Imaging for early stage breast cancer

Statistical considerations

Sample size •Non-inferiority design

•Assumed 2.5% LR rate at 5 years in control group •Exclude >2.5% increase in LR rate in either test group •80% power •2.5% α, 1-sided for each comparison of test group versus control •645 required in each group (1935 total)

Local recurrence is any invasive or preinvasive tumour in the breast

Coles et al, Lancet 2017