Monday, 8 February 2016

Hunting zebras in Texas or why doctors order so many tests


Bayes’s theorem estimates the probability of an event occurring if another one occurred before. If we hear thunder, it’s very likely that soon it will start to rain. However, biomedicine is more interested by the inverse probabilistic formulation because often when an observation is made we look for the cause: "If, for example, you find yourself on a ranch in Texas and hear the sound of trotting horses, what is the probability that a herd of zebras will approach? ”Dr. Saurabh Jha, radiologist, blogger and contributor to the BMJ, says that we hunted zebras in Texas for long enough. "The clinical reasoning of many doctors today - explains Dr. Jha - forgot the Bayesian thinking and embraced instead the thesis that it’s better to manage a false positive than a false negative."

When, in the second decade of the last century, he was asked why he wanted to climb Everest, George Mallory, an English mountaineer, replied that it’s because the mountain was there. According to Dr. Jha, many doctors feel the same as Mallory: they ask for tests because they can. In his article in the BMJ, the radiologist explained the case of a patient of Dr. Watson, a well read doctor emerged from a refined education who pays a lot of attention to detail but lacks the deductive ability of the famous Sherlock Holmes’ assistant. If you have a chance to read the letter, you will see how for the study of this case, nothing seems elementary to Dr. Watson, to the extent that for every horse he sees an opportunity for a zebra hunt. "Zebras are intellectually exciting," Dr. Jha ends up saying.


In the blog "Science-Based Medicine," Dr. Harriet Hall, includes an article published in the New England Journal of Medicine 35 years ago, where there’s a list of reasons that explain why doctors order tests in excess, and I thought it appropriate to copy it here, after the Dr. Jha’s analogy of zebras.
  1. To complete the study. It being understood that "complete" is a concept with confusing but generous limits. 
  2. They say it must be done. Without ever clarifying who says so.
  3. We’ll have problems if we don’t do it. It’s the proffered excuse for overcautious medical residents when they have to prepare a case for clinical session.
  4. Better ask for all the possible tests in case we miss something. Usual excuse in the large university hospitals.
  5. Whilst the patient is hospitalised, we should take the opportunity and ask for certain complex tests.
  6. A practice of the defensive medicine. Very common, especially with accidents that end up in court.
  7. We must complete the protocol required by the ongoing study.
  8. If it was my mother I'd ask for that test just in case. "When doctors are emotionally involved with a patient, the reasons flee through the window" (sic).
  9. We should be sure that the patient doesn’t have it - the lack of Bayesian deductive thinking, as in the case of Dr. Watson’s zebras.
  10. The more we know the better.

Asking for too many tests is not just a matter of costs, it’s also a practice that has consequences due to complex statistical interpretation of certain results, the false positive and the risks inherent to some tests themselves. But even worse is the appearance of overdiagnosis, a phenomenon caused by findings that, although true, would never have endangered the health or the life of those affected. Good time, then, to recall the Bayesian practice of medicine or, in other words, the return of Dr. Watson’s deductive ability, not the zebras.


Jordi Varela

Editor

No comments:

Post a Comment