Killer Features of the Next EMR

iphone search

I was absent-mindedly playing with my iPhone today and took special notice of a feature I have rarely used before. If you swipe all the way to the left on the home screen, you will get a search bar to search all of your iPhone. This includes contacts, iMessages, and apps. I’ve never needed to use this before—a testament to the iPhone’s ease of use. Just prior to this, I was working on some patient notes using my hospital’s electronic medical record (EMR). In contrast, each task I performed required a highly-regimented, multi-click process to accomplish.

Criticizing EMR interfaces is a well-loved pastime among clinicians. Here, however, I am going to take an oblique approach and reflect instead on what has made good interfaces (all outside of medicine, it turns out) recognized as such.


The Google Algorithm often gets credit for Google winning the Great Search Engine War. Indeed, there are whole teams dedicated to improving it. However, if you compare algorithms today, even 5 years ago, the differences in results have been only marginal. How does Google stay ahead? Speed. Google has done extensive research to determine what keeps users coming back and it is unequivocally speed of results. It has been much of the motivation for creating their own browser (Chrome) and operating system (Android). Speed means more searches and more searches means more money for Google.


With EMRs, wait times to store and retrieve data can be extremely long. Moreover, it frequently takes multiple clicks to get to the precise page you want, further compounding the problem. But how slow is slow? Research in web user behavior indicates that 47% of consumers expect a web page to load in 2 seconds or less and that 40% of people abandon a website that takes more than 3 seconds to load. It regularly takes over 3 seconds to retrieve an important piece of data from an EMR. That makes the experience constantly frustrating; I wish there was another EMR I could switch to. (As a fun aside, I often find myself logging into two computers side-by-side in the hospital to save precious seconds waiting for the computer to load.)


A highly-regimented click pattern to obtain results means that you have to be constantly aware of what you are going to click next. Compare that, for example, to the Facebook Newsfeed where the next step is always “Scroll Down.”

Fortunately, consumer web has discovered a better way for reaching your destination. Why not let people search everything in one place and guess at what they’re looking for? Examples include the iPhone search I opened this post with, the original Awesomebar on Firefox, and, of course, Google search (which lets you type in flight times or conversions like “lbs to kg” along with your usual searches). The main point is that you have one destination from which you can intelligently, semi-algorithmically branch off into any workflow you could need whether that is texting a friend or searching for the air speed velocity of an unladen swallow.

The trouble is that it is hard to do this “awesomebar” search well without knowing what the clinician is looking for. What do you mean when you start typing in “echo?” Which patient are you talking about with the first name “Rose?” In fact, this has been attempted before in healthcare in the form of Amalga, which was sadly mismanaged after being purchased by Microsoft from its founders. In principle this is very doable, vendors just need to listen to and understand their users. It’s been said before, but EMRs would do well to involve clinicians in the design.


This is often cited by users as the most frustrating part of the EMR experience and so I am not going to dwell on it. Cluttered interfaces and uncertain visual hierarchy both make it harder to use software. I hope it is clear from the above that even a nice visually-appealing interface can have problems if it doesn’t incorporate speed and search.


Of course, the ultimate problem is that EMRs are enterprise software, where the payers are not the users (clinicians) but rather the executives who are thinking about how this will improve the bottom line. I don’t see this changing anytime soon, moreover, as consolidation and physician employment (rather than self-employment as has historically been the case) becomes more popular. The popular EMRs do a good job of making sure hospitals get paid, but they do not do a good job optimizing for speed, search, and interface. If it is going to change, it will have to come from the small concierge practices that are emerging with novel clinical and financial workflows requiring custom solutions.


18 replies »

  1. Quick comment:

    You probably don’t want to use the word “killer” when talking about health technology.

  2. And was not given $28 Billion dollars of Federal Money to go out and design things up-front, like the EMR/EHS/HIT.
    One of the reasons that the swarms of e-Fleas are all over medicine was that they were given forklifts-full of cash to “get something done.” It worked as well as the same approach to the warring parties in the Iraq War, I expect.

  3. But the internet revolution did not have a pile of government regulations forcing piles of features some of which were not useful, and, more importantly it was not hobbled by a federal government forcing the use of backward, outdated development methodology. I assure you, had internet applications been hobbled by the likes of the meaningful-use certification process, that revolution would never have happened.

  4. Yes, but as the system “matures” and the patient becomes more “empowered,” all that may be just a mask for the shifting of liability back onto the individual. Attorneys WILL use that point to push the liability onto the patient, and their IT teams will come up with reams and reams of “data” showing the ease of use for the patient, etc.
    So far, it’s a courtesy – but courtesies do not live long in front of insurance companies and lawyers. In our tolerance for deepening bureaucracy, the solution is often buried under ” the patient supposedly has access, but can’t always get through to the portal.”
    I don’t like it – I just gotta tell you what’s in the crystal ball.

  5. “Remember pre-Google search? Remember your old desktop PC? Remember your dial up modem? Remember the original Netscape browser?”

    Yes, and I also remember that everybody used them voluntarily and thought they were great, an obvious and immediately beneficial step forward in sharing information. Only in retrospect do they look bad.

