By Fred N. Pelzman, MD
Apocryphal story from residency: On morning rounds in the critical care unit, the post-call resident starts to present a complicated patient admitted overnight with chest pain, and after the first bits of the history have been presented, the wise old cardiology attending turns to the gathered medical students who are just starting their first clinical rotation and asks them what they think this could be.
One medical student bravely raises their hand and says, “Acute aortic dissection and regurgitation secondary to relapsing polychondritis.”
Stunned, the cardiologist says, “That’s incredible, you’re absolutely right, how could you possibly know that?”
To which the medical student replies, “What else causes chest pain?”
For some reason, this old story popped into my head again after our most recent workgroup session on helping develop new functionality in the electronic health record.
The development team from our electronic health record vendor asked for clinical input — from those of us actually taking care of patients — about developing some of the cool new tools they are working on for the next version of our electronic health record (EHR) coming soon to our institution.
The group of primary care providers meets with their team and offers advice, recommendations, and feedback.
The EHR has lots of built-in functionality, things that they’ve designed which they think will help make our lives easier.
They group things together into these modules for clinical problems, and when you select one of them it brings in a whole menu of items across various categories, from potential diagnoses; a checklist of orders for various items such as labs, referrals, imaging, medications, and in-office treatments; some templated note functionality; and even options for choosing level of service for billing.
They think that this is going to make our lives easier, but everyone on this committee keeps scratching their heads and saying, “When am I possibly going to use this; is this really going to make my life more efficient, or is it going to end up creating some templated junk that really doesn’t do what we think the medical record is supposed to do?”
Take, for instance, the tool they have for evaluating a patient with “fever.” The workflow started with a list of diagnoses that do, in fact, present with fevers — but it was such a random selection of things.
True, the list did include influenza, cellulitis, urinary tract infection, pneumonia, and many others. Some seemed more random, such as viral infection and acute hepatitis.
But if we already know what the diagnosis is, why are we trying to figure out what’s causing their fever? Why would we be using this tool to help figure out what it is, if we already know they have influenza?
The orderable tests seemed scattered across all of medicine, some we would never order, and some I have never ordered (ultrasound rectum, routine?), no matter what I thought was going on.
The “fatigue” order set was also unhelpful, both too specific and way too vague. Diagnoses included transient heat stroke, depression, and anemia — plus neurasthenia, about the vaguest diagnosis there is.
Options for testing ran from the sublime to the ridiculous. Phosphorous and aldosterone, MRI of the brain and referral to infectious diseases. Medication choices included a vast array of stimulants for narcolepsy, and sleeping pills.
These do not seem to work either for acute presentations as first-line workups, or more intensive investigations when the diagnosis has remained elusive. This is not how patients flow through an office visit; this is not how clinical care happens.
Our interactions with our patients, and our decisions about what might be going on, and what additional testing we might need, happen in an organic, messy, real-life manner, which is really, really, really hard to put into some narrow templated framework.
There are certainly instances where it would be nice to have clustered together a group of the things that you do every time for a certain clinical situation. There are lots of things that we do in primary care — and many things in subspecialty medicine, and probably many more that the sub-subspecialists do — that are almost always done the same way. For these sorts of things, in terms of evaluation and management, diagnosis and treatment, it is certainly worth developing tech tools to make these regular and routine clinical interactions go more smoothly, and ease the documentation burden in the EHR.
Most of us agreed that concrete clinical complaints like cough, chest pain, and palpitations might work out okay with this format. But something as vague as fever or fatigue just doesn’t seem to lend itself to this sort of technical solution.
If we turn all of our clinical documentation into a bunch of checkboxes and toggle selections from lists, I fear that our charts will continue to degrade towards unreadable garbage that offers no insight into what we were thinking, what we wanted to do, and what the heck was going on with our patients.
We appreciate being asked to participate in the development of the next generation of our electronic health record, but moving forward we need to make sure that our voices are heard, that our needs and the needs of our patients are met, and that what we come out with at the other end is truly the best system to take care of our patients, not just something the computer programmers think is easy to build.
There is more poetry than accounting in the care we provide for our patients, and the art of figuring out what is going on and how we can get our patients to a better state of health will always be a challenge to force into a menu of little clicky boxes.
Because, what else could it be?