Setting: Four Canadian organised breast cancer screening programmes. Methods: Women aged 50-69 receiving dual screening (CBE and mammography) (n=300,303) between 1996 and 1998 were followed up between screen and diagnosis. Outcomes assessed by mode of detection (CBE alone, mammography alone, or both CBE and mammography) included referral rate, positive predictive value, pathological features of tumours (size, nodal status, morphology), and cancer detection rates overall and for small cancers (<10 mm or node-negative). Heterogeneity in ndings across programmes was also assessed.
Results:On rst versus subsequent screen, CBE alone resulted in 28.5-36.7% of referrals, and 4.6-5.9% of cancers compared with 52.6-60.1% of referrals and 60.0-64.3% of cancers for mammography alone. Among cancers detected by CBE, 83.6-88.6% were also detected by mammography, whereas for mammographically detected cancers only 31.7-37.2% were also detected by CBE. On average, CBE increased the rate of detection of small invasive cancers by 2-6% over rates if mammography was the sole detection method. Without CBE, programmes would be missing three cancers for every 10,000 screens and 3-10 small invasive cancers in every 100,000 screens.