Unclear methods in estimate of screening effect in women ages 40 to 49 years

Authors


Hellquist et al.1 report a 26%-29% reduction in breast cancer mortality in women aged 40-49 years in Sweden; however, the authors used unclear methods. Women in unscreened regions were matched with women in screened regions for the time screening started, but because this varied widely between screened regions, it is not clear how this complex matching was done and whether it was adequate.

The authors added and subtracted person-years based on assumed lead times, but the extent to which screening brings forward the time of diagnosis is unknown. The authors were also selective regarding the biases for which they adjusted; for example, because they did not adjust for a prescreening difference in breast cancer mortality of 6%, they overestimated the effect.

It is more reliable to count the number of deaths. Breast cancer mortality in Sweden in women aged 40-49 years has declined 36% since 1989, although only half of these women were offered screening, whereas the decline was only 16% in women aged 50-69 years who were offered screening for over 20 years.2

Denmark has a unique control group, because only 20% of the population was screened for 17 years; however, we could not find an effect of screening, and the decline in breast cancer mortality was largest in young women who were not screened.3 This was also a general trend in a review of 30 European countries.2 Similarly, we did not observe an effect of screening in Sweden or the United Kingdom,3 and a recent Norwegian study was also negative.4

Screening probably had some effect 30 years ago. However, because of improved treatments and breast awareness, it is questionable whether there is an effect today, in any age group, when the data used do not require assumptions.

Karsten Juhl Jørgensen MD*, Peter C. GYøtzsche DrMedSci*, * The Nordic Cochrane Centre, Copenhagen, Denmark

Ancillary