Of the 1804 women, 29 became ineligible after randomisation and 14 did not begin the assigned therapy. A further 269 in the placebo group and 295 in the tamoxifen group were not fully compliant. The patient characteristics were similar in the two groups, 65% were postmenopausal, 16% had positive resected sample margins after definitive surgery and more than 80% of the tumours had maximum dimensions of 1 cm or less.
At 5 years of follow-up, 83% of patients who received placebo were event?free compared with 87% of tamoxifen-treated patients. The placebo group had 130 invasive and non-invasive breast-cancer events in the ipsilateral breast, contralateral breast, or presented as metastases at regional or distant sites compared with 84 in the tamoxifen group. Women in the tamoxifen group had fewer breast-cancer events at 5 years than did those on placebo (8.2 versus 13.4%, P = 0.0009). The cumulative incidence of all invasive breast-cancer events in the tamoxifen group was 4.1% at 5 years: 2.1% in the ipsilateral breast, 1.8% in the contralateral breast, and 0.2% at regional or distant sites. In the tamoxifen group the estimated rate ratio represented 37% fewer breast-cancer events. The risk of ipsilateral-breast cancer was lower in the tamoxifen group even when sample margins contained tumor and when DCIS was associated with comedonecrosis. The 5 year survival was 97% for the two groups.