It’s been a good so far as finding things that I want to post about. Case in point is this news story. It seems that doctors have long advised women suffering from otosclerosis, a degenerative disease that causes hearing loss, to avoid becoming pregnant. It seems that pregnancy would heighten the risk of increased hearing loss, and might even lead to complete deafness.
But it seems that this is simply not true. A doctor from Ohio not only conducted a study to disprove the notion, but he also researched the literature and found that the original source for this belief was a 1939 seminar conducted by German doctors. The Nazis used the idea to promote racial purity.
What caused our perceptive physician to question prevailing medical opinion? He started a teaching job in a foreign country, and he noticed that women there with the disease who had given birth to many children didn’t seem to suffer any greater hearing loss than those with few children. That foreign country was Israel.
Trust the Israelis to distrust advice given by a Nazi doctor.
There was a study done some time ago that determined that the majority of clinical practices were not based on empirical science but came from medical tradition.
This problem arises from several sources:
(1) Performing rigorous scientific studies is expensive and time consuming. We only have the resources to study a subset of all medical problems in detail. For example, we can easily study the effects of one particular drug but it gets exponentially expensive to test the effects of combinations of drugs. Most people who are taking more than one kind of medication are in fact using an untested combination.
(2) Clinical medicine works in real-time. Doctors must make the best decision they can at the moment of illness, they can’t wait for rigorous scientific results. If, in their collective experience, something seems to have worked in the past then they go with it until they get better information.
(3) Each clinical case is unique. That is why medicine is an art. Even with well tested methods a great deal of variation exist in individual patients outcomes. This is also why rigorous scientific testing is so hard.
In the case of otosclerosis, the advise to avoid pregnancy served to confirm the false correlation. Most women probably followed the advise not to have children which reduced the pool of women who had children without harm. Doctors saw few successful pregnancies so the ones they did see they could put down to statistical luck. Preventing women with the condition from getting pregnant caused the very loss of data that would have allowed doctors to see that the correlation wasn’t valid. The Doctor in the story began to suspect the correlation wasn’t valid only by observing a population of women who had children anyway for religious regions
One solution to this problem could be data mining of medical records. If we could get enough medical records all in one effective database, researchers could just go fishing for correlations that formal studies or first hand experience would miss. There have already been some successes in this area and it might be prudent to try to figure out how to expand this access without creating a logistical, legal and privacy nightmare.