Circadian disruption and ageing: what the human studies support

An IARC classification based on eight epidemiological studies in 2007, a 2017 update across 26 studies that still reports heterogeneity in how exposure is defined, and a review that notes sleep has fallen by an hour and a half over a century. The mechanism is clear; the human effect size is not.

mixedStudies disagree, or the mechanism is contested.Revised 22 Oct 2026

A plain wooden nightstand beside a window at dusk with deep blue light outside and a warm lamp inside, a blank-faced bedside clock and a folded pale blue cloth

The clock in this photograph has no numbers on it, which is also how the exposure is measured in the epidemiology.

Chronobiology has an unusual status in ageing research: the mechanism is about as settled as mechanism ever gets in humans, and the effect size is still unsettled. Both facts come from the same place — a mismatch between what the molecular clock does and what the epidemiology can measure.

What is mechanistically settled

Garaulet, Ordovás and Madrid, in the International Journal of Obesity 2010, state the position of a decade ago cleanly. Shift work, sleep deprivation and exposure to bright light at night increase the prevalence of adiposity. Animal models show that mice with disruption of the Clock gene are prone to developing obesity and metabolic syndrome.

The 2023 review from Franzago and colleagues, in the International Journal of Molecular Sciences, restates the same mechanism from the nutrition side: the human circadian system has a period of approximately 24 hours, and lifestyle and environmental factors of modern societies — artificial lighting, jetlag, shift work and around-the-clock access to energy-dense food — can induce disruptions of that system. There is a complex reciprocal relationship between metabolism and the circadian system in which perturbations in one affect the other.

The reciprocal part is the interesting bit, and it is where the field is genuinely unusual. The same clock genes are argued to influence metabolic health and to modify the individual response to diet, and an interplay between circadian rhythm, gut microbiome and epigenome has been demonstrated.

Where the mechanism is thin on human ground is that the human experiment is small. This blog’s entry on meal timing covers Jamshed and colleagues’ 2019 crossover: eleven overweight adults, four days, early feeding reducing mean 24-hour glucose by 4 ± 1 milligrams per decilitre (p = 0.0003) and altering diurnal patterns in cortisol and the expression of several circadian clock genes (p < 0.05). A real signal, in eleven people.

The classification that is older than most of its citations

The reference point for everything in night-work epidemiology is a 2007 classification. Hansen and colleagues’ 2017 review in Current Environmental Health Reports describes it: the International Agency for Research on Cancer classified shift work that involves circadian disruption as probably carcinogenic to humans, on the basis of limited evidence from eight epidemiological studies on breast cancer plus sufficient evidence from animal experiments.

Note the wording. “Limited evidence” on the human side, “sufficient” on the animal side, and the resulting classification is “probably carcinogenic”, not “carcinogenic”. Three tiers of caution in one sentence, and almost every popular article about shift workers drops all three.

Ten years of new evidence, and what it fixed

The same review’s update is the most useful part of the field. After 2007, nine new case-control studies, one case-cohort study and eight cohort studies were published — which triples the number of studies — and two previous cohorts were updated with extended follow-up.

So the evidence base roughly quadrupled in ten years. Here is what that bought:

There is some evidence that a high number of consecutive night shifts has an impact on the extent of circadian disruption, and thereby increased breast cancer risk. But this information is missing in almost all cohort studies.
More recent case-control studies contributed results concerning breast cancer subtypes in relation to menopausal status and different hormonal subtypes.
Large differences in the definitions of both exposure and outcome may contribute to the observed heterogeneity of results.
Overall the direction points toward increased breast cancer risk, particularly after over 20 years of night shifts, or after shorter periods with many consecutive shifts.

The sentence that governs everything here

Read that list again, because the fourth bullet is the one that matters most. “The assessment of night shift work is different in all of the 26 existing studies.”

Twenty-six studies, twenty-six different definitions of what counts as the exposure. A ten-year shift worker and a twenty-year shift worker who worked four nights then three days are being treated as comparable by somebody, and the reviewers cannot tell which somebody meant what.

The review’s own conclusion about why some cohorts came back null is worth quoting: the missing information on consecutive shifts, combined with short-term follow-up of ageing cohorts, may explain why some cohort studies have null findings. A null result from a cohort that measured a decade of exposure in people who are still alive is not the same as a null result.

The population-level trend nobody designed an experiment for

Garaulet and colleagues’ observation is the one to end on, because it is the strongest evidence in this article and it is not a study at all: the amount of sleep has declined by 1.5 hours over the past century, accompanied by an important increase in obesity.

That is an ecological observation. It cannot establish that the sleep decline caused the obesity increase, and no reviewer claims it does. But it is the context in which every night-shift study sits, and it is the reason the chronobiology of obesity is not a niche concern for a longevity blog.

What these studies do not show

They do not show a quantified effect of circadian disruption on human lifespan: the strongest human evidence in this set is 26 studies with no agreed exposure definition, and the IARC classification rests on limited human evidence. They do not show that fixing your sleep timing fixes your metabolism — the eleven-person, four-day trial measured cortisol rhythm and clock gene expression alongside a glucose change, not a clinical outcome. And the 1.5-hour decline in sleep is an ecological trend with no control group, which is exactly why it belongs at the end and not the beginning.

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