Thursday, December 15, 2011

When Care is NOT Worth it Even if the End is Life

Oh dear.  I'm a fan of Dr. Peter Bach.  We trained under some of the same mentors.  But....

In this article When Care Is Worth It, Even if End Is Death, he makes so many logical errors in thinking that I just can't let it pass tightlipped.

He describes the case of a middle aged man whom he cared for some 20 years ago who presented to the ER in extremis (about to die) from what turned out to be mesenteric ischemia (some of the bowel is dying for lack of blood flow.)  His supervisor mused that the man was hours from death.  After surgical treatment and a two week ICU stay, the patient walked out of the hospital alive, and Dr. Bach uses the case to attempt to illustrate the point that treating patients near death is not a waste of time, effort, and money.  Allow me to summarize the many errors in his logic.

Firstly, he was victim of the representativeness heuristic and neglect of (or in this case, ignorance of) base rates.  The man looked like he was going to die, so they assumed he would die - that's representativeness (See:  Thinking Fast and Slow).  Later in the article, he gives the statistics (which I have not bothered to confirm, that's not the point here, but interested readers may wish to confirm or refute) showing that the survival from mesenteric ischemia is 87%.  If he had used that base rate of survival to make his estimation when he saw the patient, and ignored everything else, he would have been more accurate at predicting the outcome.  But probably, as an intern, he did not know the base rate or how to use it, so he just went by his intuitions, smitten by the chaotic physiologic mayhem unfolding before him.

Secondly, he failed to recognize the importance of whether or not there is a correctable problem or diagnosis.  Mesenteric ischemia is surgically correctable.  That makes the prognosis much less grave than other conditions for which there is no cure, such as metastatic cancer, advanced dementia, massive hemorrhagic stroke, etc.  It's like the difference between ventricular fibrillation cardiac arrest (treatable often with angioplasty if the cause is acute coronary syndrome) and asystolic cardiac arrest (usually untreatable - a final common pathway to death from other causes). A corollary to this is that prognostication should not be undertaken before a confident diagnosis is made.  Only then can you know if what you're dealing with is correctable.  (I intentionally use "correctable" here rather than "treatable".  They are not snyonyms.)

Thirdly, he judges decisions and prognostications by their outcomes rather than by the thought process that formulated them.  That the man lived does NOT prove that the judgment that he would die was wrong (indeed it was, but not for the right reasons).  Had he died, Bach and his supervisor's prediction of death would have still been flawed, for the reasons of representativeness and base rate neglect outlined above!  The judgement of a decision's rectitude is properly based on process not outcome, on what was known and knowable at the time of the decision and whether that information was properly and rationally incorporated into the decision framework.  Think Iraq and WMD in 2003.

Fourthly, the title of his article can be re-engineered as the title of this post after the backwards logic is set in forward motion.  And it can be much more meaningfully applied that way.  Allow me to use my own case to illustrate the point.

An 89 year-old woman with mild dementia comes to the hospital with pneumonia from a nursing home where she receives assistance for most activities of daily living.  The pneumonia is presumed to be due to aspiration due to airway incompetence from generalized debility and dementia (read: old age).  She is intubated for respiratory failure and extubated on day 6.  Within 24 hours she has recrudescent respiratory failure, which continues unabated despite 48 hours of non-invasive positive pressure ventilation.  She is weak and not protecting her airway, however, she is not in extremis or "hours from death".  What is her prognosis?

In this case, we can employ a proper decision framework to analyze the likely outcomes.  We know that an 89 year-old patient who has poor functional status and dementia, who is weak and not protecting her airway, who has failed extubation and who is 9 days into a hospitalization for aspiration pneumonia is unlikely to be weanable from mechanical ventilation in the short term.  This is a prior probability, based on decades of experience with similar patients.  We also know another prior probability: such a patient, if she receives a tracheostomy and goes to a long-term acute care center (LTAC), has about a 25% chance of being alive and breathing independently at 12 months.  Given specifics of her case, the posterior probability is likely much lower. (For references, see: Cox_2009_CCM; Carson_Bach_1999_AJRCCM; Chelluri_2004_CCM; Feng_Manthous_2009_Chest; Engoren_2004_Chest.)

We should also ascertain whether we have a correctable problem.  In this case we can formulate a problem list as a population of trees, or we can back up so we can see the forest.  And the fact here is that pneumonia, respiratory failure, aspiration, weakness, failure to protect the airway, poor swallowing, malnutrition, etc. are downstream results of a proximate problem - poor functional status and frailty associated with the natural aging process.  And that is not a correctable problem.  Treatable, perhaps, but not correctable.

Such may be a case when care is not worth it, even if the end is life.

1 comment:

  1. I feel I have to rebut the previous comments by saying "You are great".
    I have been following your Medical Evidence blog for some time and only just discovered this one. Your rational, scientific approach to the overinflated, biased claims so often seen in medical reporting is welcome, and was one of the driving forces behind my recent decision to start my own blog at doctorskeptic.blogspot.com.au
    Keep up the good work.

    ReplyDelete

Note: Only a member of this blog may post a comment.