Night shift work has been suspected to increase breast cancer risk but epidemiological studies have been inconsistent due to heterogeneous assessment of exposure to night work. To overcome this limitation, we pooled data of five population-based case-control studies from Australia, Canada, France, Germany, and Spain into a single harmonized dataset using a common definition of night work including 6093 breast cancer cases and 6933 population controls. The odds ratio for breast cancer in women who ever worked at night for at least 3 h between midnight and 5 a.m. as compared to never night workers was 1.12 (95% CI 1.00-1.25). Among pre-menopausal women, this odds ratio was 1.26 [1.06-1.51], increasing to 1.36 [1.07-1.74] for night shifts ≥ 10 h, 1.80 [1.20-2.71] for work ≥ 3 nights/week, and 2.55 [1.03-6.30] for both duration of night work ≥ 10 years and exposure intensity ≥ 3 nights/week. Breast cancer risk in pre-menopausal women was higher in current or recent night workers (OR = 1.41 [1.06-1.88]) than in those who had stopped night work more than 2 years ago. Breast cancer in post-menopausal women was not associated with night work whatever the exposure metric. The increase in risk was restricted to ER+ tumors, particularly those who were both ER+ and HER2+ . These results support the hypothesis that night shift work increases the risk of breast cancer in pre-menopausal women, particularly those with high intensity and long duration of exposure. Risk difference between pre- and post-menopausal women deserves further scrutiny.
Keywords: Breast cancer; Case–control study; Circadian disruption; Night shift work; Pooled analysis.