Communicating Risk and Uncertainty

Case study in breast-cancer screening

“Scales of uncertainty”

Benefits and harms of cancer screening

• ‘Shared-care’ and ‘informed choice’• New leaflet will present pros and cons• Will not make recommendation

Citizen’s ‘Jury’

• Scientific review used evidence to try to answer question

‘What is the effect of being invited for screening?’

• But relevant question for women is ‘What is the effect of attending screening’

• Preferred ‘over-treatment’ to ‘over-diagnosis?’ etc

• Etc etc

Uncertainty

But these numbers are very uncertain!

Full of ‘conditionalities’

Not quantifiable as a confidence interval

Best current estimate, but may change

Words of Estimative Probability (WEP)

• NIC scale of likelihood and confidence

How can we communicate deeper uncertainties due to the quality of the evidence?

Part of GRADE scale used in Cochrane Collaboration and 25 other organisations to assess confidence in estimates of medical treatment effects

Sandman’s 4 conclusions

1.Don’t just acknowledge the uncertainty, proclaim it

2. Proclaim how uncertain you are – from – “I’m taking a shot in the dark here” to – “I’m almost certain but there are still a few

remaining doubts to clear up.”

3. Distinguish your level of uncertainty now from the level of uncertainty earlier

4. Come across as human

1990: John Gummer – ‘beef is safe’

1992: three cows in every 1,000 in Britain had BSE

1996: government admits link between BSE and the human form of the disease, new variant CJD