Sleeping better: what the plate and the glass decide

Many people who call themselves bad sleepers fall asleep perfectly well. What they struggle with is what follows: waking around 3am, broken sleep through the second half of the night, and getting up more tired than they went to bed.

The distinction is useful, because the two problems have different causes. Falling asleep depends mostly on how alert you are at bedtime. The continuity of the night depends a great deal on what is still circulating in your blood while you sleep.

And there, three things weigh heavily and are settled in advance: what dinner was made of, caffeine, and alcohol.

This is the thread left hanging in Gaining energy: a carb-heavy dinner is paid for twice.

The mechanism is the same as in the daytime. The meal sends a lot of glucose in at once, the pancreas answers with a release of insulin, and blood sugar comes back down two to three hours later, passing below where it started. Except that at that hour, you are asleep.

Blood sugar dropping overnight triggers a corrective hormonal response, cortisol and adrenaline among others, whose job is to bring the level back up. Those are also waking hormones. Hence those middle-of-the-night wakings with no apparent cause, often around 3am, which is precisely the interval you would expect after a dinner eaten at 8pm.

The lever is not to skip dinner, which would do the opposite, but to flatten its curve: keep the protein and the vegetables, lighten the starch portion, and finish on something other than a fast sugar.

Caffeine has a half-life of about five to six hours in adults. A coffee at 4pm therefore still leaves half its dose active at 9 or 10pm, and a quarter in the middle of the night.

An often-cited trial measured this directly. Volunteers were given 400mg of caffeine, the equivalent of two to three cups, either at bedtime, three hours before, or six hours before. The result: even six hours before bed, objectively measured sleep time fell by more than an hour.

The second figure is the more instructive one. The participants noticed nothing: their sleep diaries reported no trouble at all. So you can lose an hour of sleep a night to an afternoon coffee while being convinced caffeine has no effect on you.

Alcohol is a sedative, not a sleep aid. The difference is exactly the one from the first section: it helps you fall asleep and stops you sleeping well.

Its effect is biphasic, and well described. In the first half of the night, falling asleep is faster and deep sleep increases. Then, as the alcohol is metabolized, the second half becomes markedly more fragmented, with repeated wakings. REM sleep, the kind that carries memory consolidation, is reduced across the whole night, and the more so the higher the dose.

The overall picture is therefore misleading: you fell asleep faster, and slept worse. Since what you notice is mostly the falling asleep, the evening drink passes for a help when it is one of the causes.

On this there is no trade-off to present: the effect is dose-dependent, the less there is the better the night goes, and zero remains the best case. No amount of alcohol improves sleep.

Three nutrients come up constantly on this subject. Their level of evidence is not the same, and saying so is part of the information.

Magnesium takes part in regulating the nervous system, and a deficiency often goes with poorer quality sleep. Correcting an insufficient intake is therefore justified. The idea that taking more improves sleep in someone whose status is already adequate, on the other hand, rests on studies that are few and small.

The best sources are seeds and nuts, and this site's tables give the figures:

The European reference intake is 375mg a day. The table Seeds & nuts uses a 30g serving. At that size, a handful of pumpkin seeds supplies 165mg, close to half the daily need. A handful of sunflower seeds gives 98mg, a handful of almonds 81mg.

So it is reachable, on one condition: that the magnesium comes from there. Because on this point the site runs into its own logic. Whole grains are often presented as the answer, and they do contain far more than refined ones:

Refining takes away two thirds of the magnesium, along with the fiber. But that table says only one thing: among grains, the whole one keeps what the refined one lost. It does not make grains a good source, and the ratio to carbs is what shows it.

A supplement need not be the first reflex, since the first lever is on the plate and a handful of pumpkin seeds already covers close to half the need. But if that route does not take hold, it is worth discussing, with a doctor or a pharmacist rather than alone: doses and interactions are not all equivalent, and magnesium calls for caution in kidney failure.

Two amino acids come up on this subject, for different reasons.

Tryptophan is one of the nine essential amino acids described in Protein: quantity does not tell you quality. It is the precursor of serotonin, itself the precursor of melatonin. Without it, the chain does not happen. It is found in eggs, cheese, poultry, oily fish and nuts, which is to say in the same foods recommended elsewhere on this site.

Glycine is a non-essential amino acid, but it is the one with the most concrete sleep data. A Japanese trial showed that 3g taken about an hour before bed shortened the time to fall asleep and to reach deep sleep, improved perceived sleep quality and reduced next-day drowsiness, without altering sleep architecture.

The proposed mechanism is elegant: glycine increases blood flow to the skin, which lowers core body temperature. And that drop is precisely one of the physiological signals that trigger sleep.

An order appears when you read the evidence column from top to bottom: the best established levers are all removals, and the ones that consist of adding something bring up the rear. That is therefore where to start, all the more since it is also where the effects are largest: an hour of sleep recovered from simply moving a coffee, no supplement offers that. As long as the afternoon caffeine, the late drink or the sweet dessert stay in place, they cancel most of what could be added elsewhere.

Which does not make the additions useless: correcting an insufficient magnesium intake is justified, and glycine has real data behind it, preliminary though it is.