Tuesday, June 2, 2015

Evolution Based Medicine: A Philosophical Framework for Understanding Why Things Don't Work

An afternoon session at the ATS meeting this year about "de-adoption" of therapies which have been shown to be ineffective was very thought provoking and the contrasts between it and the morning session on ARDS are nothing less than ironic.   As I described in the prior post about the baby in the bathwater, physicians seem to have a hard time de-adopting therapies.  Ask your colleagues at the next division conference if you should abandon hypothermia after cardiac arrest and rather just treat fever based on the TTM trial and the recent pediatric trial, and see what the response is.  Or, suggest that hyperglycemia (at any level in non-diabetic patients) in the ICU be observed rather than treated.  Or float the idea to your surgical colleagues that antibiotics be curtailed after four days in complicated intraabdominal infection, and see how quickly you are ushered out of the SICU.  Tell your dietition that you're going to begin intentionally underfeeding patients, or not feeding them at all and see what s/he say(s).  Propose that you discard sepsis resuscitation bundles, etc.  We have a hard time de-adopting.  We want to take what we have learned about physiology and pharmacology and apply it, to usurp control of and modify biological processes that we think we understand. We (especially in critical care) are interventionists at heart.

The irony occurred at ATS because in the morning session, we were told that there is incontrovertible (uncontroverted may have been a better word) evidence for the efficacy of prone positioning in ARDS (interestingly, one of the only putative therapies for ARDS that the ARDSnet investigators never trialed), and it was strongly suggested that we begin using esophageal manometry to titrate PEEP in ARDS.  So, in the morning, we are admonished to adopt, and in the afternoon we are chided to de-adopt a host of therapies.  Is this the inevitable cycle in critical care and medical therapeutics?  A headlong rush to adopt, then an uphill battle to de-adopt?

As I described in the post about stochastic dominance of the null hypothesis, it is my observation that discovery of associations is far more common than discovery of causal pathways in most lines of biological inquiry.  The only way to tell if something is causal rather than an epiphenomenon or association is to control it as an independent variable in a randomized controlled trial or experiment.  It is here that cognitive dissonance creeps into our psyches:  manipulating a potential causative variable holds the hope of proving the efficacy of a treatment for disease.  This hope of proving the efficacy of a therapy then supplants the disinterested questioning of the hypothesis, and objectivity is lost.  We now want this therapy to work, and we will read between the lines and promote it on the basis of shaky or inconclusive evidence, and we will then resist and neglect discomfiting but more robust evidence as it later accrues.  This pattern has been repeated over and over in critical care (and probably other specialties) during the past 15 years.

Another problem is that we overestimate our understanding of biology, and underestimate our impotence in manipulating it to beneficial ends.  We know what we know (known knowns; platelets express XYZ receptor), but we do not know what we don't know (unknown unknowns; XYZ platelet receptor has dozens of ligands and scores of downstream pathways), nor do we know the scope of the unknown problem (how many redundant receptors and pathways there are, the ratio of known to unknown pathways).  We find associations, and we implicitly assume that we have uncovered causal pathways which are susceptible to our manipulations.  But when Max Perutz said "Evolution is a brilliant chemist" his awe of nature doubled as a prescient warning.

We tend to forget that for millions of years evolution has been working on the same diseases that we endeavor to understand and treat.  Consider streptococcus pneumoniae infection.  We may find therapies in addition to oxygen and intravenous fluids that bolster recovery, but humility is in order:  most of the antibiotics that we use for this and other infections were gifts from evolution, and our major contribution has been to isolate, purify, and infuse them.

So, we should ask:  could evolution not solve the problem of anorexia and catabolism in association with acute illness?  Are anorexia and catabolism harmful?  Is it so simple as to give enteral nutrition and one-up evolution?  Or should we consider that evolution has been working on this problem for millions of years, and probably anorexia and catabolism are part of an adaptive response in acute illness?  What of hyperglycemia in critical illness?  Fever in sepsis?  Hypotension not responsive to fluids in septic shock?   Should our first assumption be that these are targets for intervention, low hanging fruit that evolution would have solved if given another million years?  Or is it a better bet that they are evidence that evolved defensive processes are already hard at work on the problem?

I am not advocating nihilism, rather humility and restraint in the application of our "knowledge" and interpretation of our "evidence".  We should not underestimate the complexity of the biological processes we seek to understand and manipulate.  The simplest explanation for negative results in any trial is that our understanding of the diseases we are studying is incomplete and that we are trying to manipulate epiphenomena rather than causal factors.

If this is true, and I believe it often is, we would be wise to follow the advice of Voltaire :  "The art of medicine consists in amusing the patient while nature cures the disease." 

No comments:

Post a Comment