= 27d567bc2d48d9f40e9ccc20ea370b15)
moridinamael comments on Open thread, February 15-28, 2013 - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (345)
So, there are hundreds of diseases, genetic and otherwise, with an incidence of less than 1%. That means that the odds of you having any one of them are pretty low, but the odds of you having at least one of them are pretty good. The consequence of this is that you're less likely to be correctly diagnosed if you have one of these rare conditions, which again, you very well might. If you have a rare disorder whose symptoms include frequent headaches and eczema, doctors are likely to treat the headaches and the eczema separately, because, hey, it's pretty unlikely that you have that one really rare condition!
For example, I was diagnosed by several doctors with "allergies to everything" when I actually have a relatively rare condition, histamine intolerance; my brother was diagnosed by different doctors as having Celiac disease, severe anxiety, or ulcers, when he actually just had lactose intolerance, which is pretty common, and I still cannot understand how they systematically got that one wrong. In both cases, these repeated misdiagnoses led to years of unnecessary, significant suffering. In my brother's case, at one point they actually prescribed him drugs with significant negative side effects which did nothing to alter his lactose intolerance.
I don't intend to come off as bitter, although I suppose I am. My intent is rather to discuss strategies for avoiding this type of systematic misdiagnosis of rare conditions. This line of thought seems like a strong argument in favor of the eventual role of Watson-like AIs as medical diagnostic assistants. A quick Googling indicates that the medical establishment is at least aware of the need to confront the under-diagnosis of rare diseases, but I'm not seeing a lot of concrete policies. For the present time, I don't know what strategy a non-medically-trained individual should pursue, especially if the "experts" are all telling you that your watery eyes mean you have have hay fever when you really have some treatable congenital eye disease.
The odds of you having any particular disease is not independent of your odds of having other diseases.
Also it depends on how much less than 1% the incidences are.
Self experimentation. If the doctor prescribes something for you, test numerically whether it helps you to improve.
If you suffer from allergies it makes sense to systemmatically check through self experimentation whether your condition improves by removing varies substances from your diet.
It doesn't hurt to use a symptom checker like http://symptoms.webmd.com/#./introView to get a list of more possible diagnoses.
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.
It is my impression that there is already software out there that has a doctor put in a bunch of symptoms, and then outputs an ordered list of potential diagnoses (including rare ones). The main problem being that adoption is slow. Unfortunately, after 10 minutes of searching, I'm completely failing to find a reference, so who knows how well it works (I know I read about it in The Economist, but that's it).
There's also poking around online to find people with similar symptoms. . When I say I think his blog is good, I mean that he shows respect for human variation and for science-- I'm not sure that he's right about particular things.
Are you referring to curetogether.com, Nancy?
This graph illustrates clusters of related systems from that site: http://circos.ca/intro/genomic_data/
That's not the one I was thinking of, but it sounds promising.