Home » Rick Weinhaus » Currently Reading:

EHR Design Talk with Dr. Rick 1/23/12

January 23, 2012 Rick Weinhaus 33 Comments

Computer-Centered versus User-Centered Design

Within the next few years, most physicians in this country will have converted from paper-based charting to electronic health record (EHR) technology. This is an unprecedented technological change in healthcare delivery. Whether this technological transformation succeeds will in large part depend on the design of the EHR software itself.

As a physician in clinical practice, my day-to-day care of patients depends in large part on how easy or difficult it is to interact with my EHR. Like many of my colleagues, I find that while my EHR provides all the necessary functionality, using it requires too much cognitive effort. In other words, the EHR design is computer-centered instead of being user-centered.

What’s the difference between computer-centered and user-centered design? Let me give an example.

Imagine that you and your very young son have recently started playing tic-tac-toe against each other on two networked computers. Your son thinks he should be winning more games, so he proposes a change, not in the rules, but in your screen view, in order to make the odds more even.

While his screen view of the tic-tac-toe grid will remain the same, your screen view will no longer be the standard three-by-three grid, but rather will be a single row of nine boxes.

He enlists his older sister, who is great with computers, to program your new user interface. Each of you can only see your own screen.

1-23-2012 8-12-11 PM

The first three boxes in your row correspond to the three boxes in the top row in his grid, the next three boxes in your row correspond to the three boxes in the middle row of his grid, and the last three boxes in your row correspond to the three boxes in the bottom row of his grid.

So, for a particular game, your respective screen views would be as follows:

1-23-2012 8-13-43 PM

1-23-2012 8-15-02 PM

All of the sudden, you find that you’re working pretty hard just to play tic-tac-toe. You’re working hard because you can no longer ‘see’ the problem.

First you have to mentally reconstruct the normal three-by-three tic-tac-toe grid, then mentally segment your row of nine boxes into three groups of three, and then transpose each segment back onto the appropriate part of the tic-tac-toe grid that you are keeping in your head. (Alternatively, you might decide to solve the problem using a different strategy, but that would still require cognitive effort on your part.)

With a lot of effort, you’re able to stay pretty even with your son, but then your daughter introduces a second challenge — a two-second time limit for each move. At this point, your son starts winning a lot more games than you, restoring family harmony.

What is interesting about this example is that, from a logical perspective, the two screen views contain exactly the same amount of information. And, in fact, if a computer program were using an algebraic algorithm to play tic-tac-toe against you, the screen view would be immaterial.

But for humans, it is clear that the grid view works better. It works because we can literally ‘see’ the solution.

If we see a tic-tac-toe grid, we can visually superimpose horizontal, vertical, or diagonal lines at will. If we are faced with the game position below, we don’t have to compute the slope of the line passing through the two Xs or solve an equation to know whether that line would also pass through the square on the bottom right.

1-23-2012 8-16-52 PM

In other words, the human brain is an extremely powerful computer, but one that evolved to help us survive in the physical world by making sense of our spatial environment. Our brain is almost always better at solving problems visually than by using formal logical or mathematical operations.

Donald Norman, a cognitive scientist and pioneer in applying human cognition to design, has written extensively on this topic. In Things that Make Us Smart, he devotes a chapter to why certain design variants are easier for humans than others, even if the variants are formally identical. He includes one diabolical example which turns tic-tac-toe into a variant of Sudoku.

Humans enjoy solving mental games and puzzles for fun, which is why we invent things like Sudoku, but we don’t enjoy them at all when they interfere with complex tasks. Physicians need to be able to devote their full cognitive attention to patients in order to help solve their very real health puzzles.

As physicians, we need user-centered EHR designs that take advantage of our innate visual and spatial perceptual abilities and stay in the background, instead of competing with patients for our finite cognitive resources. Far too many EHR designs force us to play linear tic-tac-toe.

Next post:

Why T-Sheets Work

1-23-2012 8-09-09 PM

Rick Weinhaus MD practices clinical ophthalmology in the Boston area. He trained at Harvard Medical School, The Massachusetts Eye and Ear Infirmary, and the Neuroscience Unit of the Schepens Eye Research Institute. He writes on how to design simple, powerful, elegant user interfaces for electronic health records (EHRs) by applying our understanding of human perception and cognition. He welcomes your comments and thoughts on this post and on EHR usability issues.

