Benefit of 30% may be substantial overestimate

EDITOR—Blanks et al have attempted to model the decline in mortality from breast cancer in England and Wales and to estimate the proportion of this decline due to screening.1 Unfortunately, even their estimate of a modest effect may be too large: the effect of tamoxifen at ages 55-74 may be larger than that at ages 50-54, as many of these younger women may be premenopausal and have oestrogen negative cancers, gaining less benefit from tamoxifen.

Blanks et al do not comment on the unexplained rise in mortality from breast cancer in the United Kingdom from about 1965 to 1990. The recent fall in this mortality in the United Kingdom may have been partly due to the removal of the factor that caused the rise. Such a rise was not seen in North America, where mortality from breast cancer was stable until about 1990, since when similar falls in both Canada and the United States have occurred, of the same order as that in the United Kingdom.2

Blanks et al perpetuate the unproved assumption that falls in mortality from breast cancer are due to the early detection by mammography of cancers when they are small and impalpable. It was always impossible to explain the rapid fall in mortality from breast cancer in women aged 50-64 in the health insurance plan trial3 by such an effect; recently published Canadian data cast further doubt on this assumption.4

Gøtzsche and Olsen have suggested that imbalances in many of the breast screening trials cast doubt on …