How A Charity Exploits Innumeracy
Before I get to the substance of the post I want to apologize for the dearth of postings of late. I was in Corpus Christi trying a bank case (go figure). I lost. It's been awhile since a jury branded me with the scarlet "L"; and it still feels like a punch in the gut. Anyway ...
For years we've been harping on risk, Bayes' Rule, our collective blind spot for base rates and mass confusion about the proper use, interpretation and limits, of percentages. So when we saw "How A Charity Oversells Mammography" in British Medical Journal taking on Susan G. Komen for the Cure for exploiting our cognitive weaknesses we knew instantly that it was blogworthy.
We've previously addressed the issue of why mammography leads to so many false positives but what's really interesting about the BMJ article is its take down of the following fallacy: early detection leads to a much greater chance of winning the "five year survival rate" award - the most common metric for assessing outcomes in cancer cases.
The problem is that mammography often doesn't make a difference even when it does accurately detect breast cancer. First, we all come with an expiration date. For some people the cause of expiration will be cancer because sadly for many cancers, including some breast cancers, there is no cure. Now let's say that in 2011 someone was diagnosed with a form of breast cancer that was refractory to treatment and that person died in 2012. She clearly didn't make it 5 years. But what if, thanks to mammography, she'd been diagnosed with the breast cancer in 2006 and then had succumbed to it in 2012? She'd be just as dead but would get counted as a success story.
Second, some cancers grow very slowly, some never grow at all and some even go away on their own. Thus, when mammography detects a breast cancer that was never going to be the cause of the patient's death anyway that patient gets counted as a mammography success even though her cancer had no effect on her expiration date.
Saying that mammography screening saves lives is true only because the way we count lives saved is so deeply flawed.
