LAUSR

Dear editor, Zahl et al. assumed the effect of modern screening programs to be (i) a 10% reduction in breast cancer mortality; (ii) 50–75% overdiagnosis; and that (iii) 20–80% of reduction in breast cancer mortality translate into reduction in all-cause mortality. Using 50% for the last variable, a QALY loss of 437–875 per 100,000 was expected in women screened for 20 years starting at age 50. Denmark is probably the country with the best data for estimating the effect of modern screening, because 20% of women aged 50–69 years were invited for screening during a 17 years period, whereas the other 80% of women were not invited, and opportunistic screening was rare. Register data furthermore allowed for control of prescreening regional differences. Long-term follow-up of invited compared to notinvited women showed a 20% reduction in breast cancer mortality, and a 5% overdiagnosis (recalculated to correspond to Zahl et al.’s numbers). The IARC working group on breast cancer screening found an overall effectiveness of 22% in reduction in breast cancer mortality, which is well in accordance with the Danish data. The same working group found overdiagnosis to range from 2% to 22%, though including some studies that did not include a long-term follow-up of breast cancer incidence in postscreening age. In the randomized controlled trials on breast cancer screening, the reductions in breast cancer mortality were not accompanied by statistically significant reductions in all-cause mortality, which according to Zahl et al. suggested “that not all of the reductions in breast cancer mortality translated into reductions in all-cause mortality”. This is a strange argument given that breast cancer deaths constituted 3% of deaths in the trial populations, and 20% of 3% is 0.6%; a difference that would require a very large trial population to find statistically significant. In our view, the most appropriate input data for a QALY calculation would be a 20% reduction in breast cancer mortality; a 5% overdiagnosis; and a 100% translation of breast cancer mortality into all-cause mortality. Among the scenarios calculated by Zahl et al., the one that comes closest uses a 20% reduction in breast cancer mortality; a 20% overdiagnosis; and an 80% translation of breast cancer mortality into allcause mortality. Zahl et al. found a net gain in 2816 OALY per 100,000 women for this combination. In modern breast cancer screening programs, a gain in QALY is thus a more likely outcome than a loss. Yours sincerely, Sisse H. Njor Elsebeth Lynge