The Medical Errors Debate

A recent article published in the BMJ has caused a furor in the medical community claiming that medical error is the third leading cause of death in the US.  In the article Sensationalization of Medical Errors: Breaking Down the Data In Order to Improve Patient, the author makes a careful analysis of the data used to come to these conclusions. The methodology of the data collection makes the claims of the study grossly overstated, but does deliver an important message to the medical-industrial complex. 

Wikipedia states a medical error is an error that is a preventable adverse effect of care, whether or not it is evident or harmful to the patient. This might include an inaccurate or incomplete diagnosis or treatment of a disease, injury, syndrome, behavior, infection, or other ailment.

This is where the complications arise. Medical errors are and can be dangerously detrimental to certain patients but whether this is a cause and effect relationship can be very difficult to prove. Medical errors are contributory factors.

Medical errors run the gamut of poor communication, failure to diagnose in a timely manner (a complicated legal question), improper medications, not accessing the right data at the right time (a failure of interoperability of the modern Electronic Health Records, multiple intellectual and emotional biases of the providers, system errors (most common) and etc.

Whether a medical error directly cause a death, was contributory to what degree, and/or irrelevant would have to be carefully ascertained on a case by case basis. Making generalizations on death certificates where the data is frequently incorrect leads to suspect conclusions. 

The bottom line is that the medical community should take this article as a warning shot that there are significant problems in the system. Crying foul is not a solution. Moreover, the use of the word “cause” with respect to medical error is totally inappropriate.  We well know, too well, that “proximate cause” without significant other “contributory” factors, is necessary in a tort case. It is no different here.  When the disease is the underlying etiology, and the healthcare system does its best, but fails, as it naturally does now and then, what is the underlying cause?

Certainly minimizing what are termed “errors,” but should more properly be termed “imprecisions” or “flaws” is a goal to which all strive.  But as imperfect beings, subject to many flaws, a perfect medical world is not going to happen. Preventable means zero margin for the humanity under which we all labor. All we can do is our best to keep the imperfections minimized.

Fixing the present Electronic Health Record Systems to give accurate, clinically specific data would go a long way in solving some of the problems. Artificial intelligence giving specific warnings would give the provider an immediate second opinion that may help guide the proper course. Finally society has to come with grips that medical art and science is not perfect and never will be.

Acute on Chronic Electronic Health Records Dissatisfaction

The slideshow 6 Ways IT is Contributing to Healthcare Inefficiencies examines why there is general dissatisfaction with the Electronic Health Record.

1.      Work-flow issues
a.      When a healthcare IT system impedes workflow, it becomes a major hindrance to efficiency and satisfaction. An EHR should naturally and smoothly integrate into the time-honored workflow of a facility, not the other way around. 
b.      Therefore, changing workflow for the convenience of the electronic record, for billing, for data collection, while ignoring the working process of the providers is an obvious misstep.

2.      Training that never ends.
a.      When a product is not user-friendly and needs multiple classes to teach the provider to navigate through the mess, one has a built-in disaster.
b.      In such situations, the interface is not naturally intuitive, and most providers will have to relearn the entire process after a two-week vacation.
c.       One would think that the American Heart Association’s experience with poor retention after CPR classes would have demonstrated that easier is better.
d.      Lots of visual prompts work better than lots of training and re-training. CPR has been changed to “push on the chest”, defibrillate if possible, and call 911.
e.      Success rates improve with simplicity. Providers agree that most EHRs need to simplify or provide real-time guidance through prompts and orderly flow.

3.      Finding the Information   
a.      There is lots of relevant but buried data in the E HR. But it sits underneath layers in very separate silos. These take significant know-how and effort to access.
b.      It has been noted that finding a key nursing note can be so onerous that the provider gets burned out on the process and when writing WNL actually means “WE NEVER LOOKED”.