    But now we’re only using EMRs because we’re paid to/forced to, and everybody hates them.

    Big difference.

  6. There is a huge difference, John, between what is not perfected, and what cannot be attained in the first place, given infinite resources, time and patience.
    The Technopaths deny that anything cannot be attained. They are a few generations ahead of those who built the unsinkable Titanic – absolutely unsinkable in theory, no doubt.
    The political parallel is seen in Pyongyang – ” those wishy-washy Russians and duplicitous Chinese gave up on Communism, just before victory was achieved! Only the Great Little Leader has continued on undaunted, and the Perfection of the Workers’ State is within his far-seeing vision!” The food thing, though, the food thing – that’s gonna take some more work.

  7. I caution you on that, Craig. Both the BMW and the Wartburg were German-made cars; the BMW from (West) Germany, which survived, and the other from East Germany, which did not.
    The concept of an EMR is still a Wartburg or Trabant. The dull and necessary details have to be brought into tighter specification. Otherwise, you have a clunker that is barely worth keeping going.

  8. Interesting point. I take for granted that EMRs have to contain patient data (meds, labs, problems) just as you might take for granted that a car needs a steering wheel, wheels, etc.

    There may be interesting conversations to have there, but I think the concept of the EMR is sufficiently mature to start talking about things you describe as more “peripheral.” Moreover, I hoped to demonstrate that these things are sufficiently commonplace in other domains and easy to achieve. However, they are not prioritized for reasons I describe above.

  9. The analogy I see is to the early days of the internet when it looked like that a select group of Silicon Valley companies would completely dominate the tech revolution, stifling all competition.

    Remember pre-Google search? Remember your old desktop PC? Remember your dial up modem? Remember the original Netscape browser?

    All examples of technology getting better in hurry, at the very moment when it seemed all progress was stalling and we were going to be trapped using stuff that didn’t work very well.

    Do a select group of tech companies dominate the industry? Sure. Several of the larger players have gone on to become giants, but we’ve also seen an explosion of creativity like nothing that’s been seen since the Industrial revolution.

    And yes, there’s also been an explosion of stupidity: but you can blame the market for that. I will agree that many of criticisms of the usability of these new technologies are valid. People are being expected to make tools that don’t work very well central to their jobs. The result is a predictably pissed-off group of users. But that doesn’t mean we should all climb into our time machines and try to go back to 1985.

    The dial up modem is another good analogy.

    What we’ve got right now doesn’t work very well. It chugs along. It sputters. It stalls. It needs to be unplugged and rebooted every now and then. But our little modem gives us a sense of what the future holds in store, nonetheless …

  10. I agree with this wholeheartedly, Peter. The problem is like so many others with these systems, the patient supposedly has access, but can’t always get through to the portal, etc. In the end, we need to continue the doctor/patient encounter to discuss the results.

  11. Steveo, think about what you just said. I had access, but the doc ordered the tests AND has an obligation to schedule an appointment to review the results, or do a followup phone call. The test results go to the doc first, to me it’s mostly a courtesy.

  12. There is a dark side to this cheerful and sunny gathering of data for patient access. You know, Peter, you DID have access to your records, you simply CHOSE not to review them. This will become a critical point in the litigation over care. The attorneys will snap this up like chum.
    “Ma’am, so you are telling the court that all this time you had access to the results of your breast biopsy showing cancer – you just couldn’t take the time yourself to look them up. And now you are complaining that someone ELSE should have told you the results?”
    Offering access allows for the delegation of liability. And as we all know, the IT people are not terribly pressured to make their systems usable – all they are legally obligated to do is provide a PATHWAY to the data. If you do not learn the pathway, it’s your fault.
    I do not like this.

  13. The last time I had the opportunity (there was no need) to access my EMR is when the doc emailed me that my lab test results were available online. The problem was all the different interconnected curtains of security necessary to just get there. I had several failed attempts and was just frustrated even though I’m pretty good with computer stuff.

    Did it once and that was the end.

