FAQ  /  Women's Hormone Health

Why Can't I Sleep Through the Night in Perimenopause?

The Hot Flash Part, and Why It Is Less Simple Than It Sounds

---

Vasomotor symptoms are a genuine contributor, and the cleanest demonstration of that comes from an experimental model rather than an observational one. Twenty-nine healthy premenopausal women, average age 27, were given a gonadotropin-releasing hormone agonist to suppress estradiol rapidly and reproduce menopause without the confound of aging. Sixty-nine percent developed persistent hot flashes. Each additional reported nighttime episode was associated with a 62 percent increase from baseline in polysomnography-measured wake after sleep onset, a 3 percent increase in awakenings, and a 6 percent increase in stage N1 sleep. Objectively recorded nighttime episodes correlated with measured wake after sleep onset.

A companion analysis in the same model complicates the picture in a way worth sitting with. Across 48 sleep studies, 165 hot flash episodes were recorded objectively. Most occurred during wake or stage N1, and 80 percent occurred just before or during an awakening. But objectively recorded episodes were not associated with an increased rate of transitions to wake or to light sleep. Only self-reported nighttime hot flashes were. The most plausible reading is that a woman who is already sleeping lightly and surfacing repeatedly is awake often enough to notice and remember her hot flashes, while the flashes she sleeps through do not register. The flash and the awakening keep company, but the flash is not always the thing that did it.

Which is why treating the flashes and still sleeping badly is common rather than puzzling. Something else is running underneath.

Losing Progesterone Removes a Real Sleep Signal

---

Progesterone falls earlier in the transition than estradiol does, because ovulation becomes irregular before it stops, and cycles without ovulation produce no meaningful luteal progesterone at all. A woman can be years from her last period, still cycling, with estradiol that looks unremarkable on paper, and already be spending most months without the progesterone she used to make.

That matters for sleep specifically because progesterone is not sedating as itself. It is converted to allopregnanolone, a positive allosteric modulator at the GABA-A receptor, which is the same receptor family that benzodiazepines and the z-drugs act on. Losing progesterone removes an endogenous GABAergic signal that was there every luteal phase for thirty years.

Two small trials give this a human footing. In ten healthy postmenopausal women given 300 mg of oral micronized progesterone nightly for 21 days in a randomized, double-blind, crossover design, intermittent time spent awake fell and rapid eye movement sleep in the first third of the night increased, with no effect on next-day cognitive performance. In eight postmenopausal women without sleep complaints, given the same dose for three weeks, progesterone did nothing to already-normal sleep, but when sleep was deliberately disrupted by overnight blood sampling, wake after sleep onset was 53 percent lower, slow-wave sleep duration nearly 50 percent higher, and total slow-wave activity nearly 45 percent higher than on placebo. The authors described it as acting like a physiologic regulator rather than a hypnotic, and the distinction is not cosmetic. Conventional hypnotics tend to suppress deep sleep. This restored it.

Route and timing are not incidental. Oral progesterone works as a prodrug to allopregnanolone, and the first-pass metabolism that a transdermal route is designed to avoid is exactly the step that does most of that conversion. In postmenopausal women given low-dose oral micronized progesterone, allopregnanolone exposure over the twelve hours after a dose ran 196 percent above what endogenous production alone accounts for, roughly triple, and steady-state trough concentrations sat within the range seen across a normal menstrual cycle. Timing then follows from the pharmacokinetics. Allopregnanolone concentrations peak around two hours after an oral dose, and measurable sedation appears at one to two hours, which is why the dose is usually taken about an hour before bed rather than at lights-out. That places the effect at sleep onset rather than well after it.

The honest caveats: both pharmacokinetic studies were small, one used a dose well below what is typically prescribed and the other enrolled ten healthy volunteers who were mostly men taking an extended-release formulation in a different clinical context, so treat the time course as indicative rather than exact. The two sleep trials were small as well, eight and ten women, and both were done after menopause rather than during the transition.

The Early-Morning Cortisol Rise, With Less Holding It Back

---

There is a third piece, and it explains the specific complaint of waking at two or three in the morning and lying there alert.

