NYT headline today: US Panel advises that prostate test may do more harm than good.. Here’s why, briefly: around 70% of elevated readings from the test are false positives – they do not indicate cancer. Men who get a positive reading, and the urologists who treat them, feel constrained by a positive reading to get invasive follow-up tests or even prophylactic treatment. This follow-up has generally worse health outcomes, including incontinence and sexual dysfunction, than if the test result had been ignored. But faced with that positive reading few men will accept that they may be better ignoring that 30% possibility that they have cancer, however slow growing. Urologists have a lot to lose if they recommend against treatment and nothing much to lose if the treatment has a bad outcome. So it would be better to not even know, because you don’t really know anything useful, there’s just the illusion of control in a situation which can’t, with current medical technology, be controlled.
The other side of this is interesting, too. When the same panel recommended in 2009 that mammograms be used only every two years and only for women between 50 and 74 there was a furor. Doctors groups and cancer victims advocacy groups strongly disagreed. The same thing is happening, with even more heat in this case. It’s not just that doctors stand to lose a lot of income. Probably more important is that doctors and cancer advocacy groups have a huge emotional investment in their treatment of past patients. It’s not easy to admit to yourself that you have caused large amounts of unnecessary suffering from overtreatment. Patients have an investment in believing that their treatment was justified and necessary.
The logic of the US Preventative Services Taskforce‘s case is clear, but it runs up against the emotional argument “if even one life is saved it’s worth the cost.” When that cost is suffering and worse quality of life for many, it may be a bad bargain. The problem is that risk is badly weighted, psychologically, when there’s fear involved. If you have an elevated PSA reading, do you say “I accept the risk because the balance of value goes this way” or do you say “I’ll endure the suffering to feel safer” – even if that feeling is largely an illusion.
We see this problem everywhere, it’s Pascal’s Wager. But Pascal’s wager is a bad bet; he phrases it as “I have everything to gain and nothing to lose” but ignores the costs – society warped by superstition and time wasted in church, for example. Look at the TSA’s Security Theatre. Failure to understand risk. It’s why so many people are stuck in a job they hate, or personal relationships that are unsatisfying. But it’s also why people will so strongly defend their risk-averse bad decisions. If you want to design a cult, take advantage of that. Get people to cut off a piece of themselves, metaphorically or literally, as the price of joining. I’ve met so many people who hated school but insist their children go to the same one. Remember the argument that we can’t leave Iraq because it’s cost us so much already that it would be disloyal to the fallen?
This is about Baysean analysis. If the human race has a weakness it’s that the fight-or-flight survival patterns baked in to us have some very bad decision making heuristics built in. And one of them says: “if it looks like it might be connected; if it looks like there could be a pattern there — then there is. We only see the smallest fraction of all patterns and everything is really a pattern.” Well, no. Also yes but that’s another story. The no is that we assume patterns and then look for evidence, so we always have survivor bias. Tombstone epitaph: should have applied Bayes.