As somebody who’s had to deal with doctors because of a plethora of diseases, I must say you’re absolutely right. (I also shadowed a few and am considering applying to med school.)
I don’t remember what this concept is called, but basically it posits that “one should look for horses, not zebras” and is part of medical education. That is, a doctor should assume that the symptoms a patient has are caused by a common disease rather than by a rare one. So most doctors, thanks to their confirmation bias, dismiss any symptoms that don’t fit the common disease diagnosis. (A girl from my town went to her physician because she complained of headaches. The good doctor said that she’s got nothing to worry about and recommended more rest and relaxation. It turned out that the girl had a brain tumor which was discovered when she was autopsied. The good doctor is still practicing. Would this gross example of irrationality be tolerated in other professions? I think not.)
Most doctors are not so rational because of the way their education is structured: becoming a doctor isn’t so much about reasoning but memorizing heaps of information ad verbatim. It appears that they are prone to spew curiosity-stoppers when confronted with diseases.
soren, please don’t take this the wrong way, but based on what I’ve seen you post so far, you are not a strong enough rationalist to say things like this yet. You are using your existing knowledge of biases to justify your other biases, and this is dangerous.
Doctors have a limited amount of time and other resources. Any time and other resources they put into considering the possibility that a patient has a rare disease is time and other resources they can’t put into treating their other patients with common diseases. In the absence of a certain threshold of evidence suggesting it’s time to consider a rare disease (with a large space of possible rare diseases, most of the work you need to do goes into getting enough evidence to bring a given rare disease to your attention at all), it is absolutely completely rational to assume that patients have common diseases in general. .
None taken, but how can you assess my level of rationality? When will I be enough rationalist to say things like that?
What bias did I use to justify another bias?
Again, testing a hypothesis when somebody’s life is at stake is, I think, paramount to being a good doctor. What’s the threshold of evidence a doctor should reckon?
The question is whether “people with headaches” is the right reference class. If the headache is unusually severe or persistent, it makes sense to look deeper. Also, a doctor can ask for details about the headache before prescribing the expensive tests.
More precisely, the question is whether or not the right reference class is one in which cancer tests are worth while. The headaches would have to be very unusually severe to get enough evidence.
Also, a doctor can ask for details about the headache before prescribing the expensive tests.
It was never mentioned whether or not the doctor asked for details. It’s also possible that none of those reference classes are worth looking into, and she’d need headaches and something else.
Yeah, but the other ones also get ignored if you assume it’s cancer. To my knowledge, they have to be individually tested for. If none is worth testing for individually, it’s best to ignore the headaches.
“The vast majority” != “All”. What’s wrong with “you most likely have nothing to worry about, but I suggest doing this exam the off-chance that you do”? You’ve got to multiply the probability by the disutility, and the result can be large enough to worry about even if the probability is small. (Yes, down that way Pascal’s mugging lies, but still.)
EDIT: Okay, given the replies to this comment I’m going to Aumann my estimate of the cost of tests for rare diseases upwards by a couple of orders of magnitude. Retracted.
Doctors get taught to practice evidence-based medicine. There’s a lack of clinical trials that show that you can increase life span by routinically giving people who suffer from headaches brain scans.
If I understand the argument right, then doctors are basically irrational because the favor empirical results from trials over trying to think through the problem on a intellectual level?
The vast majority of people with headaches don’t have anything to worry about.
If you had a headache, would you want your doctor to find out its cause or would you be satisfied if he told you that ‘the vast majority of people with headaches don’t have anything to worry about’ and sent you home straight away?
This ‘majority’ argument is a fallacious argument. Therefore it’s wrong, much wrong.
Edit #1: If you downvote me, I’d like to get some feedback as to why you’re doing that.
The argument by majority fallacy means arguing that something is true because many people believe it. In the example of the headaches, the argument was that it was likely true because it is true for most people.
What you would want your doctor to do is take the action that maximizes your expected utility, E[U(action)]. Let’s simplify a bit, and say that action can be either “do nothing” or “find cause”. Then the utilities could be something like:
P(not sick) = 0.99 (most people have nothing to worry about)
U(not sick, do nothing) = 0
U(not sick, find cure) = -1 (unnecessary tests, drugs, worry)
U(sick, do nothing) = -10 (possibly more headaches, or something worse)
U(sick, find cure) = -1 (still need tests, drugs, etc.)
So with these numbers I just made up, it is better for the doctor to tell you that there is likely nothing to worry about. And you can be pretty sure that in real life, people have done this calculation. Of course in real life there are many more possible actions, such as waiting for a week to see if the headaches go away, which they will likely do if there was nothing wrong. And that is what a doctor will actually tell you to do.
The doctor believed that the girl didn’t have any serious disease because most people who have headaches do not. How exactly is that not an appeal to majority?
If the doctor’s hypothesis anticipates that the girl is healthy in spite of having headaches, then the easiest way to falsify it is to ask what sign or symptom would indicate a life-threatening disease. Would you want your doctor to wait a week or so to test his hypothesis, if you had headaches that could be caused by a brain tumor?
But then again, they don’t teach falsification in med school.
iI would be an appeal to majority if and only if he was appealing to the fact that most people thought she didn’t have a serious disease. Instead, he was just appealing to base rates, which is totally reasonable.
As somebody who’s had to deal with doctors because of a plethora of diseases, I must say you’re absolutely right. (I also shadowed a few and am considering applying to med school.)
