Comparisons between aviation safety and patient safety should only go so far...

Are medical errors really the “third leading cause of death”?

There are many apt comparisons to be made between aviation and healthcare.  As a (lapsed) instrument-rated pilot, I may have some insight about the system in which planes function on a daily basis in the United States.  It is truly an impressive system which is (to paraphrase Don Berwick) well-designed to achieve the safety results that it does.  However, the kind of statement that some healthcare safety leaders make in which they compare the raw mortality rates of acute hospital care and airline fatal accident rates is not so apt.  Let me see if I can explain why. 

First, just to be clear, I have no illusions that we have a long way to go in optimizing the safety of medical care.  I am a vocal advocate for transparent safety culture and adequate resource allocation for patient, staff and family safety.  But, we don’t do anyone any favors by sensationalizing and mis-characterizing where the problems are. 

First, remember that in these aviation/medicine metaphors, airplanes are patients and “flying” is providing medical care. The thing is that the commercial aircraft that fly every day are required to regularly demonstrate that the performance of their critical systems meets or exceeds strict standards.  If a radio, compressor blade, hydraulic line, or pressurization fan is not operating within a very narrow performance range the simply plane does not fly.  With our fellow humans we of course do not have this luxury.  We don’t have too many replacement parts – and they have their own problems.  When we admit a 72 year old with heart failure, COPD and CRD we must do our best to keep them “flying” despite multiple marginally functioning critical systems. 

Imagine flying a plane in which engine power is only 50% of normal and control surfaces do not respond to input from the pilot reliably and where the radio is not always capable of receiving ground communications that are understandable to the pilots and crew.  Also, the fuel tank is leaking.  BUT you HAVE TO FLY it no matter how many systems are semi or non-functional.  At this point, you can’t simply follow the “standard procedures”, all of these accumulated problems make this plane a unique, untested system.  At some point, in the process of trying to improvise to keep this plane in the air, you may do something that results in an accident.  However, it was really, “an accident waiting to happen”.  Almost unavoidable for the brave souls who were trying their best to keep the plane flying.  This is what taking care of someone with critical illnesses is like.  It is NOT like flying a 737 that has had each and every critical system brought to 6-sigma levels of performance before it is allowed out of the hanger.

I don’t have access to the primary data that drives estimates of hospital mortality from adverse events, but if you actually reviewed the charts, how many of the “medical errors” occurred in very sick patients with multiple system disease?  Massive trauma?  A pain medicine overdose in a healthy 37 y/o in the hospital for a hysterectomy is not the same as a pain medicine overdose in a 78 year old with CAD and multiple myeloma who is admitted with a hip fracture.  We should continue to learn whatever we can from the praiseworthy culture of safety and performance of our airplanes and pilots. But let’s not take the metaphor where it doesn’t fit and let’s not assume that every very sick person who dies in a hospital from an adverse event is an example of a truly preventable death rather than dedicated clinicians trying their best to keep someone alive and eventually failing. 

ICU patient.  Every patient is a unique, uncharacterized system.  NOT the same as a 737. 

ICU patient.  Every patient is a unique, uncharacterized system.  NOT the same as a 737. 

Older Airline Cockpit.  EVERY one looks exactly the same.  Every control is in exactly the same place and responds exactly (within 1%) the same way to inputs. 

Older Airline Cockpit.  EVERY one looks exactly the same.  Every control is in exactly the same place and responds exactly (within 1%) the same way to inputs.