Sleep and immunity: what the inoculation experiments measured

In 153 volunteers given rhinovirus by nose, people sleeping under seven hours were 2.94 times more likely to develop a clinical cold, and those under 92 per cent sleep efficiency 5.50 times. A 2024 crossover in 46 people put people on four hours a night for five nights and measured IL-6 and CRP.

goodWell supported, with the usual caveats about effect size.Revised 18 Oct 2026

A plain unlabelled glass dropper bottle and a small ceramic dish on a folded pale blue cotton sheet on a light wooden bedside table

You cannot randomise people into a decade of bad sleep. You can inoculate 153 of them with a rhinovirus and count the colds.

The sleep-and-immunity literature has a property almost nothing else in this field has: you can put a virus in someone’s nose on purpose. That turns a twenty-year observational argument into an experiment with a clock running. It also means the field’s best evidence is about the common cold, which is a frivolous endpoint for a longevity blog and yet turns out to be the cleanest thing available.

The inoculation study

Cohen and colleagues, Archives of Internal Medicine 2009, enrolled 153 healthy men and women aged 21 to 55. For fourteen consecutive days they reported sleep duration, sleep efficiency — the percentage of time in bed actually asleep — and whether they felt rested. Average scores over those fourteen days became each person’s baseline.

Then the participants were quarantined, given nasal drops containing a rhinovirus, and monitored for a clinical cold, defined as infection in the presence of objective signs of illness, on the day before and for five days after exposure.

The associations were graded, not thresholded. Participants who had averaged less than seven hours of sleep were 2.94 times more likely to develop a cold than those who had slept eight hours or more (95% CI 1.18 to 7.30). Participants below 92 per cent sleep efficiency were 5.50 times more likely than those at 98 per cent or above (95% CI 2.08 to 14.48). The authors checked the obvious confounders and report the relationships could not be explained by differences in prechallenge virus-specific antibody titres, demographics, season of the year, body mass, socioeconomic status, psychological variables or health practices.

One null result in the same abstract is worth carrying: the percentage of days feeling rested was not associated with colds. Feeling rested and being resilient are different variables, and only one of them predicted anything here.

The direction also runs backwards

The 2000 Physiology & Behaviour paper from Henry Ford’s sleep disorders centre is the mirror image. Twenty-one volunteers susceptible to rhinovirus type 23 were inoculated nasally and had polysomnographically recorded sleep for five nights, with performance testing twice daily and a multiple sleep latency test each day.

In the symptomatic individuals, total sleep time fell by an average of 23 minutes, consolidated sleep by 36 minutes, and sleep efficiency by about 5 per cent during the active viral period compared with the incubation period. Psychomotor performance was impaired, and these changes were significantly greater than in the asymptomatic individuals.

So the cold makes you sleep worse as well as the bad sleep making the cold more likely. That is a feedback loop, and any single-week protocol is going to sit somewhere inside it.

Cutting sleep in a laboratory and reading the inflammation

The 2024 Brain, Behaviour, and Immunity trial from Engert and colleagues is the one that measures the mechanism directly. Forty-six healthy participants, 19 women and 27 men, aged 19 to 63, in a double-blind randomized placebo-controlled crossover with three protocols. Each protocol was a 14-day at-home monitoring phase followed by an 11-day, ten-night laboratory stay. Under the two restriction conditions, participants had five nights with a four-hour sleep opportunity, followed by three recovery nights at eight hours. Under the control condition they had eight hours throughout.

Two findings, both with the drug interaction rather than on its own. Acetylsalicylic acid taken in the evening each day reduced interleukin-6 expression in LPS-stimulated monocytes (p < 0.05) and reduced serum C-reactive protein (p < 0.01) after five nights of restriction compared with placebo in the restriction condition. It also reduced the number of cyclooxygenase-1/COX-2 double-positive cells among LPS-stimulated monocytes after two nights of recovery sleep following the restriction (p < 0.05).

The design is worth noting for a different reason. The authors set out to test whether bluntting inflammation would blunt the sleep-deficiency response, on the logic that if inflammation is the mechanism, adding an anti-inflammatory should change the outcome. The paper’s own framing is that few interventional sleep studies have investigated whether improving sleep affects inflammatory processes, and that the results of those which have suggest complementary approaches may be necessary. That is an unusual admission in a field that usually promises a mechanism.

What the review says about direction

Akkaoui, Palagini and Geoffroy’s 2023 chapter on sleep–immune cross talk is useful precisely because it is a map rather than a finding. During physiological sleep, pro-inflammatory cytokines IL-1, IL-6 and TNF-α decrease, and anti-inflammatory cytokines IL-4 and IL-10 decrease too. It is the ratios that carry meaning: pro-inflammatory activity at the beginning of the night, then anti-inflammatory during the second part of the night. NK cells leave the blood because they migrate to secondary lymphoid organs, while their activity increases.

And in the other direction, short sleep duration appears associated with increased inflammatory processes and increased risk of infection. The authors close by stating that more studies are necessary — including, notably, on insomnia disorder itself, where the evidence base is thin.

The one animal study, and why it is in this article

Bellesi and colleagues, Journal of Neuroscience 2017, used serial block-face scanning electron microscopy in mouse frontal cortex after 6 to 8 hours of sleep, spontaneous wake, sleep deprivation, or chronic restriction of about five days.

Astrocytic phagocytosis, mainly of presynaptic components of large synapses, increased after both acute and chronic sleep loss relative to sleep and wake. MERTK expression and lipid peroxidation in synaptoneurosomes increased to a similar extent after short and long sleep loss. Then the distinction that matters: chronic restriction, but not acute deprivation, produced morphological signs of microglial activation and enhanced microglial phagocytosis of synaptic elements — without obvious signs of neuroinflammation in the cerebrospinal fluid.

The authors read it as the brain clearing worn components of heavily used synapses in response to extended wake, and note that low-level sustained microglial activation can lead to abnormal responses to a secondary insult.

What the data do not show

They do not show that poor sleep causes infections in general. The Cohen study measures a clinical cold after a deliberate inoculation, over five days, in 153 people — that is a specific and short exposure. They do not show an effect magnitude for cancer, cardiovascular disease or mortality; nothing here extrapolates a 2.94-fold risk of a cold into a 2.94-fold risk of anything that kills you. They do not show that restoring sleep reverses the inflammatory shift measured in the 2024 trial, because the design tested a drug rather than a recovery schedule. And the mouse paper describes a plausible mechanism at the synapse; it does not demonstrate that the same process occurs in a human brain or that it has clinical consequences.

This article summarises published experiments and reviews. It is not medical advice and it is not a recommendation for any individual.