Mammographic screening and breast cancer mortality: a case-control study and meta-analysis

Abstract

Background: Observational studies are necessary to assess the impact of population screening on breast cancer mortality. While some ecological studies have notably found little or no association, case-control studies consistently show strong inverse associations, but they are sometimes ignored, perhaps due to theoretical biases arising from the study design. We conducted a case-control study of breast cancer deaths in Western Australia to evaluate the effect of participation in the BreastScreen Australia program, paying particular attention to potential sources of bias, and undertook an updated meta-analysis of case-control studies.

Methods: Our study included 427 cases (women who died from breast cancer), each matched to up to 10 controls. We estimated the association between screening participation and breast cancer mortality, quantifying the effect of potential sources of bias on our findings, including selection bias, information bias, and confounding. We also conducted a meta-analysis of published case-control studies.

Results: The OR for participation in the Western Australian BreastScreen program in relation to death from breast cancer was 0.48 [95% confidence interval (CI), 0.38-0.59; P < 0.001]. We were unable to identify biases that could negate this finding: sensitivity analyses generated ORs from 0.45 to 0.52. Our meta-analysis yielded an OR of 0.51 (95% CI, 0.46-0.55).

Conclusions: Our findings suggest an average 49% reduction in breast cancer mortality for women who are screened. In practice, theoretical biases have little effect on estimates from case-control studies.

Impact: Case-control studies, such as ours, provide robust and consistent evidence that screening benefits women who choose to be screened.