Ninety minutes less sleep raised insulin resistance by 15% in women

The design is what makes this study worth reading. Thirty-six women aged 20 to 75 — all of them already sleeping a healthy seven to nine hours a night, none with diabetes or heart disease — completed two six-week phases in random order. In one, they kept sleeping normally. In the other, they went to bed with 90 minutes cut from their night, landing at an average of about 6.2 hours. Every woman did both phases, so each acted as her own control. That makes this a randomised crossover trial rather than a survey of people who happen to sleep badly — it tests cause, not just association.

During the short-sleep phase, fasting insulin rose significantly and insulin resistance increased by around 15% across the group, according to the National Institutes of Health's summary of the findings. In postmenopausal women the increase reached roughly 20%. Premenopausal women mainly showed a rise in fasting insulin; in postmenopausal women, fasting glucose tended to drift up as well — the body no longer fully compensating by producing more insulin.

The finding that matters most is the one that's easiest to skim past: the effect was independent of body weight, and independent of any change in body fat across the study. Six weeks of mild sleep restriction is not long enough to make anyone gain meaningful weight, and the researchers checked — the changes in glucose metabolism were not explained by changes in adiposity. So this isn't the familiar story of tiredness leading to worse eating and less movement, which then affects blood sugar. Short sleep appears to act on insulin sensitivity through its own route, in women who were otherwise healthy and at a stable weight.

Two honest caveats. First, the trial is small — 36 women, of whom only 11 were postmenopausal. The 20% figure rests on that eleven, so treat it as a signal about direction rather than a precise measurement of size; it wants replicating in a larger group before anyone leans on the exact number. Second, this is not new research: it was published in Diabetes Care in November 2023. It circulates periodically as though it were a fresh finding, which it isn't — what it is, is good-quality causal evidence that has held up rather than a preliminary result waiting to be overturned.

For women in their forties and fifties this compounds awkwardly, because perimenopause is often the thing taking the sleep in the first place — night sweats, 3 a.m. waking, the broken nights that arrive alongside falling oestrogen. Meanwhile oestrogen decline is separately nudging insulin sensitivity downwards. The trial suggests these aren't two unrelated complaints filed under "midlife": lost sleep is itself doing measurable metabolic work. I've written more on how that overlap plays out in perimenopause and blood sugar, and on where sleep sits among the other levers in what the latest type 2 diabetes research actually supports.

The practical read is undramatic, which is usually a good sign. Nobody needs to optimise their sleep. But if you are managing insulin resistance, prediabetes or type 1 diabetes and you are routinely running on six hours, the evidence now says that is not a neutral background condition — it is part of the picture your glucose numbers are responding to, and worth raising at your next review rather than treating as the thing you'll sort out later.

What the research says