All I want is...

…a meaningfully usable and user friendly EHR.

In the EHR universe there is a terrifyingly wide chasm, as yet unbridged, between the potential benefit of the ideal EHR and the real world impacts of our current EHRs. As an EHR evangelist, I often advocate for the use of products that I know to be mediocre on their best days. Why? Because a slow horse will always beat no horse. I have been using an electronic record in my primary care practice since 2001 and, despite how much I dislike it, I would never go back to paper. Although I think adopting an EHR is a no-brainer, I am not satisfied (to say nothing of happy) with my current EHR. It needs to be more powerful, more usable and more friendly.

What would make an EHR meaningfully usable and user friendly? Here are my criteria.


The job I do is patient care.

Therefore, the EHR must be a clinical tool, designed explicitly for aggregating, evaluating, integrating, and presenting clinical information to individual  patients and their clinicians while they work together to solve clinical problems. (Note: The EHR contains the patient’s health information and is a tool to address the patient’s health  issues. Clinician’s acquire and use this information only at the request of patients. I therefore define the ‘user’ as the patient-clinician team, not the physician, office ,or institution. This has important ramifications, as we will see.)

The EHR is not an expensive word processor used to save on transcription costs by shifting data entry work from the transcriptionist to the clinician. It is not a billing instrument, used to document services to justify CPT codes. And it is not a data repository, a library to store information.  To the extent it is asked to do additional administrative work, great care must be taken that administrative tasks not interfere with its raison d’etre. If anyone doubts this, they should ask themselves this: do you want flight crew logistics or payroll information to interfere with the pilot landing your commercial jet at night in the rain?


It must be intuitive.

It just works. Yes, I know. Medical care is complicated, medical information is fractal, and patients are unique individuals, so an EHR can never be as simple as a toaster. Nonetheless, it is possible to build an EHR so the correct action is the default; it should never take extra work to do the job right.  Basic operations and navigation should not require any training. Amazon, Google, YouTube and World of Warcraft would be dismal failures if users had to attend training courses to participate on their sites. How straightforward and logical must it be? Enough that the patient, without extra training, can easily navigate, understand and contribute to his or her own record.

(It’s actually not fair to claim that current EHRs are not intuitive. They are. It is just that they are designed around the billing process and are therefore only intuitive for those fluent in the language of CMS criteria, CPT Coding, and Relative Value Units. That’s why highly trained medical professionals struggle to use the EHR.)


It must be efficient.

The processes of health care rarely follow defined paths through a static landscape. EHR workflows must accommodate this. Defined collections of templates requiring stepwise progression from one prepackaged form to the next limit the ability of the patient and clinician to follow, understand or document the patient’s story in its own context. That is, the typical EHR is actually designed to make it harder to record and understand the pertinent medical information.  (The typical EHR tries to shoehorn a patient’s story into a pre-defined billing documentation format in much the way the IMDb summarizes a film by title, author, characters, and dates. There is certainly accurate and important information there, but the IMDb page is not a substitute for seeing the movie.) 

It’s as if EHR software engineers didn’t understand the semantic web or that this is whathyperlinks were designed for! 

Here are a few things, standard in much other software, that would make use of the EHR more efficient:

  • Multiple easy hyperlinked paths through the current illness, past medical history, life style information, testing results, exam findings, medications, allergies and the like. 
  • Work can be done as it presents. It should be easy to accomplish a task that presents during a separate task, without complex navigation or losing one’s place. (For example, entering an adverse reaction to a previously prescribed medication or test results while taking a history of the course of a chronic illness.)
  • I don’t have to wait. The EHR opens and closes charts, and moves from page to page within charts, without delays. (In my current setting, 3 second waits are routine, and 20 second waits are not rare. They add up.)
  • I don’t get logged out  of the EHR in my office when I am away from my desk to go to the bathroom or see a patient, requiring multiple daily logins.
  • The most common tasks are easily accomplished.
  • Steps are minimized. In medicine, we grouse about the huge number of clicks necessary for simple tasks, sometimes called clickorrhea. In my system it takes 11 keystrokes to enter ‘Sunburn’ on a problem list, and 21 keystrokes to indicate that I have read and reviewed a document from a consultant.
  • Data is efficiently acquired, organized and integrated from multiple sources. This means that results from office devices, laboratories, radiology, other clinicians AND THE PATIENT flow easily into the record, not requiring time consuming and error prone human data entry.
  • Keyboard commands are intuitive and available for almost everything. (Using the mouse is slower, requires substantially more cognitive effort, and moves the attention away from the problem and to the screen.)
  • Parsimonious use of alerts. Automated warnings are one of the great strengths of the electronic record. It is fantastic that the system can automatically alert me to an important drug interaction or contraindication based on a diagnosis. However, the systems are insufficiently smart and do not distinguish between oral and topical medications, current and prior diagnoses, or important versus trivial issues. The result is ‘alert fatigue’ and there are multiple studies that document that clinicians ignore (and claim they never saw) more than half of high priority or urgent alerts.