4.      Alert fatigue is a dangerous issue.
a.      Warnings and alerts especially in Computerized Provider Order Entry (CPOE) modules wear the provider out psychologically.
b.      Not uncommonly, risk adverse programming triggers these bells and whistles.  Workflow takes a serious hit when the alarms are always going off.

5.      Myths: Bigger is Better; more words are better than a few.
a.      Ask any provider to point out relevant information from a 17 page document and find out what otherwise obvious key data points are only recognized after a problem comes to light.
b.      The retrospectoscope is a more functional modifier of workflow when it is viewing just a compact presentation and report.

6.      Call for a National Data Base
a.      The lack of interoperability and lack of poor, difficult to obtain, communication remains a huge problem. One proffered solution is a National-Data-Base that every E HR vendor uses as its’ clinical data repository.
b.      In that way, any provider could see a problem list, test, treatments, hospitalization, and medications in a real-time basis. Key elements from every encounter would automatically flow into the data base. Pharmacies could also list all prescriptions filled with dates, times, refills etc. The provider would know if the patient is actually filling their prescriptions and what other providers are writing for that patient.
c.       Its implementation, at least in theory would enable the EH R vendor to concentrate on workflow, navigation, and simplification. 
d.      A national CPOE that could be locally modified according to clinical settings could massively improve efficiency.
e.      What a benefit it would be for all if there were common interfaces between EHRs . Providers would not have to learn multiple systems.  But, no, vendors tend to be in favor of non-standardized interfaces. 
f.        When is the last time you tried to pay for groceries with a card swipe that worked the same as the one you used at the store down the block. Never happened. Never will.
g.      If cross-system standardization a fundamental goal, a national data base and national CPOE effort might actually work. With agreed upon standards, across the healthcare IT industry, the money that was spent on meaningful could possibly have created some actual clinical value. But no. We need to have it different on the first floor than on the third; different on this street, than on the next; different in this city than in another.  Back to the drawing board.

The Nature of Diagnosis

One of my fundamental thoughts about the diagnostic model in medicine is illustrated by this case from the week's NEJM titled The Deficient Diagnosis.  The authors' comment in the closing paragraph on "circuitous diagnostic route" could (should) be applied to many other clinical situations. That is, most diagnostic routes, except the very obvious are indeed circuitous. And it is this aspect of medical decision refinement that should (or so I think) eliminate the term "errors" from the mix.

Instructive, especially to us, is the child's prior presentation, and the ultimate necessity of coming through the ED to find the true path to a successful diagnostic "route."

Were there errors prior to acquisition of a clinically actionable entity?  I don't see them that way.  Or maybe "hell no!"  This is the nature of medicine, in general.

Compare to someone who arrives with chest pain and acute ST-T elevation. A no-brainer, so to speak. But throw into the mix someone whose EKG and two enzyme sets are  normal, scheduled for a stress-test in 72 hours, and dies of an acute MI in 40 hours. Error? I don't see it that way.  Safe-route?  Not for that patient.

So diagnosis is not static, but dynamic. Even in the NEJM case, once action ability is reached, is that the "final" diagnosis ? (Of course "final diagnosis" is a term we use but applies in most cases to the moment of discharge, highlighting its inherent temporal nature) .   How many times have diagnostic pathways forced our otherwise non-compliant minds into submission to diagnosis' dynamism? Take hypoglycemia --> poor insulin administration --> (wait) Addison's disease --> (wait again) Multiple endocrine adenopathy.  Or chest pain --> acute MI --> aortic dissection.  And on and on.

Diagnosis is not, and should not ever be, a static entity.  What are called "errors" therefore are indeed (expected?) steps off the fastest route.  (That route of course is faster if the patient comes to the ED !!).  I think it would serve the diagnostic error community (if there is one) well, and the emergency medicine community (to which we belong) well if this were better encountered theoretically.  Hence, if done, a path to actionable diagnosis is best served when the time-frame to it is the shortest possible. And the ED is the best site for that !!