Reply to: Loss of QALY in mammography screening reported by Zahl et al.

Abstract

Dear editor, In their letter, Njor and Lynge argue that mammography screening leads to 5% overdiagnosis and 20% reduction in breast cancer mortality and assume that all translates to reduced overall mortality. Njor et al. base their estimates on their own studies only, and ignore other more comprehensive studies from Denmark. Njor et al. included only 8% (57,763) of the available female Danish population when studying overdiagnosis. A later study including the complete Danish female population could not reproduce their estimate and reported 48% overdiagnosis. Their other study reporting 20% reduction in breast cancer mortality only included data from Copenhagen. In another study including all women invited to screening in the 1990s (twice as many), no effect of mammography screening on breast cancer mortality was seen. Systematic reviews of mammography screening trials show 20–30% overdiagnosis. Detection rate of cancer including ductal carcinoma in situ has increased from 3 per 1,000 in the screening trials to 8–11 per 1,000 in modern mammography screening. This is chiefly due to improved mammography techniques (see our references in the Appendix of our published paper). Thus, assuming 5% overdiagnosis in a present screening program using modern techniques as in Denmark is simply not plausible. Finally, Njor et al. claim that the reduction in breast cancer mortality fully translates to reduction in total mortality. However, in meta-analyses of all the randomized trials, no reduction in overall mortality is seen. One reason may be that overdiagnosis leads to overtreatment and increased risk of dying of other causes. Another reason is that cause-specific mortality is a biased outcome because of lack of blinding. Unfortunately, the randomized studies in breast cancer screening were done in an era before the use of modern treatment and mammography techniques, so we may have to rely on data from observational studies to inform today’s practice. Results from observational studies vary, as do their use of different designs and methods, which influences their trustworthiness. We believe there are compelling arguments that the Njor et al. studies may not provide the most appropriate estimates for effect and harms of mammography screening. Yours sincerely, Per-Henrik Zahl Mette Kalager