In the current EHRs on the market, these tasks are generally organized by data engineers; each set of tasks is separate and largely unlinked, organized not according to clinical work flows but according to data and task categories. One enters medications in the medication module and smoking on the life style page. 


It is just helpful enough. 

When I use my EHR, my goal is patient care. I am not striving to reach some geeky form of Nerdvana tool mastery. The Goldilocks Rule applies here. Real time program support should be “not too big, not too little, but just right.” 

  • Dialog boxes should appear on beside the active window and not obscure one’s primary task. 
  • Alerts should be informative. (“This action is forbidden.” is not as helpful as “This medication cannot be added because it is no longer available.”) 
  • Most important, the amount of help offered should be controllable. Sometimes I want to know if foods impact a medication, sometimes I don’t. If I want to know if the red color of a pill might be a cause of a drug interaction, I should be able to see - but I should not get a top priority alert for eery red medication because the patient once had an allergic reaction to a medication that was red. (It may be hard to believe, but that is what my current state of the art EHR does. The vendor and my IT department know, but have no plans to fix this.)


The interface is simple and focused.

The interface should be designed consistent with what we know about cognition and decision making (behavioral neuroscience). Most EHRs (including the one I use) are designed by a committee of special interests, with multiple members trying to maximize the prominence of their domain. 

A basic but rarely considered principle is that of interference.  Behavioral neuroscience tells us that there are finite (and surprisingly narrow) limits to how much information humans can perceive, absorb, process and remember. Attention is a zero sum game. We simply do not multitask. To the extent that we attempt to address secondary tasks, our performance on the primary task degrades. EHR design routinely ignores this. The EHR should clearly present what is pertinent and ONLY what is pertinent. It should allow me to easily filter what is displayed based on the task at hand. 


Some common violations of this principle include:

  • Real estate should not be squandered on information not used in clinical encounters. Amazingly, the top of the screen is devoted to a ‘banner’ with demographic and insurance information that I NEVER need during an encounter. Why clutter up my mind with insurance and phone numbers?
  • More information is not automatically more informative. This is why we only resort to the OED and a magnifying glass when other easier resources fail.  It must easy to find what is important on a screen with an initial quick glance. For example, when I see a lab or X-ray result, I want the patient name, date, test name and result to be the first and most prominent things I see. I should NEVER have to scroll through multiple screens to find the result.
  • Simplicity. Going forward, patients will be ever more involved in their own care and records. The EHR must be navigable and understandable by the patient whose information it contains and whose health it serves.  An excellent demonstration of how well this can work can be seen in this TED talk
  • Page layout must be consistent and predictable.  The same keyboard commands should work for the same tasks no matter where I am.   
  • The EHR must allow me to design and use filters, so I can determine what collections of information I wish to see for any given task. I don’t want to be forced to slog through pages of information about treatment of an infected ingrown toenail while evaluating a lung mass or reviewing a chronic illness like hypertension. 


A brief concluding note about tails and dogs.

The EHR is a clinical tool for the patient and the clinician. We don’t work for the billing office, the auditors, the insurers, the regulators, or the software engineers.  We certainly don’t work for the EHR itself. The EHR must work for us. Every item on every page, every mouse click and key stroke, every piece of requested or displayed information should be justified in terms of a simple question: how does this help with the care of the patient?



Links to more on this topic::