EHR Cutting & Pasting, in Perspective

I've started to think the medical record is akin to DNA. Maybe 10% (or less) is useful information; the rest is junk. When folks want to find a sequence of significance, risk or reassurance, they've got to search for the good stuff and filter out all the garbage.

But junk DNA is believed to have a purpose. Some regions of junk DNA are highly conserved -- found in organism after organism -- suggesting an important function. In medical charts, conserved regions are also repeated. And they also serve an important function.

It's this repetition that Dr. Bryan Vartabedian called "Cut and Paste medicine" in his excellent recent post. He's concerned that all these computer-generated phrases of historical elements, exam findings and decision-making makes all patients look alike, and hurts continuity of care, as it becomes harder to discern what's actually going with the patient.

It's a reasonable concern. This problem, created by documentation regulations, compounded by declining reimbursements, and exacerbated by quick-fix features of some electronic records, can be solved through technology, too. Just as researchers and geneticists built tools to sift through DNA, to find the small section they're looking for, we need to easily search through records to show the details of patient care relevant to us.

Maybe this solution will simply highlight free-text sentences and paragraphs, and gray-out all the checkbox-generated prose. Maybe these searches will involve natural-language processing, or complex filters based on provider or position. I'm hopeful this problem will be solved, because medical records aren't getting shorter.

And here's where my analogy with DNA breaks down: junk DNA is also called "noncoding regions" because those sequences don't code for proteins. But in the medical chart, those junk sections are actually designed for coders -- they have key phrases that medical billing companies look for, to show to the insurance companies, to pay us. You could be providing the most competent, compassionate, time-consuming care in the world, but if those phrases aren't in the chart, nobody gets reimbursed. Other instances of these oft-repeated, computer-generated phrases in medical charts are designed to protect against legal liability, which also serves the financial interest of providers and healthcare institutions. 

The good stuff, the free-text prose that describes what the doctor is thinking, may only be a short paragraph in a sea of vital signs and lab results and macros and checkbox-generated text. While this section is the most important part of the chart to future caregivers, from a medical billing perspective, it's essentially a noncoding region.

There was a time when medical charts were short and designed soley to communicate patient care to  future providers. Medpundit once wrote of a mentor who could boil down an encounter to two terms, like "ROM - Amox" (right otitis media, given amoxicillin). Years later, a similarly simple encounter would have to run for 10 or 20 lines of prose.

By 2008, Peter Viccellio wrote:
When an ED visit for a cough, with diagnosis of pneumonia, consumes 17 pages of print, something has gone awry. (Or perhaps things went awry when pulmonary edema was no longer considered an emergency unless there was a documented family history, social history, and 10 reviews of systems.)
True. Things have gone awry. But I can't see any way back. So tell me: why should the medical chart ever printed, in a hospital? (Besides the increasingly rare scenario where a patient moves from an electronic part of the hospital to a paper-chart area). And if the chart's not printed, well, why should anyone on the care team have to scroll through 17 screens' worth of prose? Caregivers should see the parts of the chart really relevant to patient care.

Let's recognize the vast majority of the chart for what it is: coding regions that keep the hospital or the practice afloat, and comparatively safe from prosecution. These sections are not really important for patient care, and they're the last thing I want to see when I pull up a chart about a prior visit. Recognizing that, and building EHR search and display functions around that, and you'll solve a good deal of the frustration around electronic records and their cut-and-paste confusion.

The biggest reason we haven't done this already, I think, is pride. It's too painful to admit that most of what we're doing on the computer is not directly relevant to patient care. Let's get past that, recognize the checkboxes and macros serve a primarily financial function, and give those sections the low status they deserve when we run our searches and pull up our charts.