Randomization in the Canadian National Breast Screening Study: a review for evidence of subversion.

The authors assess the randomization strategy that had been used in the Canadian National Breast Screening Study (NBSS). Document experts at a private investigation and security company were hired to assist in reviewing instances in which names of subjects were altered in the "allocation books" (the basic instrument used to assign, at random, participants to either the mammography or the usual-care arm). The review was restricted to records from 3 NBSS centres where women assigned to the mammography arm had a distinctly higher (not necessarily significant) number of deaths from breast cancer than those assigned to the usual-care arm, and to records from 2 centres where, for limited periods, administrative problems were reported. In most cases the underlying, original name could be identified. The document experts found no evidence of a deliberate attempt to conceal the alterations. A search of the NBSS database for the underlying and superimposed names revealed that only 1 of the women whose name had been deleted or superimpsed died of breast cancer. She was in the mammography arm. The authors' thorough review of ways in which the randomization could have been subverted failed to uncover credible evidence of it. They conclude that even if there had been acts of subversion, they could only have been few in number and, given that there was only 1 death from breast cancer in the group reviewed, the alterations could have had only a trivial effect on the study findings as reported in 1992.