The Four Hours Before Bed
A longevity physician's three small evening changes—earlier dinner, dimmer light, a paper wind-down—that finally reset my sleep.
For three years I woke up at 3:40 in the morning. Not occasionally. Most nights, close enough to the same minute that I stopped looking at the clock, because looking made it worse.
I assumed it was age. I was in my mid-forties, I had a practice to run and a book in progress, and 3:40 felt like the invoice for that. Every physician I know has a version of this story, where we diagnose ourselves with something vague and move on, because the alternative is taking our own advice.
What it turned out to be had almost nothing to do with what I was doing at bedtime. It had to do with the four hours before it. I changed three things across that window and the 3:40 wake-up stopped. It has not come back.
I’m an internist, and I now ask every patient who tells me they can’t sleep to walk me through their evening rather than their bedtime routine. The answer is nearly always in the earlier part.
1. Moving the last meal earlier than felt reasonable
I used to eat at 8:30 or 9, because that’s when the day was actually over. Then I’d be in bed by 10:30 wondering why I felt warm and restless.
The study that reorganized my thinking here is a small, tightly controlled crossover trial out of Johns Hopkins, published in the Journal of Clinical Endocrinology and Metabolism in 2020. Twenty healthy young volunteers, ten men and ten women, spent two separate stays in a research unit eating the identical meal at either 6 PM or 10 PM, with a fixed 11 PM bedtime and blood drawn hourly through the night.
Same food. Same calories. Same people.
The late dinner produced a peak glucose roughly 18 percent higher, and overnight fat oxidation dropped by about 10 percent. Plasma cortisol went up. The effects were most pronounced in people who habitually went to bed earlier, which is to say the people for whom 10 PM was genuinely late relative to their own clock.
That’s a single meal, in healthy twenty-somethings, measured over one night. I want to be careful not to inflate it. But the authors’ framing is the part I quote to patients: if these effects recur chronically, they plausibly contribute to obesity and glucose dysregulation. Most of us are not eating one late dinner. We’re eating four hundred of them a year.
The mechanism also explains the 3:40 specifically. A large late meal drives a glucose rise and then a compensatory fall in the early hours, and that fall triggers a counter-regulatory cortisol and catecholamine response to pull blood sugar back up. That response wakes you. Separately, core body temperature has to drop for deep sleep to consolidate, and digestion is heat-producing work that runs directly against it.
I moved dinner to 6:30. Three hours before bed is the number I use with patients, and it produced the fastest effect of anything I changed. Not gradual. The first week.
The pushback is always about family schedules and about not being hungry at 6:30, and both are legitimate. When someone truly can’t move dinner, I tell them to move the size instead. A smaller late meal behaves differently than a large one. It’s not as good as an earlier one. It’s not nothing.
2. Cutting overhead light after sunset
This is the one that sounds like wellness theater. It is the best-evidenced item on the list.
Gooley and colleagues at Brigham and Women’s and Harvard published a study in the same journal in 2011 in which participants lived in a research center for at least five consecutive days under controlled lighting. Ordinary room light in the hours before bed, the kind in every kitchen in America, shortened the duration of melatonin production by roughly 90 minutes compared to dim light. Room light during habitual sleep hours suppressed melatonin by more than 50 percent.
Ninety minutes. From the ceiling.
The reason this happens is that melatonin release is gated by specialized retinal cells that care about intensity, wavelength, and crucially the angle light arrives from. Light coming from above, which is what every overhead fixture produces, reads to that system as midday sun.
I did not want to believe this mattered as much as it does. I put lamps at eye level and below in the rooms I actually use after dark, switched to warm bulbs, and stopped touching the overheads. My wind-down went from something I had to manufacture to something that arrived on its own around nine.
What I tell my patients: you don’t need candlelight. You need to stop being lit from above after sunset. It’s a lighting change, not a lifestyle change, and it costs about sixty dollars.
The screens matter too, but less than people assume and much less than the ceiling. A phone is a small light source six inches from your face. A kitchen ceiling is a large one filling your entire visual field.
3. The wind-down I resisted for years
Here’s the one I fought hardest.