HIStalk Featured Sponsors


Currently there are "33 comments" on this Article:

  1. Great post!
    And one more gripe: Both the health system in general and the EHR in particular encourage a whole lotta silly and redundant “information.” It sounds fabulous to have everything at your disposal, but in practice what breaks down is not only the presentation/fractionation of this information. What causes a breakdown is that to get to the meat requires wading through an incredible quantity of noise.
    Make one more change to your example: Within every square is not just a single X or O; within every square is the entire alphabet, and you have to look through them to see if an X or O is represented there.

  2. The tic-tac-toe example is cleverly cute, but assumes something important: there is a single goal. However, even with a single goal, the assumptions about the ability to use the interface is of little import if you don’t know the goal — which I argue below is the problem for complex systems. Indeed, this is highlighted by anyone who has played tic-tac-toe with a child: the “information” and the “grid” don’t mean much if you don’t understand the goal. It is only once the user understands the “goal” that the player sees its advantage over the linear representation.

    The challenge for any EHR system, and for complex systems more generally, is figuring out what the goal is. I’d say that at some level this is an achievable goal for solitary tasks. But, the analogy for an EHR might be better stated has having a series of games lined up tic-tac-toe, Go, Monopoly, Chess, kickball, shuffleboard, pong, etc. and players forced to play them skillfully in rapid succession.

    Finally, even if you could identify all the goals, EHRs face other hurdles: very few rules (and lots of “house” rules), a much grander scope of information, and huge variations on “choices” (not just x, o, and empty). Moreover, user preference largely will drive information interprets as useful and usable.

    All of that said, EHRs can do a much better job, of course.

  3. Great article on application design human factors engineering. Having worked in IT, and in Healthcare for over 20 years my only comment, or question is “who is teaching Healthcare workers Tic Tac Toe”? I do not disagree with the analogy, however it is based upon the premise that we are all taught the same game as a child. In my experience with physicians, nurses, specialist ,etc – they all have been taught their own version of TTT. Intuitive design, efficiency, effectiveness, supportability, are all based upon the premise of consistency from an IT perspective. I have witnessed “discussions” on standardized formulary or order sets, etc, and as an IT guy come away with respect for the clinical knowledge, but dismay on the individual interpretation. I have yet to meet a technologist that wants to “write a bad application” and still uncertain how we ultimately solve this.

  4. Rick,
    Excellent – certainly applies to RNs as well. Explains why we are still carrying paper notes in the acute care setting even when our documentation is 99% online. I’m sharing this with my MSN students as well as my nursing informatics colleagues. Now if we could get the application vendors to listen…

    Looking forward to your discussion on T-Sheets especially since I spent so many years in ED and like their system.

  5. Thanks for all the comments!

    Skeptic, I agree completely that there is an incredible amount of noise in many EHR applications. This is one of the topics that I will be posting on in the future.

    mmm, you are absolutely correct that the EHR design needs to be goal-directed. This was just a first introductory post to present the general issue of computer-centered versus user-centered design. My tic-tac-toe example was not intended to be an analogy to an actual EHR user interface. I do believe, however, that it is possible, at least broadly, to define users’ goals for EHRs. It is also critically important to make a distinction between goals and tasks.

    IT Guy, I am very interested in your perspective. Could you give a specific example of how different clinicians might have individual interpretations

  6. IT Guy, I am very interested in your perspective. Could you give a specific example of how different clinicians might have individual interpretations of the same user interface or data set?

    CherylRNPhD, thanks! Absolutely all the points in my post apply to RNs as well. I am interested not just in the physician’s experience, but in the perspective of everyone who uses the EHR, including all clinicians, RNs, techs, front-desk staff, and so forth. I was just trying to keep this first post simple.


  7. There is, of course, published guidance from NIST on how to incorporate UCD into development processes: http://www.nist.gov/manuscript-publication-search.cfm?pub_id=907313

    as well as guidance on how to perform design evaluation and human user performance testing (summative or validation testing) on EHR systems:

    Whether system developers follow this guidance or purchasers or certifiers of systems require that developers follow it is another issue.