  14. Thanks, Bill, wonderful points.
    It is vital for those pimping the New Medical Revolution to insist that doctors are opposed to meaningful reform for selfish parochial purposes.
    That “New Revolution” is merely the same story that has gone on since Medicare and insurance came on board, and the testing and utilization question became adversarial between doctor and payor. But that is a long and somewhat tangential story.
    You touch on a vital point with AI and diagnosis, although it is not the medium itself – reading Harrison’s is enough to get one well into any subtleties of rare and arcane diagnoses.
    Medical IT is balanced on a huge quantity of spectacularly bad assumptions that will crash the system as surely as all the bad ideas of Marx, followed as gospel by those who half-understood his theories, crashed the Communist system.
    There are axioms of the scientific method which are ignored – if even understood – by the HIT mavens. The first of this is that there is an infinite amount of information, but it never expresses MEANING until it can be used to answer a question. No matter how vast a pile of data is, it is finite.
    It is up to the human mind to develop the context in which the data takes meaning. Scientific method is based on logical positivism – until the question is properly asked, the data has no value. Human minds express methods and algorithms by which the universe can be ordered. Computers, no matter how clever, are not algorithm-generators, but are algorithm-appliers. The insistence that the computers are “more right” than humans is a terrible expression of technopathy. The “algorithms from above” are not implanted by the computer; rather, they are laws created by men and women who consider themselves the wisest, and codified into an electronic format. No doubt they are wise; but they are not all-seeing.
    Algorithms do not span the space of the human condition. At best, they provide a bijective map between ideal clusters of symptoms, and ideal diagnoses. They are “mile-markers,” no better.
    Real medicine is determining what is in between. The worrisome trend I see in medicine is the concept that what the patient expresses is not identifiable in reality. Skill in medicine is the ability to close the lacunae of diagnostic understanding in the care of the individual, the Actual Patient. Nowadays, chronic abdominal pain is treated in the format – “I checked everything on the algorithm, and came up empty. Therefore, the chronic abdominal pain does not exist.” That sort of approach is unsophisticated, unintelligent, and horribly expensive.
    The definition of an instrument or tool is that which can be used to do a thing. A useful instrument is one that allows one to accomplish something easier, quicker, more efficiently, etc. In the New Revolution, tools are not offered to physicians, but rather as a threat to physicians – something that anyone can wield that can replace the physician. EMR/EHR is not a tool for medical care. It is a $30 Billion Dollar make-work slush that pretends to be the electronic bridge between IT and Medicine.
    I am fascinated that, as a physician, I get into argument with HIT people who insist that they can offer me a swath of historical information on a patient – much of which is not interesting to me, and much of which it is inaccurate. I say I do not need it; they say I do. It is tremendously useful, though, for someone compiling centralized records of people – such as a “citizen database” in some mythical totalitarian country – to track its own people.
    I have just completed a job application, and filled out 18 pages of paper, much of which reduplicated the 11-page electronic on-line application for credentialing; which in turn repeated a pre-application, all of which contained my life in sequence since high school. I am over fifty now, and I have told my life history so many times.
    I received a phone call from the people with whom I had corresponded. They also had my resume. They asked me for my home address, which was on the CV, medical license, and probably in 12 different places elsewhere. The computer can multiply useless redundancy faster than we can keep a lid on. IT may well accelerated the grinding of American Medicine to a rusty and demented halt by the end of this decade. Schade.

  15. I keep thinking that maybe the sweet spot in the degree of digitization of the medical record was about two years ago before the grand digital mandate. At this earlier time we had incorporated about as much computerization as the natural evolution of medicine demanded. It fell into place and was not forced. We had laboratory results, radiology, anatomic path, echos, EKGs, vascular studies, EEGs, drug incompatibilities, tumor staging formats, discharge summaries, and a few more items I have forgotten, all in digital formats or at least in text which could be scanned. Alas, now it feels artificial and forced and we are installing clumsiness that might last a long time because we can’t pay the programmers forever.

    Docs were not late adopters. We were early. E.g. We had a DEC PDP8 computer in my lab in 1971. 8k’s of memory in ram! We color coded our office charts and wrote cryptically, but the memories in the chart were as accurate as page after page of copy and paste in today’s repetition hell.

    After a few dozen more hacking jobs and a few expose books on the leakages of vital personal data, like the movie “As Good As It Gets”, I believe we are sure to have some fierce backlash on digitization and its super goal of porosity, interoperability. Because of this, people will demand less easily read health records…yes, to go backward! Specifically patients will ask us to handwrite certain portions of the chart. Gasp! There it shall be restrained a bit from entering the world’s big data.

    We also desperately need some AI to help us diagnose. Come on, we can’t remember all the types of familial Meditarranean fevers or all the hemoglobinopathies. We need this before we have to put in digital stone the fact that Mrs. Doe had a D&C that contained products of conception. Bureaucrats probably don’t know what I am saying. But think priorities.

    Does the government have any idea how damaging some of these records can be? Did it think? Employers want to know. Insurers want to know. Fiances and spouses want to know. Political opponents want to know. You can’t just order all these records to be written the sky so that the world sees them. Do you have any clever cryptography in mind?

    Anyway this artificial push to swim in EHRs will see a pushback. Hopefully policy will ascend the learning curve.

  16. At the risk of sounding unkind, these are peripheral aspects which do not express much about the guts of the EMR. My next car should have windshield wipers that sense the variability of a light mist, based on the power needed to sweep the wipers across the glass, and adjust them accordingly.
    My next car should also have great airbags and brakes and passenger restraints so we don’t get killed.
    EMR’s, like the Microsoft Office products, have a massive amount of features that are unused, and do not offer a benefit to the user. The medical EMR does not seem to have a very clearly defined mission, but is a hodge-podge of addons masquerading as an entity.
    Who can describe the EMR/EHR under a clear and small mission and purpose? Then, let’s talk. Nice wipers won’t help.

  17. While we’re on the topic – an EMR that revealed pricing information for tests, procedures etc – would do wonders to hold health care spending in check

    This IS something that should be tied to certification.