THE REVIEW OF RANDOMIZATION IN THE CANADIAN NATIONAL BREAST SCREENINGSTUDY - IS THE DEBATE OVER

THE RANDOMIZATION PROCEDURE in the Canadian National Breast Screening Study (NBSS) is assessed in this issue (see pages 193 to 199) by Drs. John C. Bailar III and Brian MacMahon. They conclude that although the re was ample opportunity for the randomization process to be subverted , no evidence of subversion was found. This is unlikely to allay all c oncerns about randomization, because there are still puzzling differen ces between the arms of the NBSS in a number of baseline variables. Fo r example, the existence of prior health claims for breast cancer for women who entered the NBSS in Manitoba has raised the possibility that subversion occurred. Although the question may never be resolved, one lesson is clear: randomization in clinical trials should be managed i n a manner that makes subversion impossible. As for the clinical impli cations of the NBSS for women in their 40s, physicians may now look to the results of randomized trials that have been published more recent ly. A meta-analysis of these results suggests that screening mammograp hy reduces deaths from breast cancer among women in their 40s, but con tinued follow-up over the next few years will be needed to settle the debate.