Model of outcomes of screening mammography: Authors' reply

BMJ 2005; 331 doi: (Published 04 August 2005) Cite this as: BMJ 2005;331:351
  1. Alexandra Barratt, associate professor of epidemiology (alexb{at},
  2. Kirsten Howard, research fellow health economics,
  3. Les Irwig, professor of epidemiology,
  4. Glenn Salkeld, associate professor of health economics,
  5. Nehmat Houssami, clinical associate and honorary senior lecturer
  1. Screening and Test Evaluation Programme, School of Public Health, University of Sydney, NSW 2006, Australia

    EDITOR—Zahl and Mæhlen and G⊘tzsche and J⊘rgensen cite a variety of evidence, consistent with our model, to propose that overdetection (and potentially overtreatment) may be substantial in breast screening. Furthermore, Zahl and Mæhlen say that the biological mechanisms underlying the observed overdetection may include regression of small cancers, as well as non-progression.

    Although the frequency of regression is as yet very unclear, evidence from epidemiology and cancer biology are indeed consistent in pointing towards biological mechanisms that control early cancer.1 2 In future, molecular markers may help clarify which screen detected cancers will progress and therefore warrant treatment, and which can be treated minimally or even left alone because they will remain asymptomatic or regress. Evidence already exists that screen detected cancers with certain mammographic features (small, stellate lesions in particular) have an excellent prognosis regardless of treatment regimen.3 Future research findings will provide much needed information on this issue for screening policy, management decisions, and participants in screening programmes.

    Our optimistic estimate of the effect of breast cancer screening is plausible and defensible. We probably overestimated the impact on total mortality as there is no allowance for harmful effects of screening, particularly the mortality (and morbidity) from treatment of screen detected disease. As noted above, the extent of this is currently unclear.

    We agree with Lockwood et al that these are only estimates. However, potential participants should be aware of the uncertainty surrounding outcomes of breast screening. Decisions still have to be made, and our estimates, although imperfect, represent a realistic general picture.

    False negatives may be viewed as an important harm of screening, included under interval cancers in our model. However, overdetection and the potential for overtreatment are important but largely ignored outcomes of screening that warrant attention. Public views about overtreatment are largely unknown, and it remains to be seen how people view the possibility of receiving cancer treatment unnecessarily. Early evidence suggests women want to be told about it and want to take it into account when considering screening.4


    • Competing interests None declared.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    View Abstract

    Log in

    Log in through your institution


    * For online subscription