Monday, February 17, 2014

The (Un)Learning Process

The Gist:  Despite Harvard's medical school dean, Dr. Burwell, warning students, "Half of what we are going to teach you is wrong, and half of it is right. Our problem is that we don't know which half is which," medical education doesn't do a very good job preparing us for unlearning the wrong half.  Recent publications revealing the reversal of common medical practices demonstrate the need for this skill among clinicians [1-4]. Although unlearning is a part of the learning process, it is difficult, can feel personal, and may mirror the stages of grief [6].  Recognizing the obstacles to unlearning and arming ourselves with an enhanced awareness of ways in which we can overcome these barriers may mitigate our difficulty unlearning.

A study by Prasad et al examined studies testing the "standard of care" in publications in the NEJM and found that 40.2% of articles testing a practice that existed as "standard of care" reversed the practices, whereas 38% of articles testing the practice reaffirmed the standard [1].  The follow up commentary demonstrate that physicians should be open to unlearning and should treat studies with skepticism [2,3].  This is not to say that trainees and physicians should perform full, independent critical appraisals of each article they read but rather, beware of the barriers to scrutinizing our practice.  Through Free Open Access Medical Education (FOAM), one may keep a finger on the pulse of medical literature, through the primary literature and voyeurism into how others process the information.

The Case:  I read the Nielsen et al paper on therapeutic hypothermia (TH) in cardiac arrest in the New England Journal of Medicine with astonishment after the FOAM world exploded with chatter on the paper.  Just days before, the residency conference featured a presentation outlining the evidence behind TH, including the landmark papers by Bernard et al and by The Hypothermia After Cardiac Arrest Study Group.
This paper made me feel uncomfortable.  Medical school, Rosen's, and Tintinalli's made no mention of this aspect of medicine - how to unlearn something well ingrained. This is not to say I think we need to unlearn TH, but to re-examine the cherished practice to identify parts that create a difference in patient care. Less than one year into my residency, there have been countless interventions, diagnostic algorithms, or pathophysiologic explanations, taught to me as a medical student, that are now widely accepted as untrue or bad practice:  activated protein C in sepsis, the pathophysiology behind the hypoxic drive in COPD, the new left bundle branch block in a not so sick patient. I am not an expert, but these struggles with unlearning (or relearning) caused me to develop practices to keep my mental flexibility in check.

