User Expectations and EHR Design

Ideally, an EHR system should make one more efficient and productive.  Under less-than-ideal circumstances, workarounds, errors and “gotchas” appear and complaints mount.   Addressing EHR design issues requires mapping complaints to specific EHR features and design decisions.  However, mapping is not necessarily a straightforward process.  Some complaints are easily mapped (e.g., the print is too small) while others present more of a challenge (e.g., offering provider-specific problem list views).  Improving EHR designs requires both a list of complaints and a map.   Software vendors necessarily expend energy to collect and categorize complaints, but their findings are not shared, and as a result, EHR design research limps forward.  Fortunately, over the last few years, more informatics researchers have turned their attention to EHR system design flaws, and while the number of papers is still small, the results are useful (1,2,3).

In the post EHR Design and Personal Work Habits, I discussed two papers that looked at the types of workarounds that occur with EHR systems (4, 5).  Both papers provide categorizations of workflow adjustments and offer information that helps with the mapping problem.  Smith and Koppel (6), in their paper Healthcare Information Technology’s Relativity Problems: A Typology of How Patients’ Physical Reality, Clinicians’ Mental Models, and Healthcare Information Technology Differ, offer additional mapping assistance.
In the paper’s introduction, the authors make the following statement.

The goal of useable, effective, safe and interoperable healthcare information technology (HIT)  remains difficult to achieve.  We suggest one of the barriers to this goal is the temptation to focus on tidy use cases of predictable orderliness, which fail to convey the complex reality of medical care.  Looking at what happens in real HIT-in-use settings yields a large set of scenarios in which things do not work according to design, to original understanding of workflow, or to efficient operation. To improve HIT, we must be able to organize problems into a systematic typology so we can understand and remedy them. This paper seeks to catalog and organize these messy obstacles, and perhaps illuminate structures underlying them—and by doing so, to overcome some of HIT’s significant difficulties.

They go on to identify 45 scenarios that capture key usability disruptions and then organize them into five categories.  Each category is listed below (RW=real world; MM=mental model, COW=computers on wheels).

Type I:  IT too coarse — Significantly different scenarios in RW and/or MM are represented in the same way in IT.  Examples: (1) Problem lists that do not permit sufficient qualification of classifications, for example, left side CVA versus just stroke, or inactive asthma, or, (2) Only indicating the patient has cancer is woefully insufficient to be useful to oncologists

Type II: IT too fine — Scenarios identical to the clinician are represented significantly differently in IT. Examples: (1) Very granular categories within ICD-10 may reflect a level of certainty or understanding that does not exist for a specific patient. The (false) specificity may misguide other clinicians. (2) Unconfirmed suggestion of one very specific subcategory of several possible cancers may lead to premature closure of analysis

Type III: missing reality — Scenarios or scenario details significant to the clinician are not represented at all in IT. Examples: (1) Only lab reports and medications are listed; not symptoms or history. (2) The EMR implicitly assumes COWs are always network connected, but the clinician encounters reality where they are not.

Type IV: multiplicity — Different communities of clinicians may construct different mental models (and hence infer different realities) from the same representation in the IT. Example: the EMR reflects misleading/distracting judgments by staff or family members in addition to many lab reports with alternative interpretations.

Type V: looking glass — When a clinician scenario is reflect into the IT and back, it becomes something rather different and surprising. Example: clearly incorrect sensor data, which a clinician would reject, becomes enshrined in the EMR, which now describes a reality that never existed.

Having looked at the five categories and 45 scenarios, I think the authors offer a great initial stab at identifying common usability issues and tying them to specific EHR features and design decisions.    As the authors state in their introduction, many of these usability misalignments are workflow disruptions.  With the understanding that workflow issues are at the root of many usability problems, it is possible to begin the move toward a standard set of terms and concepts to guide future discussions using workflow patterns.  There are 126 workflow patterns covering the flow and control of task sequences, data movements, and resource interactions (resources include people).

The concept of “state” is essential to workflow.   Processes have a definite beginning and a definite end and produce a specific outcome.  Since many EHR systems do not have explicit representations of processes, they also lack a representation of state.    Lacking process awareness, EHR systems are left with only data stores to keep track of what is occurring.   Unfortunately, data stores are binary–either a data value is present or it isn’t.    Thus, the idea that an event is underway, but not completed has to be managed in software by the developer on an ad-hoc basis.  Workflow engines address these concerns directly and insulate state/process management from the user interface, data stores and other software components.

One can see the value of workflow patterns in analyzing usability issues by using patterns to review the scenarios offered by Smith and Koppel.   For example, Scenario 1-Vomiting, is presented in the paper as follows:

The reality of a patient receiving medicine can be rather messy (literally). For example, consider a patient whose orders indicate she is to receive a pill. The clinician may administer the pill—after which the patient vomits. Has the medicine been administered, or not?

Here, two states are involved—that of the system and that of the patient.   The state of the system should indicate a medication was administered and that, that process is complete.  However, the patient’s state has undergone two changes–she was given a medication, and later she vomited.   A well-designed system would record all three state changes.   The analytical value of workflow patterns in this situation arises from the fact that they make one consider state explicitly when analyzing workflows and designing systems.

The informatics literature is beginning to offer solid data on problems encountered by those using EHR systems.   Workflow patterns are mathematically-based and provide the terms and concepts needed to represent usability issues in a formal, standard manner.  Even better, with Colored Petri nets and YAWL, there are great tools available for rendering usability issues as visual models that are capable of representing state as well as task sequences, data movements, and resource interactions.   Adoption of workflow patterns for software design could help to move EHR design from an art to a science.  Hmmm… EHR science–has a nice ring to it, don’t you think?

  1. Friedman A, Crosson JC, Howard J, Clark EC, Pellerano M, Karsh BT, Crabtree B, Jaén CR, Cohen DJ. A typology of electronic health record workarounds in small-to-medium size primary care practices. J Am Med Inform Assoc. 2013 Jul 31.[E]
  2. Hill RG Jr, Sears LM, Melanson SW. 4000 Clicks: a productivity analysis of electronic medical records in a community hospital ED. Am J Emerg Med. 2013 Sep 20. [E]
  3. Elkin PL, Beuscart-Zephir MC, Pelayo S, Patel V, Nøhr C. The usability-error ontology. Stud Health Technol Inform. 2013;194:91-6.
  4. Saleem JJ, Russ AL, Justice CF, Hagg H, Ebright PR, Woodbridge PA, Doebbeling  BN. Exploring the persistence of paper with the electronic health record. Int J Med Inform. 2009 Sep;78(9):618-28.
  5. Flanagan ME, Saleem JJ, Millitello LG, Russ AL, Doebbeling BN. Paper- and computer-based workarounds to electronic health record use at three benchmark institutions. J Am Med Inform Assoc. 2013 Mar 14. [E]
  6. Smith SW, Koppel R. Healthcare information technology’s relativity problems: a typology of how patients’ physical reality, clinicians’ mental models, and healthcare information technology differ. J Am Med Inform Assoc. 2014 Jan 1;21(1):117-31.


  1. Grat article.

    I’ll come back to this site often as there is a wealth of information!


    Bennett Lauber
    The Usability People, LLC

    1. Glad you find the site useful. Please spread the word!

Leave a Reply

Your email address will not be published. Required fields are marked *