(Over)Simplifying EHR Usability

Dr. P patted the middle aged patient on the back, helped him off the elevated exam table and guided him to the chair by the sink. He picked up the chart and using the exam table as his desk he flipped through the chart, pulling out several pieces of paper, spreading them to his right, while making small talk with his patient. He reached into his pocket and pulled out a battered silver recorder and without any warning started dictating: “Mr. H is a 60 year old mildly obese gentleman presenting with…..“. He had a pen now in his right hand, and as he was talking into his recorder, shuffling the various papers in front of him, he was also writing orders and prescriptions as fast as he was dictating. “….follow up in two weeks” was the last thing he said. He didn’t write that one down, but turned around, handed the patient a bunch of scripts, told him to stop by the front desk and make an appointment two weeks out and stop by the lab on the fourth floor to pick up a container for the urine test. Two minutes, tops, including the small talk. It was my turn now and I was sweating bullets because I knew exactly what he is about to say. “Can I do this in the EMR?”

--------------------------

EHR usability has finally arrived to Washington as the guest of honor at the most recent ONC HIT Policy Committee hearing. ONC seems to be considering the regulation and certification of EHR usability. NIST has created a testing procedure and just like its Meaningful Use testing procedures, it is superficial and doesn’t really test anything of any consequence. Those who represented “providers” and patients argued for the need to improve usability and those who represented academia and grant funded research argued for more funded research. Predictably, usability experts, argued for hiring more usability experts. Large vendors eloquently stated their objections to government mandating what EHRs should look like and small vendors argued that the more mandates, the better, since this will automatically remove the built-in competitive advantage of those with larger budgets and larger usability departments. As is customary, EHRs were compared to ATM machines, cars, iPhones, Google and a variety of “other industries” that are all so much more advanced than health care when it comes to usability.

When usability, or lack thereof, is discussed, most actual users of EHRs (oddly, there was only one of those at the ONC hearing) think about too many clicks, too many screen changes, convoluted workflows, stilted terminology, finding needles in haystacks, slow and freezing software, crashed servers, disappearing information, mind numbing alerts and lack of functionality. But wait, there is more… There should be out-of-the-box interoperability, ability to customize everything, thousands of templates, no bugs, no need for training, no need to document all that crazy billing stuff, and it wouldn’t hurt if it looked pretty and colorful (as opposed to “dull”), and it should work on the iPhone, iPad, Blackberry, Android, Mac and Windows 98 too.

There are two questions facing all involved: How to measure usability, and who should define and measure usability. The ONC committee is presumably exploring whether government should be the answer to both questions. Before you cheerfully agree that government should indeed regulate EHRs through an FDA approval process, let’s take a minute and explore what it is that we want government to regulate. No doubt, we want government to ensure safety of patients. Since EHRs are part of the clinical process, the FDA has, in my opinion, a clear and definitive role in making sure that EHRs do not endanger people’s lives. Usability, however, is a much larger aspect of a product than safety. To use the completely inappropriate analogy to automobiles (more on that later), it is pretty obvious that government should mandate that cars come with airbags and seatbelts, but it is less clear that government should mandate that all vehicles come with heated seats or automatic transmission, even if manual transmission and freezing bottoms may be tied to some types of accidents, for particular types of users, in particular circumstances, at particular times of day. And here is a trickier question: should the government fund and engage in the design of a preferred seatbelt, and then require that all automobile manufacturers use the exact same design?

Back to the more general question of usability and how it should be measured. ONC is funding projects and the government is paying for contracted work to provide an answer to this question. The initial outcomes as presented at this hearing consist of a rather strange standard form for assessing effectiveness (success/failure), efficiency (time to completion) and satisfaction (subjective) for several use cases based on narrow Meaningful Use criteria as defined by NIST testing procedures for certifying EHRs for Meaningful Use incentives. For example, an evaluator would be asked to prescribe a statin for a patient, or record vital signs, or execute a similarly granular sub-step of real life clinical scenarios. I don’t think I need to belabor why this exceedingly simplistic approach provides no indication for evaluating usability of the EHR. However, as one participant stated during the hearing, it seems that it is better to measure something than nothing. If you are reading this and you are a physician, this way of measuring things out of context, just because we can, would be akin to measuring the percent of patients sitting in your waiting room at a random date and time with a blood pressure under 130/70, and deciding that you are a good doctor if they all do, or a bad one if they don’t, whether you are a pediatrician, a geriatrician, or if you practice in a posh suburb, or tending mainly to indigent and homeless folks, or if it just so happens that this is the time when you do sports physicals for the local boys’ lacrosse team.

To continue on this path to oversimplification, there is a much circulated drawing in the circles of EHR usability experts (created by a former colleague of mine, Eric Burke), depicting three screens: the first shows an Apple screen with one word on it - “touch”, the second shows Google’s famous home page with nothing but a Search button, and the third is a cluttered data entry screen supposedly belonging to an EHR. This drawing is supposed to impress upon us how horrific EHR designs are by comparison to “other industries” and other software products we use in our daily lives. I’m not totally sure what the Apple screen is supposed to symbolize since touching a blank screen does nothing for me (sorry, Eric). I do understand the Google search screen and I agree that if you only want to do one thing, you should only have one button. When you want to do many things, many business and enterprise type things, it would be more meaningful to compare an EHR screen to say, SAP, or Siebel, or Epicor, or Photoshop, or any serious CAD application. The results of such comparison may surprise some usability experts, who seem to have all the answers. EHRs are not leisure applications for consumers and EHRs are not gaming platforms. To use the automobile example one last time, EHR is to iPhone and Facebook what a Ford F-150 is to a Little Tykes Cozy Coupe.

In conclusion, I would like to leave you with a screenshot of a widely used EHR. It indeed defies almost every single usability expert generated opinion on what good design should look like. However, if you look very carefully at the top-left of the screen, you will see that this is a screenshot from VistA, the VA EHR, designed and built by clinicians for clinicians. I have not met a single doctor who used VistA and did not really, really like it.

Click picture to enlarge

So let me ask again, who do you think should decide what a good and usable EHR should look like, a Government usability expert or Dr. P?