  8. My apologies – my reference was more directed at workflow. The data may be consistent, and placement, colors, applicant functionality may follow NIST standards with the best HE work on the planet, but different medical disciplines and individuals tend to have been trained differently, so their perceived perfect work flow varies. A good example is as a child I was taught to get ready for school, but the way I go about it and when I brush my teeth and hair versus when I eat and dress vary differently than my Neighbor. Some may say these experiences and variability is what makes a great physician better than another – however trying to develop software and the workflow within it for all physicians, and nurses, produces a high degree of variability in design, workflow, reporting, presentation, etc. Which translates into confusion for some and greatness for others.

  9. Leave it to an Ophthalmologist to tell other people how to create readable charts….

    @ITGuy: you’re right in that there is a consistent way to review documentation, but no consistent way to review raw data. So a family practitioner who prefers flowsheets showing trends over time will review the same data differently as an orthopedic surgeon who prefers snapshots of key data, for example. There is a high degree of variability from one physician to the next, which is a big reason why the usability debate is so incoherent.

  10. Blah, thanks! I am looking forward to posting on a regular basis on HIStalk. I will be starting with high level design issues and then proceeding to mid-level and lower level concerns. A problem with many discussions of EHR interaction design is that there are no screenshots or actual examples. In these posts, I look forward to presenting lots of screenshots and examples of what I consider to be good designs as well as ones that could be improved. This will no doubt generate lively discussion. I also look forward to proposing some of my design ideas to the HIStalk community for discussion and debate. I look forward to your input.

    QA, thanks as well! The NIST papers you cite are excellent. As guidelines, they are by definition general in nature and do not propose specific EHR design solutions. I think of my posts as being practical extensions of NIST and similar EHR design guidelines.

    IT Guy, thanks for your clarification. I completely agree with you. It has always been puzzling to me that two excellent clinicians can be using the same EHR and while one finds it extremely helpful, the other can’t stand using it. I don’t think this difference is attributable to lack of commitment or training. Rather, I would think that different mental models of patient care largely account for this difference. That being said, some EHRs do a better job than others of supporting disparate mental models.


  11. Rick – the NIST guidance in “Technical Evaluation, Testing and Validation of the Usability of Electronic Health Records” explicitly DOES NOT provide design guidance.

    What it DOES provide is SPECIFIC guidance on how to evaluate whether chosen designs result in “use error” that could have potential negative implications on patient care.

    A key difference between your tic-tac-toe example and that of EHR design is that the worst case scenario for “use error” in the tic-tac-toe game is an unintentional move that results in losing the game. The worst case scenario for “use error” with EHRs is of course much more grave.

    There are an unlimited design alternatives for anything, including an EHR. Vendors should perform formative testing to evaluate design alternatives as part of a comprehensive UCD-driven development process.

    In addition, they should perform summative testing – as NIST has described in “Technical Evaluation, Testing and Validation of the Usability of Electronic Health Records” – prior to release of their products to demonstrate that they have made sound design choices and that their products are free from critical “use error.”


  12. QA — Thanks for your clarification. I stand corrected. While my first post falls squarely within the realm of usability, for me the most interesting questions are how to design an EHR interface consistent with the clinician’s mental model of patient health and how to ensure that the EHR design addresses not just the tasks, but the high level goals of the clinician and other users.

    You are absolutely right that there are unlimited design alternatives, which is one reason that interaction design can be such a thorny discipline. One of my major goals in these posts is to discuss what I consider to be the implicit mental model of patient health that most EHRs present to the clinician and to suggest an alternative mental model for discussion. I very much look forward to your continued input.


  13. Terrific post, Rick. Thank you for finally bringing design talk to the table. The healthcare industry is certainly in need of this education and I look forward to your future posts.

  14. Great post Rick! To the point regarding proper user interface, some of the design limitations of EHRs can be addressed via extensibility. To this regard, we recently developed a solution to integrate health calculators with the EHR: http://blog.galenhealthcare.com/2012/01/25/ecalcs-integrated-health-calculators/. Some EHR vendors also offer integration capabilities with RIS systems as well: http://blog.galenhealthcare.com/2010/11/17/allscripts-enterprise-ehr-imagelink-demonstration/.

