The Myths We Tell Ourselves

My essays on how patient comments define how clinicians define how patients view them got me thinking about the other stories we tell in hospitals that explain things that may or may not be accurate.  That is, myths about hospitals don’t just arise out of patient perceptions or misperceptions, but about out of how clinicians define their performance and patient outcomes.  The more I worked with hospitals in various capacities, the more surprised I was that hospitals and clinicians have less appreciation for facts and data than one would expect.  To be clear, medicine is a science and science is based upon structured observation and testing, heavily reliant on data-collection.  Clearly clinicians are gathering important information, both from lab work, testing, and observation.  All these things drive the clinical next steps, ideally ending with successful outcomes.  Even so, there are blind spots that can color the reasonable questions and acceptable answers that go into this healing process.

One of the common popular myths among the patient population is that, once a therapy or pharmaceutical is established as a useful intervention, this new approach cascades through healthcare quickly; on Day One, a cure is created and on Day Two, everyone gets the cure.  This is simply not the case.  In some very specific cases, like a new high-impact treatment for a very specific disease (like a specific type of cancer), it can take as little as two years for something to become adopted as standard.  The conventional wisdom (or myth) is that it takes 10-17 years for a new therapy to take hold as common practice.  Even if you subtract the time in research and testing, it still takes three to seven years to become standard. 

This is not meant to impugn clinical staff.  It is rare for there to be a new intervention that is startlingly more effective than an old one.  Further, it is unreasonable for one to expect doctors and nurses to be up-to-date on every single development that floods in daily.  So, it takes time for new learning to filter down to the individual bedside and for the new intervention to supplant the old one.  This may feel new, but it isn’t.  Take the quote below by someone whose name will be familiar.  It was delivered at the Rush Medical College commencement, only 116 years ago.

“The sum total of medical knowledge is now so great and wide-spreading that it would be futile for one man to attempt to acquire, or for any one man to assume that he has, even a good working knowledge of any large part of the whole. The very necessities of the case are driving practitioners into cooperation.”

— Dr. William J. Mayo, June 15, 1910

That it takes a village to cure a patient is certainly understood by healthcare professionals, even if it may be underappreciated by the public.¹ What is less obvious, though, is how to square this fact with the realization that the more clinicians you have in the room providing care, the more confusion over what best-practice is.  Not only does this new information need to filter down to everyone, it also must overcome preexisting knowledge or heuristics of that team.  An unexpected roadblock to this is the body of myths associated with healthcare that is embraced by healthcare.  Let us explore one example.

No matter how you define it, sepsis is a killer.  It is confusing and difficult to spot, since it can be triggered by a bacterial, viral or fungal infection.  In reaction to the infection, one’s own body can, in essence, attack itself in a variety of different ways.  Part of the problem is that sepsis can be difficult to diagnose, since it can be triggered by infections in the respiratory, GI, or urinary tract, as well as the nervous system and even the skin.  It also has a host of generic symptoms, including some conflicting ones.  For example, one can have a high fever or low body temperature.  While modern science has been successful in bending the mortality curve for a lot of diseases, sepsis mortality rates have remained stubbornly consistent.²

What is telling about sepsis treatment, though, is that the strategies for addressing sepsis have been consistent for a few decades and have been established as best-practice.  CMS has sepsis best-practice compliance as one of its mandatory measures.  Since this essay is about why we are slow to embrace best-practices and not best practices themselves, I won’t define the specifics of the 1-hour, 3-hour, or 6-hour bundles for sepsis treatment.  I will simply use them as a shortcut for ‘established medical best practice.’  While I will call out some elements of these, my point is about acceptance of best-practice and not what the best-practice is.  There are better sources for sepsis protocols than this essay.

There are two myths that underscore my challenging journey on engaging the medical staff in efforts to improve performance on sepsis mortality.  About a decade ago, when I tried to spearhead clinical change in sepsis approach, I was confronted by these two storylines that prevented some from understanding the value of clinical best practice.³

Sepsis Context

The broadest impediment to understanding the best-practice were two firmly entrenched beliefs about what ‘really’ caused sepsis death.  Both can be summarized as NOT OUR FAULT.  The first was patient shaming.  OK, maybe shaming is too strong a word, but the focus was on all patient-related factors like age, gender, comorbidities, etc.  I remember attending a sepsis mortality report-out on the sepsis deaths a one hospital over the preceding period.  It was a recitation of a lot of numbers.  The person listed the gender and age breakdowns, the hospital length of stay, and a list of comorbidities most-often observed.  It was essentially a spreadsheet disguised as a meeting.  At the end of this torrent of data, the CNO asked me my thoughts.  I mentioned that none of this data meant anything to me, since I had no context.  I could not evaluate these numbers without getting the same numbers for patients who survived sepsis.  Saying that more men than women died of sepsis, or that most of those who died were above 75 years old was not useful unless it highlighted a variance.  For example, saying that 60% of the deaths were male means nothing, if 60% of those surviving sepsis were also male.  But, if 80% of the survivors were female, well, there is an interesting difference.  At this point the nurse doing the report-out looked at me and with a dismissive tone said, “this is a sepsis mortality meeting.”

