Response to Hersch et al.

Inevitably, Hersch et al. ¹ disagree with our criticism ² of their Lancet study's conclusions. ³ In fact, we agree with much of their response but disagree on the purpose of a decision aid and on whether the one used in their study was balanced.

We believe that ‘decision aids’ should help people make a decision, while Hersch et al. think that providing information and education is enough. A leaflet that leaves a substantial proportion unable to decide whether the intervention offered is good for them has not aided decision making and may cause anxiety. We question whether it is ethical to offer screening while implying that it is a difficult decision to make. When screening is a ‘close call’ (as with prostate cancer), we believe that sending invitations is inappropriate.

With regard to the balance of information, we have concerns over four areas:

The estimates of overdiagnosis and lives saved: the figures quoted are considerably less favourable than those in the UK Independent Review ⁴ (which the authors cite) and it is unclear how they were derived.

Failure to communicate the magnitude of the harms and benefits: the use of icon-arrays implies that the there is an equivalence between the harm of overdiagnosis and the benefit of not-dying from breast cancer. Most women strongly prefer overdiagnosis to underdiagnosis, ⁵ and the view that ‘I prefer taking the risk of [being] “overdetected” than… risk of dying’ was felt by Hersch et al. in an earlier paper to be representative of women. ⁶ We believe that the harm of dying from breast cancer (and all that that entails) greatly exceeds the harm of having an overdiagnosis (including the harm of its treatment).

Failure to distinguish between the treatment of overdiagnosed breast cancers and that of advanced breast cancers: The decision aid describes breast cancer treatments, including radical mastectomy and chemotherapy, but fails to mention that virtually all overdiagnosed breast cancers will be localised and therefore treated by local excision without chemotherapy. Additionally the aid does not describe what it is like to die from breast cancer, nor does it point out that screening could mean the difference between dying at age 82 instead of age 62.

Lack of endorsement: A provider should not be offering screening if it has not concluded that it is a good idea. For many people, the fact that independent experts support screening is important information for making a decision and should therefore be included in any decision aid.

We cannot endorse a decision aid (for an intervention deemed to be a major public good by independent experts) that 36% of women considered to be slanted away from the intervention. In our opinion, it is possible to promote informed choice while endorsing mammography screening and pointing out the potential harms (including overdetection).