I don’t remember what this concept is called, but basically it posits that “one should look for horses, not zebras” and is part of medical education. That is, a doctor should assume that the symptoms a patient has are caused by a common disease rather than by a rare one. So most doctors, thanks to their confirmation bias, dismiss any symptoms that don’t fit the common disease diagnosis. (A girl from my town went to her physician because she complained of headaches. The good doctor said that she’s got nothing to worry about and recommended more rest and relaxation. It turned out that the girl had a brain tumor which was discovered when she was autopsied. The good doctor is still practicing. Would this gross example of irrationality be tolerated in other professions? I think not.)
Most doctors are not so rational because of the way their education is structured: becoming a doctor isn’t so much about reasoning but memorizing heaps of information ad verbatim. It appears that they are prone to spew curiosity-stoppers when confronted with diseases.
soren, please don’t take this the wrong way, but based on what I’ve seen you post so far, you are not a strong enough rationalist to say things like this yet. You are using your existing knowledge of biases to justify your other biases, and this is dangerous.
Doctors have a limited amount of time and other resources. Any time and other resources they put into considering the possibility that a patient has a rare disease is time and other resources they can’t put into treating their other patients with common diseases. In the absence of a certain threshold of evidence suggesting it’s time to consider a rare disease (with a large space of possible rare diseases, most of the work you need to do goes into getting enough evidence to bring a given rare disease to your attention at all), it is absolutely completely rational to assume that patients have common diseases in general. .
None taken, but how can you assess my level of rationality? When will I be enough rationalist to say things like that?
What bias did I use to justify another bias?
Again, testing a hypothesis when somebody’s life is at stake is, I think, paramount to being a good doctor. What’s the threshold of evidence a doctor should reckon?
What gross example of irrationality? The vast majority of people with headaches don’t have anything to worry about.
The question is whether “people with headaches” is the right reference class. If the headache is unusually severe or persistent, it makes sense to look deeper. Also, a doctor can ask for details about the headache before prescribing the expensive tests.
More precisely, the question is whether or not the right reference class is one in which cancer tests are worth while. The headaches would have to be very unusually severe to get enough evidence.
It was never mentioned whether or not the doctor asked for details. It’s also possible that none of those reference classes are worth looking into, and she’d need headaches and something else.
Cancer isn’t the only solvable problem which could get ignored if headaches are handled as a minor problem which will go away on their own.
Yeah, but the other ones also get ignored if you assume it’s cancer. To my knowledge, they have to be individually tested for. If none is worth testing for individually, it’s best to ignore the headaches.
“The vast majority” != “All”. What’s wrong with “you most likely have nothing to worry about, but I suggest doing this exam the off-chance that you do”? You’ve got to multiply the probability by the disutility, and the result can be large enough to worry about even if the probability is small. (Yes, down that way Pascal’s mugging lies, but still.)
EDIT: Okay, given the replies to this comment I’m going to Aumann my estimate of the cost of tests for rare diseases upwards by a couple of orders of magnitude. Retracted.
I’m pretty sure that, in this case, the probability is smaller than the disutility is large. Getting tested for cancer doesn’t come cheap.
Doctors get taught to practice evidence-based medicine. There’s a lack of clinical trials that show that you can increase life span by routinically giving people who suffer from headaches brain scans.
If I understand the argument right, then doctors are basically irrational because the favor empirical results from trials over trying to think through the problem on a intellectual level?
MONNAY.
The question is, whose utility?
There’s also the problem of false positives. Treatments for rare diseases are often expensive and/or carry serious side effects.
I was thinking of diagnostics, not treatment, though from DanielLC’s reply I guess I had underestimated the cost of that, too.
If you start diagnosing and find false positives than you are usually going to treat them.
If you had a headache, would you want your doctor to find out its cause or would you be satisfied if he told you that ‘the vast majority of people with headaches don’t have anything to worry about’ and sent you home straight away?
This ‘majority’ argument is a fallacious argument. Therefore it’s wrong, much wrong.
Edit #1: If you downvote me, I’d like to get some feedback as to why you’re doing that.
The argument by majority fallacy means arguing that something is true because many people believe it. In the example of the headaches, the argument was that it was likely true because it is true for most people.
What you would want your doctor to do is take the action that maximizes your expected utility,
E[U(action)]. Let’s simplify a bit, and say that action can be either “do nothing” or “find cause”. Then the utilities could be something like:Then:
So with these numbers I just made up, it is better for the doctor to tell you that there is likely nothing to worry about. And you can be pretty sure that in real life, people have done this calculation. Of course in real life there are many more possible actions, such as waiting for a week to see if the headaches go away, which they will likely do if there was nothing wrong. And that is what a doctor will actually tell you to do.
The doctor believed that the girl didn’t have any serious disease because most people who have headaches do not. How exactly is that not an appeal to majority?
If the doctor’s hypothesis anticipates that the girl is healthy in spite of having headaches, then the easiest way to falsify it is to ask what sign or symptom would indicate a life-threatening disease. Would you want your doctor to wait a week or so to test his hypothesis, if you had headaches that could be caused by a brain tumor?
But then again, they don’t teach falsification in med school.
iI would be an appeal to majority if and only if he was appealing to the fact that most people thought she didn’t have a serious disease. Instead, he was just appealing to base rates, which is totally reasonable.