall asleep, but with them he took only four.

    ↩︎
  28. Fellow programmers: you would particularly benefit from using Redshift/F.lux. If you don’t believe my anecdote, read some testimonials.↩︎

  29. My own experience was positive, and teenagers are frequently sleep-deprived due to factors like their circadian rhythms wildly conflicting with school schedules (see my rough notes on the topic). However, children & adolescents have melatonin secretion rates 5-9 times greater than middle-aged or very old adults, and those very high rates combined with large supplements like 3mg may (and I use this word advisedly) be able to push melatonin concentration to high enough levels that some negative chemistry may happen (extremely high doses have often been used in clinical trials with no noted side-effect eg. 40mg/day but these are almost all adults). In addition, there’s general uncertainty about how melatonin interacts with puberty and why the concentration falls so much with age.↩︎

  30. Melatonin has been played with in extremely high doses for purposes other than sleep modification; for example, at <20mg it may help with thermal injury, and at 75mg and higher (doses up to 200mg are mentioned), it’s a female contraceptive. My personal experience with taking 9 mg, is that I will sleep for 10-12 hours and will then awake with a nasty headache that lasts for the rest of that day. (My experience reminds me of the descriptions given by a lot of people who discover that the dose they tried of 3mg or 1mg was too high: with the too-high doses, they woke feeling bad and with a headache. So as one would expect, the right dose differs from person to person.) Melatonin seems to quickly lose effectiveness and soon becomes counterproductive; Wikipedia summarizes:

    Large doses of melatonin can even be counterproductive: Lewy et al[68] provide support to the ‘idea that too much melatonin may spill over onto the wrong zone of the melatonin phase-response curve’ (PRC). In one of their subjects, 0.5 mg of melatonin was effective while 20 mg was not.

    My own dose of 1.5mg is not due to any real kind of experimentation; rather, I created a batch of about a thousand capsules using split 3mg pills, have not yet used them up, and have no real interest in throwing them out or otherwise changing them. When I run out, I will experiment with different doses like 1mg.↩︎

  31. “Low, but not high, doses of melatonin entrained a free-running blind person with a long circadian period”, Lewy et al 2002↩︎

  32. “Entrainment of Free-Running Circadian Rhythms by Melatonin in Blind People”↩︎

  33. Sack et al 2000 found that 10mg succeeded in forcing circadian rhythms in the blind; they tapered the dose down to 0.5mg for a few subjects over 4 months, and the rhythms seemed to persist.↩︎

  34. From press release summary of ConsumerLab.com’s melatonin testing:

    Can melatonin help you sleep? “Melatonin supplements may help some people get to sleep sooner, particularly those with chronic sleeping problems, but don’t just buy any supplement - they vary [substantially] in strength, dosage, and cost,” says ConsumerLab.com President, Tod Cooperman, M.D. ConsumerLab.com recently selected and tested nine different melatonin supplements. The testing showed that all contained their labeled amounts of melatonin, but the suggested daily dosage ranged from 1 mg to 50 mg [!]; and cost ranged from just 4 cents to $1.36 for an equivalent dose of melatonin. This means you may not be using the right dose for your needs and you could be paying as much as 33 times more than necessary. ConsumerLab.com also found that one supplement failed to properly disclose all of its ingredients.

    ↩︎
  35. I say very pessimistic because my actual price is ~1.4¢ per night. In December 2009, Tommy Health sold 120x3mg for $2.49; I bought 6, with shipping, at $20.89. Remembering that I take half a pill, the price per night is , or, 1.38¢. May 2013, I tried Now Foods’s 250x1mg melatonin+niacin+B6+magnesium formulation for $12.16 or 4.9¢ per night; the additional parts didn’t seem to make much difference. In March 2014, I bought Nature’s Bounty 4x180x1mg for $19.93, or 2.77¢ per night.↩︎

  36. Like all models, this is wrong. There are many ways in which it could be wrong or misleading to value an hour of your time at minimum wage or higher. While all models are wrong, some are more useful than others and this is an useful model.↩︎

  37. Here’s another place I am being overly conservative to guard against optimism on my part; the minimum wage was boosted to $7.25 in 2009, which obviously makes the gained hour that much more valuable and consequently all the figures will favor melatonin that much more.↩︎

