Reply to Kopans

Excerpt

Kopans claims that the Canadian National Breast Screening Study [1] was compromised due to non-random preferential assignment of prevalent breast cancer cases to the screening arm [2]. There is no evidence to support this, only speculation. But for the sake of argument let us suppose that he is right, that a significant proportion of the prevalent cases were palpable and that many of these were sent to the screening arm and an equal number of women with no cancer were sent to the non-screening arm. We would expect an excess number of deaths in the screening arm but most of these would occur in the first 5 years since randomisation. From year five until year 25 the benefit of screening would kick in. That is, we would expect the annual breast cancer mortality rate in the screening arm to exceed that of the unscreened arm transiently and (provided screening works) to fall below that of the unscreened arm between years 5–20. The annual mortality rates are presented in Fig. 1. There is no such pattern.

…