"Trust no one. Assume nothing."
This was the constant refrain of one of my attendings in residency whenever I answered any of his questions about patients by referring to what a prior radiology report had said. His advice has served me well in radiology, where you can expect a certain rate of things that go unreported.
Its application in real life is even more important, where we're dealing not with honest mistakes, but with un-noticed biases and even deliberate misrepresentation. That's why I simply don't take at face value any figure put out by advocacy groups. This is easy for groups I disagree with, but I'm making it a point to treat the groups on my side with the same skepticism.
As an example of an advocacy group on my side of things misbehaving, take the case of the British children's cancer group (Children with Cancer, UK). According to this episode of the More or Less podcast, the group basically pretended an increase in absolute numbers of cancers in young adults and children (driven by increase in population and earlier detection) was an increase in cancer rate in children. The program asked for an explanation from Professor Denis Henshaw, the scientific director of the charity, who in interviews with the popular press, had said things like "We were shocked to see the figures, and it’s modern lifestyle I’m afraid."
In response to whether the popular headlines conflating absolute numbers with rate were correct or if it was right to use the more anxiety-inducing total number in the headline of the society's press release, even though mortality from childhood cancers are falling, even though the rate is much lower than 40%, Dr. Henshaw said "We are certainly interested in drawing attention to figures whatever they are...Prevention is better than cure...so we think drawing attention to this issue either in general or in particular is the right thing to do...I think the purpose of this press release was to encourage people to come to the conference."
"Trust no one. Assume nothing."
No comments:
Post a Comment