  38. If , then x=~0.973.↩︎

  39. Applying the usual ‘net present value of a future sum’ formula, which goes , we find that 60 years from now, the $2000 is worth $92: . We can be pessimistic and value the annual return at $200, and sum the 60 years in Haskell with sum (map (\x -> 200 * (1-0.05)^x) [0..60]), or $38245.↩︎

  40. Many people swear by working in public places like libraries or cafes. Neeti Gupta remarks in “Grande Wi-Fi: Understanding What Wi-Fi Users Are Doing in Coffee-Shops” (2004) that:

    …when we are alone in a public place, we have a fear of “having no purpose”. If we are in a public place and it looks like that we have no business there, it may not seem socially appropriate. In coffee-shops it is okay to be there to drink coffee but loitering is definitely not allowed by coffee-shop owners, so coffee-shops patrons deploy different methods to look “busy”. Being disengaged is our big social fear, especially in public spaces, and people try to cover their “being there” with an acceptable visible activity.

    ↩︎
  41. Professional coach P.J. Eby remarks thatMost Akrasia Techniques Are Subject To ‘Meta’-Akrasia. If you procrastinate taking your pills or doing your exercises, your hygenic method is unstable: the more you delay, the more likely you are to delay some more. The same is true for maintaining your ‘trusted system’ in Getting Things Done, breaking your tasks into Pomodoros, or whatever other focusing method you use. And of course, if you put off doing your motivation technique, it’s not going to motivate you.”↩︎

  42. To clarify: In the hypothetical case, why would you be willing to pay one hour of your time (which could be spent doing anything, even high-quality work or experiences and perhaps especially high quality work, given all the research on good sleep and mental performance) to gain a few cents by not buying melatonin, but not willing to pay one hour of easy Wikipedia editing to gain many more cents?↩︎

  43. See the multiple large studies cited in “Vitamin Pills: A False Hope?”; note that at least 3 studies showed increased disease & mortality rates associated with multivitamin usage, and Wikipedia mentions a few downbeat reports & commissions:

    Interestingly, antioxidants show no anti-cancer effects (eg. lung, prostate, breast, or colon cancer) and no mortality reduction in very large datasets and Cochrane reviews (sometimes increased mortality), and only weak evidence in randomized trials measuring elderly infection rates; the long-running physicians trial found an overall cancer reduction for its multivitamin patients but no benefits to the vitamin E or vitamin C or beta carotene patients (nor in women), with vitamin E & beta-carotene fingered as increasing mortality when consumed in >RDA amounts.

    That’s when they don’t actually cause problems with cancer (eg. negating chemotherapy and destroying benefits from exercise like endurance training, see also the post “Antioxidants and Exercise: More Harm Than Good?” or the review “Antioxidants and Skeletal Muscle Performance: ‘Common Knowledge’ vs. Experimental Evidence”).

    Multivitamins don’t have much research backing, and a risk with multivitamins is that they are a moral hazard and an example of risk homeostasis - because people mistakenly expect benefits (large or small), they feel free to act in unhealthy ways:

    The study, published online today in the journal Addiction, describes two experiments run by the authors. In the first experiment, run as a dummy health-food test, 74 daily smokers were given a placebo, but half were told they had taken a Vitamin C supplement. The smokers then took a one-hour unrelated survey during which they were allowed to smoke. Those who thought they had taken a vitamin pill smoked almost twice as much as those who knew they had taken a placebo (the control group) and reported greater feelings of invulnerability. The second experiment was an expanded version of the first, with 80 participants taken from a larger community and half told they were taking a multivitamin pill. The one-hour survey also contained questions about attitudes to multivitamins. The smokers who thought they had taken a multivitamin once again smoked more than the control group. But this time, researchers found that among the multivitamin group, smokers with more positive attitudes toward multivitamins experienced a higher boost in perceived invulnerability and smoked even more than their less enthusiastic counterparts. In other words, the amount of extra smoking rose if the smoker expressed a conscious belief that multivitamins increased health.

    Of course, such contrarianism receives its own critics; Phil Goetz says that 3 of the major anti-vitamin studies have fatal statistical flaws.↩︎

Similar Links

[Similar links by topic]