Mammography screening and breast cancer mortality in Australia: an aggregate cohort study.

BACKGROUND: Evidence that mammography screening reduces breast cancer mortality derives from trials, with observational studies broadly supporting trial findings. The purpose of this study was to evaluate the national mammographic screening programme, BreastScreen Australia, using aggregate screening and breast cancer mortality data.
METHODS: Breast cancer mortality from 1990 to 2004 in the whole Australian population was assessed in relation to screening exposure in the target of women aged 50-69 years. Population cohorts were defined by year of screening (and diagnosis), five-year age group at screening (and diagnosis), and local area of residence at screening (and diagnosis). Biennial screening data for BreastScreen Australia were related to cumulated mortality from breast cancer in an event analysis using Poisson regression, and in a time-to-event analysis using Cox proportional hazards regression. Results were adjusted for repeated measures and the potential effects of mammography outside BreastScreen Australia, regionality, and area socio-economic status.
RESULTS: From the adjusted Poisson regression model, a 22% (95% CI:12-31%) reduction in six-year cumulated mortality from breast cancer was predicted for screening participation of approximately 60%, compared with no screening; 21% (95% CI:11-30%) for the most recently reported screening participation of 56%; and 25% (95% CI:15-35%) for the programme target of 70% biennial screening participation. Corresponding estimates from the Cox proportional hazard regression model were 30% (95% CI:17-41%), 28% (95% CI:16-38%) and 34% (95% CI:20-46%).
CONCLUSIONS: Despite data limitations, the results of this nationwide study are consistent with the trial evidence, and with results of other service studies of mammography screening. With sufficient participation, mammography screening substantially reduces mortality from breast cancer.