  15. Vincente — Thanks so much. By the way, I found your summary of the ONC Certification/Adoption Workgroup’s Meeting on Usability last April https://histalk2.com/2011/04/27/readers-write-42711/ to be extremely informative, especially the recommendations to vendors. I look forward to your comments and input.

    Justin — Thanks! I very much like the model of integrating health calculators, images, and other external information directly into the EHR user interface in order to present the user with a unified view of the data.


  16. Great post Dr. Weinhaus.
    I’m afraid it’s a bit worse than what you describe. sometimes the game is played with x and o, but mostly you first have to figure out what cute little icon represents the x and which little box represents the o, which may be different in different parts of the application.
    The grid itself may have 9 squares, but not always, which means the game is rigged from the get go.
    And then, of course, the entire image is camouflaged inside an all encompassing Where is Waldo? game.

  17. I am an ER doc married to an IT guy. Not until he read your article did he have an appreciation for my constant complaint that our EHR increases medical error, when the provider is a spacial relationship thinker for instance, when the EHR changes the order of my patients on my track board ( which it routinely does trying to rank the patients by acuity) error is inevitable. In my mind the last patient I saw should remain at the top of my list. Order is everything in the er. Add to this the problem of moving patients to hall beds after examination. The doctor who relies on spacial relationship for order of thought….. Is screwed, along with the patient.

  18. Jjerdoc — Thanks for your comment! You bring up two extremely important issues.

    First, you point out that unless you are actually using an EHR yourself, it is very hard to understand what the design problems are. Furthermore, as EHR users we know that certain interface designs don’t work very well, but it’s not always easy to articulate exactly what the problems are. A major goal of mine in these posts is to help articulate these design problems as I see them.

    Instead of saying “I know it’s not perfect, but it works,” I would like administrators and vendors to say, “Now I get it. What can we do to make the design better?”

    Second, your example of your track board changing the order of your patients is a textbook example of an ‘unnatural mapping.’ It’s as if you’re a chef and you have a row of knobs controlling a row of cooktop burners and then someone, without your input, keeps randomly rewiring the knobs so that they control other burners.


  19. As a software development ‘chick’ who has worked on dozens of major consumer applications in Silicon Valley, I recently came over to the EHR world. I remember my initial shock seeing the mazelike pathways, overcrowded real estate and myriad drop down lists in even the big name EHR products UI. I was astonished that vendors were able to foist SW that was not built for ease of use on such a highly educated user base. I am relived to have come across this thread to get some context how the current state of EHR came to pass in the same country where Apple turns out products that people respond to so positively/emotionally that they keep them, if not in their bed, then at least nearby.

  20. muddslide,

    I made a similar transition about 4 years ago and I share your sentiments. There is a deep need and desire for change in HIT and I’m grateful to be a part of an industry that can benefit from my contributions as an Interaction Designer.

    I believe the momentum for positive change is building. One example of this is the 2nd Annual Healthcare Experience Design Conference which will be held in Boston this March (http://www.healthcareexperiencedesign.com). It’ll be my first year attending and I’m hoping that it will continue to grow.

    Another example that I think is worth mentioning is Sam Basta’s Healthcare Innovation by Design (www.healthcareinnovationbydesign.com). This community is broader in scope and covers policy issues and the overall experience of patients and practitioners in healthcare systems, and I find that there’s much to be gleaned from the perspectives shared there.

    It’s certainly an exciting time to be a part of this industry and I welcome you to the small but growing community of professionals that care about and work toward better experiences in HIT.

  21. “I believe the momentum for positive change is building. One example of this is the 2nd Annual Healthcare Experience Design Conference which will be held in Boston this March (http://www.healthcareexperiencedesign.com).”

    As much as I’d like to believe that momentum for improved usability – at least SAFETY – of EHR systems is growing, I am pessimistic.

    Poor usability and safety of EHR systems isn’t an innovation problem. Nor will it be solved from the bottom up through challenges and pep talks, even from government officials.

    The federal government – through certification requirements or direct regulation – or perhaps large, powerful groups of health system purchasers (less likely) demanding safe and usable systems will be needed to truly drive change.

    Funny how no one from the IOM Committee on Health IT and Patient Safety is on the panel…

  22. muddslide, Vicente, and QA — thanks so much for your comments and Vicente, thanks for the links as well.