I did not want an evening routine. It struck me as something for people with more time than I had, and there’s a version of it circulating now that’s so elaborate it becomes another job. If your wind-down has nine steps, you haven’t solved your sleep problem. You’ve given yourself homework.
What I eventually adopted takes twenty minutes and has two parts: I write down whatever is unfinished, on paper, in a list I don’t organize or prioritize. Then I read fiction.
The writing half has better evidence than I expected. Scullin’s group at Baylor published a polysomnography study in the Journal of Experimental Psychology in 2018 with fifty-seven adults who spent five minutes before bed writing either a to-do list for the coming days or a list of tasks already completed. The to-do list group fell asleep measurably faster. And the more specific and detailed the list, the faster they fell asleep, which is the opposite of what you’d predict if writing about unfinished work simply stirred up worry.
The interpretation the authors favor, and the one that matches my experience, is offloading. The tasks don’t get done. The brain stops rehearsing them because they’ve been externalized somewhere reliable.
It’s fifty-seven young adults on a single night in a lab, so I hold it loosely. But it costs five minutes and a piece of paper, and I’ve now watched enough patients report the same thing that I ask about it routinely.
The fiction half has no trial behind it and I won’t pretend otherwise. It’s a boundary. I needed something my mind could not usefully continue at three in the morning, and medical journals failed that test badly.
What didn’t work, despite the hype
I want to name what I tried first, because I spent money and about eighteen months before I addressed anything above.
Melatonin supplements did essentially nothing for me, and in hindsight that’s unsurprising. Meta-analyses of melatonin for sleep onset show a genuine but small effect, on the order of several minutes. My problem was not a melatonin deficiency. It was a melatonin timing problem, created at 7 PM by my kitchen ceiling, and a three milligram tablet at 10 PM does not fix a signal that was already scrambled.
A sleep tracker made things actively worse for a while. There’s a recognized phenomenon here, named orthosomnia by Baron and colleagues in 2017, in which the pursuit of perfect tracked sleep produces the anxiety that then degrades sleep. I became a person who woke and immediately checked a score, and on nights the score was poor I felt worse than the sleep itself warranted. I took a two-month break from the ring and slept better during it.
Magnesium helped a little. I still take it. It’s a modest assist on top of a fixed foundation, and I had the order backward for a year and a half.
Start with the earliest one
If you change one thing, change the earliest one in the day.
That’s the principle I’ve landed on. Evening problems compound forward. A late meal makes the light exposure matter more, and both make the wind-down harder, and by the time you’re lying in bed doing breathing exercises, you are trying to solve at 10:45 a problem that was created at 6:30.
So move dinner. Not by three hours, not perfectly, not every night. Move it ninety minutes for two weeks and watch whether the hour you keep waking at changes.
Mine moved to 5:15, which was still annoying. And then it stopped.
If the four hours before bed are quietly breaking your sleep, you’re not alone. In this essay I walk through three small evening changes—earlier dinner, dimmer light, and a paper wind-down—that stopped a 3:40 a.m. wake-up that had followed me for years
Dr Shiv Kumar Goel is a board-certified internist practicing functional and integrative medicine at Prime Vitality Care in San Antonio, Texas. Follow his Substack, healingthesplit, for more of his writings.
References
• Gu C, Brereton N, Schweitzer A, et al. Metabolic effects of late dinner in healthy volunteers: a randomized crossover clinical trial. J Clin Endocrinol Metab. 2020;105(8):2789–2802.
• Gooley JJ, Chamberlain K, Smith KA, et al. Exposure to room light before bedtime suppresses melatonin onset and shortens melatonin duration in humans. J Clin Endocrinol Metab. 2011;96(3):E463-E472.
• Scullin MK, Krueger ML, Ballard HK, Pruett N, Bliwise DL. The effects of bedtime writing on difficulty falling asleep: a polysomnographic study comparing to-do lists and completed activity lists. J Exp Psychol Gen. 2018;147(1):139–146.
• Baron KG, Abbott S, Jao N, Manalo N, Mullen R. Orthosomnia: are some patients taking the quantified self too far? J Clin Sleep Med. 2017;13(2):351–354.