Stages of Unlearning and Practices to Overcome These Barriers
1.  Denial - We gravitate towards literature, dogma, experts, and practice patterns that reflect our own biases.  Thus, we may not come into contact with the information prompting a change in our practice or, alternatively, we may outright dismiss the assertion or data without a solid look.
Action:
  • Remind ourselves that publications may lead us astray if we are not careful [4,5].  
    • The first study makes the biggest splash but the second (third, fourth, etc) are the most important - don't overlook these [4].  After all, replication is the key to science, the foundation of medicine. 
    • Pay particular attention to "negative" studies, which are published far less frequently than positive studies.
    • Statistics may be easily manipulated.  This exists as data dredging, too much reliance on clinically meaningless statistics such as the p value, or misrepresentation of data or statistics in conclusions [4,5].   
  • Stay abreast of current literature which may include using others to help curate and manage the influx of evidence Examples in FOAM, such as use of Twitter, Emergency Medicine Literature of Note, Richard Lehman's Journal Scan, EM Nerd, can be found at the end of this post
2.  Anger - We tend to become defensive when our beliefs and practices come under scrutiny or are challenged - it can feel like we're being attacked.
Action:
  • Accept that we will do things that may, in hindsight, be called "wrong" even though this was not known at the time.
    • This takes a certain level of intellectual and emotional vulnerability.  Historically, this is not only an issue at the individual level, but also a barrier for institutions to unlearn a practice [6].  
  • Reassure ourselves that medical interventions do not define us as providers.  Recognize that evolution of medical practice involves a continuous state of learning and unlearning, despite our best research and efforts. Do not take individual medical interventions personally; rather, attempt to take our overall commitment to good, patient-centered medicine seriously.  
  • Beware of zealotry for a medical intervention.  If question over one of an intervention causes us unease, it may indicate that we are tied more to an idea than to patient care.  For example, I had a visceral response when prompted to re-work my thinking and use IVC ultrasound, in isolation, as a marker of fluid responsiveness.  It was scary to think how I had become so invested in a practice as a medical student.  This may be more difficult for individuals with industry ties or research embedded in one nidus of interest.
3.  Bargaining - We may engage in mental trade-offs with the evidence, using our own experiences/anecdotes or mental frameworks in an effort to trump the data driving the unlearning.  We may utter something along the lines of, "But, I saw Drug X work, in front of my very eyes," despite data demonstrating lack of efficacy.
Action:
  • Recognize the cognitive biases we have developed, our heuristics and anecdotes that may cause us to anchor in our learned practice patterns, and attempt to set these aside while we examine the data.  These cognitive short cuts and experiences certainly comprise part of one's clinical gestalt; yet, there may be times in which they act as a crutch.  Once we have a successful, miraculous clinical story of an intervention or diagnosis, we are at risk to become unconsciously attached to the steps that led us to the victorious save.  We have a similarly intense reaction when we experience a negative outcome.  Clinical experience and the best available evidence may work in concert but they also often have a tenuous relationship, filled with bargaining.
  • Similarly, recognize the familiar crutches of pathophysiology based answers or surrogate markers/endpoints.  When we dredge up complicated explanations that we were taught in medical school, with several extrapolations to fit the current thinking, this is frequently a marker that our understanding, to that point, was incomplete to begin with (and therefore, should be prime for re-examination).
4.  Depression - We may feel guilty or defeated by this unlearning process and assume that it translates into either a reflection on ourselves or demonstration of the futility of medicine, research, or evidence based medicine.
Action:
  • Recall that while we have a seemingly innate desire to intervene, sometimes even apparently harmless interventions carry risks.  
  • Reinforce that this is part of the process of practicing medicine, which is a dynamic environment teeming with uncertainty.
5.  Acceptance - Once we acquiesce to the notion that it's necessary to unlearn a practice or thought process in medicine, we are at risk of becoming complacent and failing to unlearn yet again, resulting in a vicious cycle.  Perhaps we should never be fully comfortable with the support for what we do, as that may allow us to become complacent and think that we understand when we don't. We may share in creating a dogma to replace the one we have just unlearned.
Action:  
  • Given the changing landscape of medical practice, it's likely wise to regard all of our practices with a skeptical and curious eye. This may allow us the mental flexibility to alter our practice when warranted. 
  • FOAM may play a role in allowing one voyeurism into how others are adopting or processing studies so we can identify interventions that may be nearing an inflection point for reversal.
    • The use of a filter and "sounding board" for literature does have limitations:
      • Selection bias - perhaps only papers in specific areas of interest/popularity are disseminated.  Furthermore, one may only encounter those disseminated by others with similar opinions, serving only to confirm our own biases.  
      • Premature adoption - As Ioannidis and Prasad warned, medical interventions undergo reversal frequently so aggressive adoption may expose patients to harms and providers to a bruised ego [1-3].  There's an elusive "sweet spot" between changing/unlearning a practice at a dangerously early time and too late. A previous post addresses the use of FOAM and local "authority" to change practice, and the cautions regarding changing practice discussed therein are also apply here.
      • Predigested information can make staying current on literature and ramblings in the medical community easy and practical for busy clinicians but deprives one of the importance of thinking for oneself - it's in this area of work and effort that learning actually takes place.  It's also tempting to simply nod behind others with similar opinions or adopt an expert's view, although this is certainly an issue outside of FOAM as well (e.g. local experts, clinical policies, etc).
References:
1. Prasad V, Vandross A, Toomey C, et al. A decade of reversal: an analysis of 146 contradicted medical practices. Mayo Clin Proc. 2013;88(8):790–8.
2.  Ioannidis JP a. How many contemporary medical practices are worse than doing nothing or doing less? Mayo Clin Proc. 2013;88(8):779–81.
3.  Vinay Prasad V, Ioannidis JP.  Evidence-based de-implementation for contradicted, unproven, and aspiring healthcare practices.  Implement Sci. 2014; 9: 1.
4.  Ioannidis JP a. Why most published research findings are falsePLoS Med. 2005;2(8):e124. 
5.  Nuzzo R.  Scientific method: Statistical errors Nature 506, 150–152 (13 February 2014) doi:10.1038/506150a 
6.  Rushmer R, Davies H. Unlearning in health care.  Qual Saf Health Care. 2004 December; 13(Suppl 2): ii10–ii15.

1 comment: