Sunday, February 24, 2013

How Big Data's Fueling Complacency

"What's the most important finding on this chest x-ray?"

There he was, standing before 5 ICU residents, each peering at a chest film on displayed on the over-sized computer screen.

"Um, the pleural effusion?" whimpered a third-year resident.

"No!" barked the attending.

The others, standing dumbfounded in front of the computer display, searching for another finding but finding none, stood silently.

"Come on, folks!  Look!"

And try as they may, no one saw it.

"The name, folks, the name!" the attending said impatiently.

And there it was, a tiny reminder of whose x-ray it was, quietly lurking in tiny print in the upper right corner of the computer screen, unmagnified. 

But wait, the name was correct.  What the heck was he talking about?

Closer inspection showed another critical piece of information, totally lost on almost everyone standing there: the date of birth of the patient.  It was not the same as the patient being discussed. They were looking at the wrong patient's chest x-ray.  Never mind that their patient had a chest tube placed on the opposite side that wasn't shown on the displayed chest x-ray.  Yet they were already trying to make decisions for care.


I recently taught an EKG reading class and had a similar experience to the one above.  Since July, I've been teaching the basics of EKG reading at least once a month: rate, rhythm, axis, intervals - you know the drill, right?

But I (once again) asked about the axis of an EKG tracing we were discussing some six months later.  A room full of at least twenty residents sat quietly.  No one answered.

I kept my composure.  I prodded them gently, hoping to hear an answer yet none came.  Were they on call?  Distracted by their cell phones or pending work?  Am I THAT boring?

Still nothing.

So I reviewed how we determine EKG axis, and quickly, a few remembered the concept and gratefully, responded correctly.

But these experiences got me thinking about the effects Big Data is having on our residents today and its tendency to build complacency.  Why learn something if you're always spoon-fed it right?

Admittedly, our medical data explosion has prevented us from knowing everything there is to know about anatomy, physiology, pathology, treatment options and the like.  There is a role for access to Big Data.

But increasingly the data we feed our residents and medical schools is nothing but printed characters: x-ray reports, EKG interpretations, study results like "ejection fraction:" all limited to the 256 ASCII character set.  Residents no longer feel the need to look at the raw image and formulate their own opinion - they'll just look at the printed report.  They expect the data to be fed to them in printed format.  They expect the reading to be correct.  In a way, they're growing up expecting to be spoon fed just the black-and-white answers rather than the brilliant data provided by pictures.  Just "google it."

Never mind the computer says "atrial fibrillation" because the original EKG contains noise.

Such an "Big Data-entitled" approach to health care is extremely dangerous, especially if the data upon which decisions are based, are wrong.  Residents should never forget two things my father always told me:

"Garbage in, garbage out" and "expect what you inspect."



John Mandrola said...

Dr Charles Fisch, one of the legends of ECGs, used to teach us to use all the data available on the recording.

The narrow-based t-waves and long ST segments were not the only clues to the electrolyte abnormality. The other was that the ECG was recorded on 4E--the renal floor.

It's always hard to teach mastery of the obvious. It's especially hard these days, when all are so focused on little white screens and check boxes.

Keep teaching my friend. Some day, one of those students might be putting in our pacemaker--if we are so lucky to have lived so long.

Andrew Kaplan, M.D. said...

The irony is that Dr. Eric Topol is going big guns about how high tech solutions will replace our knowledge and intellectual skills and that patients will be able to diagnose their own MI and arryhthmias with portable devices. What are we to do when we who continue to employ "old school" approaches to healthcare are denigrated by our own?

Quan said...

The idea is to use digital data and solution to facilitate decision making. Machine will be better in many ways in pattern recognition than human experts. We certainly do not read Holter recording any more, and will review computer report to make clinical decision.