    As you all know, while the subject matter of EHRs can be daunting, the basic interaction design principles for EHRs are the same as for other fields because there is only one design for the human brain. What is needed is a critical mass of interaction designers working in healthcare.

    QA, the safety issue is critically important and not easily solved. I strongly believe, however, that good design in itself greatly reduces errors.

    Along these lines, a 2009 report from the National Research Council found that “healthcare IT programs provide little support for the cognitive tasks of the clinicians or the workflow of the people who must actually use the system. Moreover, these applications do not take advantage of human-computer interaction [HCI] principles, leading to poor designs that can increase the chance of error . . . ´
    http://books.nap.edu/openbook.php?record_id=12572&page=R1 , page S-3.


  23. Having said what I said above, allow me to interrupt this slug fest a little bit.

    EMRs are not in any way similar to consumer apps from Silicon Valley. They are not comparable to Turbo Tax, or a shopping website, or a social media website. Actually, EMRs are not consumer apps at all. They are enterprise transactional software. If you want to compare to state of the art in that field, SAP would be a good place to start. Yes, I know about salesforce.com, but that one only handles a tiny piece of the business, and you still need a financial package and a supply chain package, etc.

    Sure, EMRs can be improved (everything can be improved), but don’t underestimate what the task at hand is, and it is NOT another iPad gizmo.

  24. I agree, Margalit 🙂 with the caveat that “use error” in most BPM/ERP software won’t result in harm or death.

    One of the real challenges of building – or evaluating – safe and usable EMRs is that measures and assessments of ease of use and productivity miss the most critical design aspects of EMR software (i.e. safety).

    Although this is difficult, one would assume that EMR vendors are conducting such assessments of usability (including safety) BEFORE they release their products to healthcare organizations. What tests or evaluations are they using? This shouldn’t be a trade secret…

    Rick – one must measure the usability of software to know whether new, interesting or novel designs actually result in improved usability (including safety). For example, does embedding advertisements in an EMR display distract clinicians / result in decreased productivity / more or different kinds of errors? What might seem like a good design in other domains might not stand up to healthcare use cases.


  25. My turn to agree, QA 🙂

    It is about safety (mostly), and as such, I believe this is something the FDA should take responsibility for.
    Maybe EMRs don’t warrant a Class III device status, but they should come up with something and it shouldn’t be just guidelines on where to place buttons and how to solicit input from users.
    Usability is just one part of safety testing. Functionality, form factor and proliferation of plain old bugs are as important. I think we need one solid roof over all these things.

  26. Margalit, QA, muddslide, and Vicente —

    Thanks for all your comments! I don’t see these last comments as being a slug fest. I don’t hear anyone proposing that designing an EHR interface is not orders of magnitude more difficult than designing a consumer app or proposing that evaluating usability and safety is not critically important. Conversely, I don’t hear anyone proposing that better user interfaces don’t have the potential to reduce cognitive effort and to improve safety.

    While in theory I am in favor of the FDA regulating EHR safety, it’s not clear to me how this would sort out in practice. I fear that more federal regulations for EHRs would add, without any proven benefit, yet another layer of requirements and documentation for our already beleaguered clinicians.

    I find it hard to discuss these issues abstractly. I need concrete examples. To this end, my posts will include lots of actual examples of widely used EHR design paradigms. New post coming soon!


  27. Rich – the “concrete examples” that would be most helpful are real world incident reports (reported to vendors) of “use error” and near misses.

    Or summative testing results from vendors for the next release of their products.

    This isn’t going to happen voluntarily…

  28. Forgive the long comment…but hey, no one is forcing you to read this.

    But to be nice…here’s an executive summary:
    • The general populace is beginning to recognize the importance and value of design
    • Demand is growing for well designed products
    • Consumer driven response to design has crossed over into healthcare
    • It’s not the “end all, be all” answer, but more designers are needed in healthcare

    QA, I certainly understand your pessimism, especially considering the long track record of EHR “unusability”. However, in the consumer market, we’re seeing a growing demand for well designed products and this influence is clearly crossing the barrier into the enterprise healthcare realm. The introduction of products like the iPad and the goal-directed workflows of well designed applications have seeped into the cultural consciousness and have created a stark contrast between what EHRs are now and what they could be given the proper considerations.