Cortisol is not flat overnight. Secretion begins climbing in the second half of the night, several hours before waking, and that climb is tied to sleep structure rather than independent of it. Nocturnal cortisol pulses cluster around the transitions out of slow-wave sleep and into lighter sleep and wake, so the rise and the shift toward arousable sleep happen together. Interrupting sleep pushes it further: experimentally disrupted sleep alters the overnight ACTH and cortisol pattern rather than leaving it intact, which means an awakening is not a neutral event for the axis.

Separately, cortisol appears to rise during the menopausal transition itself. In 169 women followed prospectively with monthly urine collections, cortisol increased across the late transition stage compared with the year before it, a change not seen around the middle stage or the final menstrual period.

Put those alongside the loss of allopregnanolone and the picture is coherent. The same hours of the night when cortisol is climbing and sleep is naturally lightest are the hours that used to be buffered by an endogenous GABAergic tone that is now largely gone. Less inhibition against a rise that was always there.

That synthesis is worth labeling plainly. Each link in it is established individually. No single study has demonstrated the assembled chain in perimenopausal women, and in fact the study showing the cortisol rise found that the women with rising cortisol did not differ from the others in reported sleep. So treat this as a mechanism that fits the physiology and fits the symptom, not as a proven cause.

Two Age-Related Layers Underneath

---

Two further changes are arriving at the same time and are not hormonal in the ovarian sense.

The first is the ordinary age-related change in sleep architecture. Deep sleep becomes shorter and shallower and sleep becomes more fragmented with age in both sexes, which is why the progesterone work above was framed around restoring disturbed sleep rather than improving good sleep. A woman entering perimenopause is losing a sleep signal at the same age her sleep was becoming more fragile anyway.

The second is sleep-disordered breathing, and this one is regularly missed. In the Wisconsin Sleep Cohort, 589 women underwent in-laboratory polysomnography. After adjustment for age, body habitus, smoking, and other confounders, the odds of having five or more apnea and hypopnea events per hour were 2.6 times higher after menopause and the odds of fifteen or more were 3.5 times higher, both statistically significant. The perimenopausal odds ratios, 1.2 and 1.1, were not significant, so the evidence for a rise during the transition itself is weaker than the evidence for a rise after it. What the data support is that a woman in this age range with snoring, unrefreshing sleep, or daytime sleepiness deserves an actual evaluation rather than an assumption that this is hormones.

What This Is Not, and Why That Matters

---

The reason to pull these apart is that they do not all respond to the same thing.

Sleep that is fragmented by untreated apnea will not be fixed by progesterone, and sedating it is worse than useless. Sleep broken by night sweats responds to whatever controls the vasomotor symptoms. Sleep that is short on the GABAergic signal responds to restoring progesterone. Insomnia that has become a conditioned pattern, months of lying awake in a bed that now cues alertness, responds to cognitive behavioral therapy for insomnia and not much else, regardless of hormonal status. Most women in perimenopause who are sleeping badly have more than one of these running at once.

So the useful question is not what is causing this, singular. It is which of these are contributing, and in what proportion, which is answerable with a careful history, a look at the cycle, and a low threshold for a sleep study.

Why a Sleep Aid Often Disappoints

---

Women in this situation are frequently offered a hypnotic, and the result is frequently partial. There are two reasons for that, and both are mechanistic rather than a matter of finding the right pill.

The first is that most hypnotics shorten the time it takes to fall asleep, while the complaint here is usually not falling asleep. It is staying asleep, specifically through the window when cortisol is rising and sleep is lightest.

The second is that conventional hypnotics tend to suppress slow-wave sleep while increasing total time in bed asleep. You get more hours and less restoration, which is how a person sleeps seven hours on medication and still wakes unrefreshed. Replacing the missing signal is a different strategy from sedating over the gap, and it is the one that fits the physiology of what was lost.

Dr. Darrell Wilcox · @wellnessdoc_4everyoung on Instagram

This content is for educational purposes only and does not constitute medical advice, diagnosis, or treatment. Progesterone and hormone therapy are prescription treatments with their own indications, contraindications, and monitoring requirements, and persistent insomnia warrants evaluation by your own physician, including assessment for sleep-disordered breathing where indicated. Individual results differ from trial averages, and the studies described here were conducted in specific populations that may not match your own situation. Dr. Wilcox is licensed to practice in multiple states. See About for current licensure.