Night shift work is a 'probable' carcinogen. The human evidence remains far from settled

Night shift work is officially classified as "probably carcinogenic to humans" — Group 2A on the International Agency for Research on Cancer's scale — a designation its expert working group set out in 2019 [s1]. But "probable" is a precise word, not a hedge for "yes," and a 2026 meta-analysis of night-shift nurses shows exactly why the human evidence stays cautious: the breast cancer signal that everyone worries about appears only after decades of shift work, and it fades toward nothing once the analysis corrects for the biases in the underlying studies [s2].

What the classification means, and what it doesn't

The IARC places agents into groups by the strength of evidence that they can cause cancer, not by how dangerous a given exposure is in practice. Group 2A — "probably carcinogenic to humans" — sits below Group 1 (established human carcinogens such as tobacco) and reflects a working group's judgement that the evidence, while suggestive, falls short of proof [s1]. The 2019 IARC Monographs Volume 124 group placed night shift work in that category [s1]. The classification is about hazard identification: it says the evidence points toward a real carcinogenic potential, not that any individual shift worker is likely to develop cancer.

Breast cancer has been the focus of that concern, on the theory that circadian disruption and suppressed night-time melatonin could influence hormone-driven tumours. The classification is what generated headlines; the meta-analysis is what shows how strong — or weak — the human data beneath it actually is.

What the 2026 meta-analysis found

Healthcare workers are the ideal population to test the question, because so many of them do years of night shifts and their occupational histories are relatively well documented. The 2026 review pooled 12 studies covering 12,132 breast cancer cases and estimated pooled relative risks by duration of exposure, compared with never having worked nights [s2].

The gradient is the whole story. For ever having done night shift work, the pooled relative risk was 1.05 (95% confidence interval 0.96 to 1.14) — essentially no increase [s2]. It edged up with cumulative exposure: 1.11 (0.96 to 1.28) for 10 or more years, 1.25 (1.01 to 1.55) for 20 or more years, and 1.68 (0.77 to 3.65) for 30 or more years [s2]. Only the 20-plus-years estimate reached statistical significance, and even that is fragile: the reviewers report it lost significance in an influence analysis and approached unity — no effect — after correcting for possible publication bias [s2]. The 30-year figure looks alarming until you notice its confidence interval runs from 0.77 to 3.65, spanning everything from a protective effect to a near-quadrupling, on just three studies [s2].

Why the caution is warranted

The review is candid about the quality of what it pooled. Sixty per cent of the included studies were rated as "definitely" or "probably" at high risk of bias in how they characterised exposure — that is, in the basic task of pinning down who actually worked nights, how many, and for how long [s2]. Higher risk estimates tended to come from case-control and nested designs and from studies that reconstructed lifetime occupational history, the very approaches most vulnerable to recall and selection biases [s2]. The authors' bottom line is deliberately two-sided: long-term night shift work may increase breast cancer risk, but the association is far from established — and if it were real, it would still account for a substantial number of breast cancers given how many people work nights [s2].

Holding two ideas at once

The honest position is not "night shifts cause cancer" and not "it's a myth." It is the one both sources actually support: there is enough mechanistic and epidemiological signal for the world's leading cancer agency to flag night shift work as a probable carcinogen, and simultaneously the human evidence for the flagship outcome, breast cancer, is inconsistent, driven by the longest-exposure subgroups, and sensitive to the biases in the data [s1] [s2]. That is what a Group 2A classification is designed to convey — a credible concern that has not been nailed down.

For the millions who work nights, the practical implication is not a screening recommendation, which this article does not make, but a reason the research keeps running: the exposure is common enough that even a modest true effect would matter, and the current studies are not yet good enough to confirm or dismiss it.