    I don’t believe the answers will be found in stricter certification requirements or greater federal regulation and oversight (nor do I think they want this responsibility). I’m afraid that this would only create more havoc and, as Rick stated, somehow become yet another unnecessary burden on the clinicians that are already heavily taxed. Ultimately, it will come down to the systems that hospitals and healthcare organizations continue to support with their dollars and there are brilliant teams working now to give them better options in the future.

    So, to your point about usability and EHR safety not being an innovation problem, I’d have to say I disagree. Both the usability and safety of EHRs will require innovative solutions to strike a careful balance between meeting existing and future regulatory requirements while meeting business goals of healthcare organizations, and maintaining the goals of care givers and patients…of which safety should be the highest priority. Further regulation will add greater complexity to an already complex problem.

    “What is needed is a critical mass of interaction designers working in healthcare.”

    I couldn’t agree with Rick more. About a year ago I met with some insiders at two major EHR vendors. One of them told me that they had one “User Interface Designer” that supported 900 developers. The claim seemed outrageous to me at first, but given how terribly designed their products were, I’m now inclined to believe it. The other vendor wasn’t much different with a ratio of about 300 to 1.

    Speaking from personal experience, four years ago I created the UX role at one HIT vendor that had been in business for nearly 27 years (that’s 27 years of software designed by engineers), and supported 7 in-house and 25 off-shore developers. Even with the help of two extremely talented and experienced Business Analysts, it was difficult to keep pace with the development team and many deployments were less than ideal from a usability perspective.

  29. All very good points, Vincente, and I’m won’t disagree that more thought and resources are being dedicated to EMR design, usability and safety than they have in the past. I wouldn’t call any of the leading EMR vendors truly user-centered in their development approaches, but maybe they’ve hired the HF consultant that they’ve used on a contract basis. Or suplemented to the team in Pune with some folks in the US.

    This, unfortunately, has been spurred, in my observation, less by some realization by vendor executives that usability and safety ARE really important (finally – FINALLY! – listening to the pleas of that single “user interface designer” among the hundreds of developers) but by fear of certification / MU / regulatory / legal implications of being completely back-assward.

    Prospective customers still aren’t asking for usability or accessibility in, uh, meaningful ways. So it’s not a real priority.

    I say that this isn’t in innovation problem because there exist plenty of really tranformative, innovative designs for health IT out there. Contest and award winning ones. Ones being used by vendors too small or new to make the final review list for any major purchaser. Design paradigms that are used in other industries or applications that could be adapted to EMRs. Apps that were truly designed from the ground up to leverage the strengths (and limitations) of non-keyboard and mouse interfaces.

    Innovation is out there. The incentives for this industry to truly transform aren’t. In fact…just the opposite.

  30. Vincente – I also agree that the approach that the government has taken to certification / MU (i.e. specification of functions to the nth degree) has been a counterproductive one (especially with regard to innovation) and has been un-necessarily burdensome to all involved.

    All federal oversight or regulatory activities don’t need to follow this misguided approach.

    I would LOVE to see requirements that push (require) vendors to open their systems and data models to allow cross product functionality. I guess then we’d need to all agree on standards. And that’s not gonna happen on it’s own, either…

  31. Yet again! Lovely topic, lovely presenation, and a lovely feedback volley between you and your followers. It is past mid night here and I am sleepy already, or else I would have read the whole deal. I do intend to read it anyway! I have bookmarked it 😉
    Take care,

Founding Sponsors


Platinum Sponsors























































Gold Sponsors











Reader Comments

  • Brian Too: One thumb up for buying cut-rate Easter candy. This is always worth doing. Two thumbs up for taking said candy in to w...
  • ellemennopee87: re: shooting in Ohio ER. I wish I could unsee the horror I just watched. That was the most unprofessional behavior ...
  • mb: So proud of the VA and their ability to break a Cerner implementation. Cerner thought they could just steamroll the VA, ...
  • Was a Community Hospital CIO: ET, EST, & EDT all refer to Eastern time. Arizona & Hawaii are in Mountain & Aleutian-Hawaii time zones resp...
  • B: Came to the comments to say the same thing. Why would you pass up an opportunity to look at a real axolotl?...

Sponsor Quick Links