Saturday, March 22, 2014

Antifragile but Exposed: A Framework for Understanding Disease that Can Improve Diagnostic Decision Making

Ruination IPA by Stone
It is said that the history reveals the diagnosis 90% of the time, but it is not stated why this is so.  Herein I will explain the logic behind this aphorism.

I used to tell residents that most of the time, the diagnosis is something common, or something related to something we already know the patient has.  In addition, when teaching history taking, I told the medical students to view history taking as an exercise in determining what the patient is exposed to in his or her environment, for these exposures weigh on the probabilities of potential diagnoses.  These principles explain the basis of the history aphorism.

Nassem Nicholas Taleb was right in his book Antifragile - evolution has made our species beyond robust - antifragile, a term which he coined.  We resist disease, we repair injury, we get stronger when exposed to stress (the antifragile principle).  But there are ailments that humans suffer for which evolution has not worked out a defense or a solution, or those which result because evolution is helping the same organisms which attack us become antifragile just as it selects us for antifragility.  This is why infectious maladies are at the top of the list for adult (and I suspect pediatric) internal medicine admissions.  Pneumonia, UTI, URI, etc.

Other problems evolution has not had a chance to work out commonly afflict us too, and we know their names.  Heart disease is the number one killer in the US.  I suspect, as an aside, that the cascades of causal factors that lead to heart disease make us stronger during the reproductive years, but work against us in senescence - because after we have reproduced, evolution cannot work on the problem anymore.  I also suspect that evolution favored inflammation, robust immunity, and clotting because infection , injury and bleeding were major threats during reproductive years for most of our evolutionary history.  Thus, maladies characterized by excessive inflammation, clotting, and exuberant immunological reactions plague us today.  A very interesting example of evolution solving an early problem and creating a later one is the notion that hemochromatosis was selected for in certain populations where there was a tendency for nutritional iron deficiency - and a byproduct of that is that carriers are now taller, as reported in the NEJM last year.

Cancer is another problem of senescence that evolution has not had a chance to work out.  In any case, if you want to know the common afflictions of humanity that evolution didn't have a chance to fix, just look at the most common causes of death in the US.  Other problems appear on the list:  COPD, AIDS, alcoholic cirrhosis - but this is because these diseases result from exposures that are recent in evolutionary history - alcohol, smoking, HIV, etc.  Which segues me to the next paradigm.

The history that we take seeks to identify exposures in the patient's environment which increase the probability of certain diseases.  The social history is concerned with substance abuse exposures, the surgical history with anatomical deviations, the medication history with pharmacological exposures and side effects, the occupational and travel histories with exposures in those realms, and so on.  In a similar vein, the history clues is into endemnicity, or endemic diseases, those which are not common overall, but which are common in certain locales as a result of an exposure environment that is enriched in specific ways.  In all cases, these exposures increase the probability of certain diseases.

The past medical history is especially important, and it's why I used to tell the residents that if it's not something common, it's likely to be related to something we already know the patient has.  If a patient presents with shock and they have dialysis dependent renal failure, obstructive shock which is generally rare rises on the differential because patients with renal failure are wont to develop pericarditis and tamponade.  Likewise, patients who are immunosuppressed for any reason are likely to get otherwise rare infections.  We are antifragile, but we are exposed.  (Genetic exposures, revealed by the family history, are likewise paramount.)

A corollary principle is related to Ochkam's Razor, and the independence of probabilities.  Ockham's principle of parsimony states states that the simplest explanation is most likely to be true.  So, if the choice is between one disease and complications or associated disease phenomena versus two distinct diseases, the simplest and most likely correct answer is the former.  Likewise, if two events are independent, the probability of both is the product of their independent probabilities.  One parachute fails one in a thousand times, the main and the reserve parachute fail one in a million times (0.001*0.001).  But not if the same guy packed both chutes and he was drunk.  The independence condition is not met in that case and the probability of both chutes failing to open is less than the product of the individual probabilities.  So if two things are happening at once, it is statistically more likely that they are linked than it is that they are independent events.  Indeed, I think this is the basis for what Ockham was thinking when he said that plurality shall not be postulated without necessity.

So from now on, when you are trying to determine a provisional diagnosis, think of common things that everybody gets, things that evolution has not had a chance to work out, and think of things that may result from exposures (and known diseases) that the patient has in his specific environment that increase the probability of other diagnoses.  Only when the work-up of these things fails to confirm these suspicions should you look for zebras like lymphangioleiomyomatosis or hemophagocytic lymphohistiocytosis (HLH).

1 comment:

  1. In this unknown problem case on the New York Times website 2/4/16, I used this framework to immediately solve the problem with a google search. I did not even know, as a pulmonologist that Cubicin is associated with eosinophilic pneumonia - but I knew that Cubicin was an exposure so all I had to do was google it, and like several other commenters, immediately figured it out. Even though it is a bit atypical for an eosinophilic pneumonia. Indeed, one commenter from Princeton, NJ, opined that if he could figure it out with a few mouse clicks, shame on the PCP and ER Doc in the case history.