Further, there was nothing in this summary that spoke of interventions.  What percentage of these patients got the 3-hour bundle?  How many of these diagnoses presented at admission versus were discovered during their stay?  Are there doctors with better or worse mortality numbers?  These questions elicited a blank stare.  It was clear that the dominant thought in this room was that sepsis mortality was like death and taxes.  After all, as stated above, sepsis is a bastard; it is difficult to identify and treat.  But if we define sepsis as just a fact of life, especially with our older and more medically fragile patients, we effectively remove ourselves from any responsibility. 

This inspired an interest in me to understand sepsis as well as the organization’s perception of sepsis.  I discovered more unhelpful “conventional wisdom” influencing decision-making.  For instance, the mothership hospital in the system had higher sepsis mortality rates than the other hospitals in the system.  This was explained away by the observation that all the other hospitals would ship off their hardcase patients (described to me by a nurse as having one foot in the grave and another on a banana peel), so they would die on the mothership’s watch and make the other hospital’s numbers look better. 

When I dove into the data, though, I discovered that none of this context mattered.  I build a database of all sepsis patients, both healed and dead.  I brought in several variables, from comorbidities to how they came to the hospital.  I discovered that most of the mythology regarding sepsis patients at the hospital was wrong. 

It turned out that sepsis patients who were transferred into the hospital had the best survival numbers.  This is not surprising because they came through the front door with a sepsis diagnosis.  The ones who had the worst survival rates were those who came into the hospital with another diagnosis and sepsis was only discovered later. 

Further, while comorbidities did pose some explanatory value, the number one variable that predicted survival was the use of the 3-hour bundle.  It alone had a 30+% improvement in survivability.  So, context was NOT as important as execution of best-practice medicine.   

Doctors be Doctoring

Further, the doctor assigned to the patient did matter in survival, though, not because they had a nose for sepsis or some magic touch.  It was because they generally were quicker to call for blood cultures and therefore were quicker with broad spectrum antibiotics.  They were more likely to push IV fluids.  Maybe they had a sixth sense when it came to sepsis, or maybe it was their years of experience, or maybe it was their default lab request in the face of confusing numbers.  In the end, though, it was clear that more consistent execution of the bundle would likely save lives.  Not all lives, but according to my math, if the bundle were consistently executed at the mothership hospital alone, about one more patient a week would go home.

But almost fifteen years after being called out as best-practice and about ten years after CMS made it a mandatory measure, the execution of the 3-hour bundle is still inconsistently utilized.  Five years after it was implemented as a mandatory measure, it is only being executed about 50% of the time.  Even now, depending on the study, it is being used only between 40% and 70% of the time.  So almost twenty years after its creation, somewhere between a third and half of all sepsis patients are not receiving best-practice treatment.

There are several reasons why it struggles with compliance.  Some are related to when the sepsis diagnosis is made and others are connected to the pass/fail nature of the measure, so patients getting some or even most of the bundle, but not all of the bundle, are still counted as failures.

But one of the biggest reasons for the lack of clear adoption is the healthcare professionals themselves.  Doctors often cite the 3-hour bundle measure as too constraining.  Doctors in opposition will reference the measure’s preference for a protocolized approach above clinical judgment and individualized care. 

This puts healthcare and patients in a delicate bind.

• On one hand, we have the time-worn notion that the best doctors trust their instincts.  They are humans and not robots.  It is their reputation at stake, and they will have to answer for any mortality inquiry, and they don’t want to be told how to practice.  Not by patients, not by best-practice alerts and not by decision-tree protocols. 
• Best practices are, by definition, best practices.  While they get tweaked over time (in fact the 3-hour bundle was replaced by a 1-hour bundle), they are what the science says is the course of action most likely to be successful.  There are plenty of protocols about when to set a broken leg or remove an inflamed appendix, so why do doctors get to make this call against best practice?

Administration and patients are loathe to tell a doctor what to do, but anxious about decisions that do not align with best medical evidence.  The evolution of medicine itself has put the physician at the top of the pyramid, so even in situations where they go against conventional medicine, they are not likely to suffer for their failures, so long as they can justify their decision-making. 

It is interesting that over one-hundred years ago, Dr. Mayo said that healthcare needed to be more collaborative, but we still default to a previous thought pattern and structure.  To be clear, this is not meant to be a hack on doctors.  I, personally, don’t want to be treated by BPA medicine.  I want my doctor to insert their own judgement.  But at the same time, I have an expectation that my doctor is reviewing other sources of information other than their gut.  This essay is not calling for machine-learning medicine.  It is simply illustrating how many of the myths that healthcare professionals embrace can actually be an impediment to progress and not a boon to progress.

¹Media doesn’t help with this, since most medical dramas, legal dramas, police procedurals, or action movies lean heavily on the “one man takes on the establishment” motif, where lives are saved and criminals punished because of one person, usually in the face of intense opposition.

²Some research indicates that sepsis is on the rise, even as mortality has dropped.  The problem with much of this research is that it does not acknowledge the increase in cases in explaining mortality.  As sepsis gets a greater focus, it gets more widely-diagnosed.  This captures some who may not actually have sepsis.  When they survive, they get counted in the numerator, even as they should never have been in the denominator.  Research that creates a standard definition of sepsis by attribute and not ICD-10 show essentially no growth in cases nor any change in mortality. 

³Please do not mistake this me vs they as some sort of braggadocio.  I am not casting myself as the chief protagonist in the trope I referenced in Endnote 1.