How do cardiovascular risk estimates differ when shift work exposure is defined by schedule type versus cumulative night-shift counts?

Why the way you count night work changes the number you get

"Shift work" is not one exposure. A schedule-type definition asks a yes/no question: does this person work nights at all, or work rotating shifts? A cumulative-count definition asks how much: how many nights over ten years, how many per month, how many consecutive nights in a row. These are different measurements, and they can produce different risk estimates even in the same population because they capture different things — holding a night-shift job versus accumulating a dose of night work over time.

The cumulative approach is built on the idea of dose. A meta-analysis of 81 studies found moderate-confidence evidence of increased inflammation, dyslipidaemia and impaired cardiac excitability among night shift workers, and reported dose-dependent effects, meaning the intensity and duration of night work appeared to contribute to cardiovascular risk [1]. That is a claim about accumulation, and it is the kind of claim a schedule-type comparison cannot test directly.

Counting nights gives a modest, often imprecise signal

In the German Gutenberg Health Study, researchers sorted workers by how many nights they had worked in the ten years before baseline: low (up to about one year's worth), middle (one to three years), and high (more than three years). Compared with people who never worked nights, the adjusted hazard ratios for a first cardiovascular event were 1.26, 1.37 and 1.19 for the low, middle and high groups [3]. Every estimate points above 1.0, but every confidence interval crosses 1.0, so none is statistically distinguishable from no effect. The pattern is also not a clean staircase — the middle group is highest, not the top group.

A much larger Danish cohort of health care workers, using day-by-day payroll records rather than self-report, found a similar picture. Overall incidence rate ratios for coronary heart disease were 1.06 for female night workers and 1.22 for male night workers compared with day workers [5]. When the researchers looked within night workers at number of monthly night shifts, cumulative night shifts, years of rotating night shifts, years with any night shift, and consecutive night shifts, they found no consistent exposure-response relation for either sex [5]. The authors concluded that in this low-exposure population, reducing the extent of night work would not be expected to reduce coronary heart disease risk [5].

That last point is worth pausing on. The Danish cohort averaged under two night shifts per month and less than four years of night work [5]. A dose-response analysis in a population with little dose has limited room to detect a gradient, so the flat result is about this population as much as about night work in general.

Schedule comparisons add a different kind of information

Schedule-type comparisons are cruder but can capture something a night count misses: the disruption itself. A controlled laboratory study put 14 healthy young adults on three days of either a simulated day or simulated night schedule, then measured over 400 blood lipids. After the night schedule, mean triglyceride levels were generally higher and mean phospholipid levels lower than after the day schedule, and the timing of certain triglyceride rhythms shifted [2]. The authors describe these lipid changes as possibly premorbid to the elevated cardiovascular risk seen in night-shift workers epidemiologically — a suggestion about mechanism, not a demonstrated cause [2].

This is where the two definitions diverge most sharply. A three-day simulated schedule has essentially no cumulative dose to count, yet it still produced measurable lipid changes [2]. A schedule-type definition can register that kind of short-term disruption; a cumulative-count definition, by construction, would score those three days as almost nothing.

Reviews of the broader literature point in the same direction at the level of disease rather than biomarkers. A narrative review of studies from 2013 to 2023 linked shift work to increased carotid artery plaque, arterial stiffness and carotid intima-media thickness — signs of atherosclerosis progression — and suggested shift work belongs alongside hypertension, diabetes and sedentary lifestyle as a cardiovascular risk factor [4]. That review frames shift work as a category of exposure, not as a counted dose.

So which definition should you trust?

Neither is simply better; they answer different questions. If you want to know whether working nights at all is associated with cardiovascular harm, schedule-type comparisons are the natural tool, and they generally show small elevations — 1.06 in women and 1.22 in men in the Danish cohort [5]. If you want to know whether more nights means more risk, you need cumulative counts, and there the evidence is weaker than the dose-dependent framing in the meta-analysis might suggest: the German cohort's estimates were all compatible with no effect [3], and the Danish cohort found no consistent gradient across five different count-based metrics [5].

One practical implication follows from the Danish data: in a population working fewer than two nights a month for under four years, cutting back further showed no expected benefit for coronary heart disease [5]. That does not tell us what happens at heavy, long-term exposure, where the German high-exposure group was too small and too imprecise to settle the question [3]. The honest summary is that schedule type and cumulative counts each capture part of the exposure, and the disagreement between their estimates is mostly about precision and population, not about whether night work is linked to cardiovascular risk at all.

Sources used in this answer

Related research questions