It's only fair to share…

Recently, a paper written by researchers at John Hopkins asserted that medical error was the third leading cause of the death in the United States.  This received – as you might imagine – considerable coverage in the media.  The researchers proposed that death certificates should include a qualifier or indicator that medical error was linked to the death, if in fact it was, so that better statistics could be obtained. Here is a figure from that article.



I certainly can’t argue with the fact that we do not have good data about how frequently medical error occurs, or how frequently such errors contribute to serious disability or death.  However, the paper also offered a case illustration which did show how un-illuminating the death certificate is, but in my view, did not actually demonstrate a preventable error.  I’ll share the following ideas* about error reporting and error prevention:

Human errors cannot be completely eradicated, and most errors are probably of no consequence.  However, it is precisely for that reason that we must improve our abilities to study errors – there is much to learn from the thousands of errors that do not ultimately harm patients.  How were they detected?  What contributed to them happening in the first place?  Clearly, we must do better in this area.  And, as we deal with the ever-exploding demands of the electronic health record and complex billing requirements, we simply must insist that the technology and bureaucracy of the business of healthcare work for doctors, so we can do the best for the patient, instead of being distracted by the paperwork and confused by the computer system.

Which errors should be counted? All of them, of course.  This causes intense debate among clinicians when the patients in question are critically ill or injured, because we know we cannot save everyone, and moreover, that not all bad endings are the result of errors.   So why include these?  Because it can be true that a patient is critically ill, and also true that an error killed the patient.

A world-renowned patient safety expert describes the following vignette to illustrate:  “Imagine you are in a car crash and you are very severely injured.  The Life Flight helicopter picks you up to take you to the trauma center.  You are in very bad shape and likely to die.  In the middle of the air transport your gurney falls out of the chopper. You plummet 2,000 feet, crash into the ground, and you die.  We obviously cannot say  “Oh, he was likely to die anyway; can’t count that.”

Who should decide whether errors were made? And who should decide whether they were preventable? The authors propose that the coroner or medical examiner could be designated for this purpose.  That is very tricky.  For one thing, medical decisions are complex and nuanced, and there is often no single “right” or “best” decision.  Medical care is a series of many decisions, and data unfolds in a sequence – much of it the result of tests or interventions that are performed.  But the medical examiner at the autopsy has the benefit of knowing the outcome, and therefore is subject to hindsight bias.  Things that were not at all obvious at the time the situation was evolving have a way of seeming very obvious when all is said and done.  We hold an unrealistic standard of what someone “should have known” or “should have done”.

Secondly, medical interventions carry some inherent amount of risk.  Any time you have a procedure, no matter how small, there is risk of bleeding, infection, and other problems.  These are not errors (though negligent sterile technique would be).  These are simply unfortunate adverse outcomes – complications of interfacing your imperfect body with an imperfect science.

Third, it is very popular to compare healthcare to aviation, but as any aviation professional will attest, planes with mechanical problems are grounded, and not permitted to fly until they are fixed.  Planes are a completely understood entity, with a full instruction manual describing their every feature.  Humans are neither without health problems, nor with an accompanying manual, and they “fly” all the time – right into our offices, emergency departments, clinics, and operating rooms.

Is this a problem, or is it progress? One contrary view is that everyone dies from something, and therefore, if physicians are doing the best job at curing all curable conditions, the only remaining way to die will be by way of medical error.  Therefore, seeing medical error rank highly on the list is a sign of progress!  While that was relayed to me at least partially in jest, it is true that all of us are going to die, and most of us are going to die in proximity to healthcare.  So figuring out which deaths represent the natural course of very serious diseases and which ones were caused by preventable errors is very difficult.

What do you think?  How can we capture more information about errors – even those that don’t harm patients – so we can learn how to prevent them? How can we make unbiased assessments about whether bad outcomes are due to error, and whether those errors were preventable?