%0 Journal Article
%T Night shift work and female breast cancer: a two-stage dose-response meta-analysis for the correct risk definition.
%A Moon J
%A Ikeda-Araki A
%A Mun Y
%J BMC Public Health
%V 24
%N 1
%D 2024 Jul 31
%M 39085800
%F 4.135
%R 10.1186/s12889-024-19518-2
%X BACKGROUND: The hypothesis of this study is night shift work exposure can increase the risk of female breast cancer. To validate this hypothesis, the authors conducted a two-stage dose-response meta-analysis with improved quality on this topic.
METHODS: The medical librarian searched PubMed, EMBASE, and the Cochrane Library on December 30th, 2022. The eight inclusion criteria were determined and strictly applied to the selection process. A reliable dose-response meta-analysis methodology was applied.
RESULTS: Reliable 10 cohort (total cases: 15,953, and total person-years: 6,812,138) and 11 case-control reports (total cases: 9196, and total controls:12,210) were included in the final analysis. The pooled risk ratio (RR) of female breast cancer (from cohort studies) for 1, 10, 20, and 30 years of night shift work exposure was 1.0042 (95% CI 1.0014-1.0070), 1.0425 (95% CI 1.0138-1.0719), 1.0867 (95% CI 1.0278-1.1490), and 1.1328 (95% CI 1.0419-1.2317), respectively. The pooled odds ratio (OR) of female breast cancer (from case-control studies) for 1, 10, 20, and 30 years of night shift work exposure was 1.0213 (95% CI 1.0108-1.0319), 1.2346 (95% CI 1.1129-1.3695), 1.5242 (95% CI 1.2386-1.8756), and 1.8817 (95% CI 1.3784-2.5687), respectively.
CONCLUSIONS: This study has several strengths from the perspective of a dose-response meta-analysis: Strictly applied eight inclusion criteria, separately synthesized RRs from cohort studies and ORs from case-control studies, clearly defined exposure dose, years of night shift work for each risk estimate, a reliable dose-response meta-analysis methodology, and careful considering of selection, exposure, and outcome biases and confounder adjustment for each study. This careful consideration of potential biases and confounding led to the exclusion of unreliable two cohort and five case-control studies.