Tuesday, April 26, 2005
Math + Medicine = Confusion
"Why do so many people have trouble with the notion of probability and chance? Mathematicians chalk it up to innumeracy, the arithmetic equivalent of illiteracy. Simply put, people are uncomfortable with mathematical concepts like probability because they never learned them in the first place.
Innumeracy explains much of the public's confusion about the risks of various drugs and medical treatments. But not all of it. In a classic 1966 study, a group of subjects was told that a man had parked his car on a hill and that the car had rolled back into a hydrant after the man had left. The subjects were sympathetic to the man.
But a second group of subjects, told that the car had rolled into another person after the man walked away, held him responsible, even though the cause was the same.
People might chalk up a minor mishap to chance, but they are reluctant to blame a serious event on bad luck. Someone or something has to be held responsible."
I can say something relating this to medical malpractice, but I won't.
The article states that dealing with probability and chance does not come naturally. People like absolute yes/no answers - and medicine often does not come prepackaged that way. One of my favorite sayings is that "the only absolute in medicine is that there are no absolutes".
I often watch law shows on TV where the physician witness is being hositally cross-examined: "Is it possible that [xxx] can happen?". The answer is always yes! Anything is possible. It's also possible that I can win the lottery or find the cure for cancer, but the chances are slim. A better question that lawyers should consider: "Is it probable that [xxx] can happen?" Of course, that makes too much sense, and thus, won't happen. (via shrinkette)
"Why do so many people have trouble with the notion of probability and chance? Mathematicians chalk it up to innumeracy, the arithmetic equivalent of illiteracy. Simply put, people are uncomfortable with mathematical concepts like probability because they never learned them in the first place.
Innumeracy explains much of the public's confusion about the risks of various drugs and medical treatments. But not all of it. In a classic 1966 study, a group of subjects was told that a man had parked his car on a hill and that the car had rolled back into a hydrant after the man had left. The subjects were sympathetic to the man.
But a second group of subjects, told that the car had rolled into another person after the man walked away, held him responsible, even though the cause was the same.
People might chalk up a minor mishap to chance, but they are reluctant to blame a serious event on bad luck. Someone or something has to be held responsible."
I can say something relating this to medical malpractice, but I won't.
The article states that dealing with probability and chance does not come naturally. People like absolute yes/no answers - and medicine often does not come prepackaged that way. One of my favorite sayings is that "the only absolute in medicine is that there are no absolutes".
I often watch law shows on TV where the physician witness is being hositally cross-examined: "Is it possible that [xxx] can happen?". The answer is always yes! Anything is possible. It's also possible that I can win the lottery or find the cure for cancer, but the chances are slim. A better question that lawyers should consider: "Is it probable that [xxx] can happen?" Of course, that makes too much sense, and thus, won't happen. (via shrinkette)
Posted by Kevin |
link
| Comments (6)
| 
Comments:
Dear Kevin,
NYtimes offers deeplinks voor weblogs. Then your visitors don't have to register/login to view the articles.
F.e. : Linkgenerator
NYtimes offers deeplinks voor weblogs. Then your visitors don't have to register/login to view the articles.
F.e. : Linkgenerator
The test for whether a person can be held responsible for harm resulting from negligence isn't where the harm was was "possible", nor whether the harm was "probable".
The test is whether the harm is reasonably forseeable, a higher standard than possible and lower standard than "probable". A harm can be uncommon and be reasonably forseeable. A harm can be possible and yet not reasonaby forseeable.
The test is whether the harm is reasonably forseeable, a higher standard than possible and lower standard than "probable". A harm can be uncommon and be reasonably forseeable. A harm can be possible and yet not reasonaby forseeable.
# posted by 
: 9:31 AM
: 9:31 AM
Kevin,
The ambiguity continues even with the suggestion of whether an outcome is "probable." What is the threshhold for the probability? The list of possible outcomes from treatment protocols is often endless, and no one outcome could be considered "probable" in the sense of being over 50% likely. I think this is an area where even people familiar with statistics have a hard time. When a patient asks what diagnosis is "most likely" and the doctor says cancer, they may not understand that the chance of cancer is only 25%, but that is significantly higher than the next option at 10%. But there is still a 75% chance that it is not cancer, so although this is "most likely" it is still not "probable." (Obviously we need better than 25% certainty to render a diagnosis, but the point is patient confusion.) Short of making a list with possible outcomes and putting approximate percentages next to each one, you can't accurately convey the probabilities in your mind to the patient. Instead we use categorical terms which in my experience patients don't always view the same as the physicians. I think the reason is that physicians spend years speaking this language of terms expressing probabilities, and for someone who is not familiar with the dialect it is easily misinterpreted. Maybe we do need to hand patients lists with approximate probabilities on them.
The ambiguity continues even with the suggestion of whether an outcome is "probable." What is the threshhold for the probability? The list of possible outcomes from treatment protocols is often endless, and no one outcome could be considered "probable" in the sense of being over 50% likely. I think this is an area where even people familiar with statistics have a hard time. When a patient asks what diagnosis is "most likely" and the doctor says cancer, they may not understand that the chance of cancer is only 25%, but that is significantly higher than the next option at 10%. But there is still a 75% chance that it is not cancer, so although this is "most likely" it is still not "probable." (Obviously we need better than 25% certainty to render a diagnosis, but the point is patient confusion.) Short of making a list with possible outcomes and putting approximate percentages next to each one, you can't accurately convey the probabilities in your mind to the patient. Instead we use categorical terms which in my experience patients don't always view the same as the physicians. I think the reason is that physicians spend years speaking this language of terms expressing probabilities, and for someone who is not familiar with the dialect it is easily misinterpreted. Maybe we do need to hand patients lists with approximate probabilities on them.
# posted by : 12:58 PM
: 12:58 PM
As a statistician, I'm always dismayed by doctor's lack of understanding about probability and causation.
Kevin is a case in point. The example he gives about the fire hydrant has nothing to do with understanding probability, but with the rubrics we use to assign blame.
For whatever reason, people who engage in risky behaviors but injure no one are viewed differently from those who engage in risky behaviors but do. This is not necessarily irrational. After all, if Adolf Hitler just sat in his room as an evil man who wrote bad political theory, would we judge him differently?
Moreover, Kevin is wrong to say that things just happen by chance. EVERYTHING has a cause. (remember, this is science). The question is what causes we can recognize and with what degree of certainty.
Kevin is a case in point. The example he gives about the fire hydrant has nothing to do with understanding probability, but with the rubrics we use to assign blame.
For whatever reason, people who engage in risky behaviors but injure no one are viewed differently from those who engage in risky behaviors but do. This is not necessarily irrational. After all, if Adolf Hitler just sat in his room as an evil man who wrote bad political theory, would we judge him differently?
Moreover, Kevin is wrong to say that things just happen by chance. EVERYTHING has a cause. (remember, this is science). The question is what causes we can recognize and with what degree of certainty.
# posted by : 8:37 PM
: 8:37 PM
Well, if that's how they do it on TV, then that must be how it's done in real life.
The fact that a TV show is the source of your knowledge speaks volumes. I watched ER last night, and I think I'm ready to perform surgery. Are you hiring?
Post a Comment
The fact that a TV show is the source of your knowledge speaks volumes. I watched ER last night, and I think I'm ready to perform surgery. Are you hiring?
