An interface-driven analysis of user interactions with an electronic health records system
This is a review of An Interface-driven Analysis of User Interactions with an Electronic Health Record System. 
- 1 First Article Review
- 2 Second Review
- 3 Third review
- 4 Fourth review
- 5 References
First Article Review
This was a study that investigated user interactions or experience with an electronic health records (EHR) system called the Clinical Reminder System (CRS) in terms of its user interface (UI) and application Flow (AF) design.
Do user interface (UI) and application flow (AF) design deficiencies influence end users’ day-to-day clinical practice? Does a poor fit between UI/AF designs influence the way clinicians navigate through an electronic health record (EHR) system? Can user behavior be used to improve the system’s usability?
Environment The study site used was West Penn Medical Associates, an ambulatory primary care clinic at Western Pennsylvania hospital, a teaching hospital where all primary users were internal medicine residents.
Recurring UI navigational patterns were uncovered using sequential pattern analysis (SPA) and a first-order Markov chain model. SPA was used to search for recurring patterns; consecutive EHR features access events at given points in time. EHR usage patterns and data were recorded and analyzed during a 10-month period.
Using the EHR’s transaction database, the study utilized computer-recorded event sequences of different features of the EHR by users to show patterns in the way they were accessed.
In all, seventeen main EHR features were recorded. During the study period, users of CRS recorded 973 patient encounters. Using the transaction database of CRS, sequential patterns across encounter sessions were constructed and data analyzed. Patterns of consecutive feature access events were recorded; eleven were identified as maximal with levels of support ranging from 15 to 51.16 percent; Within-session recurring rates of sequential patterns showed levels of support between 51.35 and 70.22 percent.
User interface (UI) and application flow (AF) design deficiencies may influence end users’ day-to-day clinical practice. Poor fit between UI/AF designs may influence the way clinicians navigate through an electronic health record (EHR) system. Users demonstrated consistent UI navigational patterns when performing different clinical tasks. Some of these patterns deviated from the EHR system’s original UI/AF design principles. User behavior and such deviations should be considered when designing health information technology (HIT) systems.
UI/AF design has the tendency to influence how the user navigates an EHR system. UI navigational patterns might have an effect on clinical practice. End users may not have the full benefits of an EHR implementation from poorly designed UI and AF. When this happens, decreased time efficiency, coupled with user dissatisfaction can adversely affect the quality of care and patient safety. Superior user experience is eminent with appealing, intuitive UI and AF designs. User behavior can be used to improve an EHR system’s usability.
The study objective was to expose and evaluate the different patterns in user interactions (UI) and applications flow (AF) of an Information System namely Clinical Reminder System (CRS) which was re-engineered to meet the clinical practice end users. 
EHR was integrated to capture comprehensive UI interaction events. The events were recorded based on time in sequences. The study site was an ambulatory primary care clinic in West Penn Medical Associates, part of Western Pennsylvania Hospital. EHR's were primarily accessed by Internal medicine residents.
Over 10 month period, sequences of events for usage of generic EHR features in CRA were retrieved from the EHR’s transaction database, which are feature labelled and chronologically ordered.
The study used Sequential Pattern Analysis (SPA) and First-order Markov Chain Analysis, along with Analysis of Within-session Recurrence Rates to evaluate recurring UI navigational patterns.
Of the 17 main features identified, 70% of them consisted of “Assessment and Plan” (21.18%), “Order” (17.17%), “Diagnosis” (16.36%), and “Medication” (14.53%). The SPA uncovered 11 maximal sequential patterns and the most frequently used packed features were “Assessment and Plan” and “Diagnosis,” “Order” and “Medication,” and “Order” and “Laboratory Test.”. These were consistent when assessed within-session recurrence rate. The 7-Step Markov Chain displayed an EHR Feature Spectrum, where “History of Present Illness”, “Social History”, “Assessment and Plan” were consistently accessed among other features.
Two important areas were observed for the imbalance between the in practice user’s navigational pathway and anticipated pathway in the original UI/AF designed.
- One was unexpected individual user discernment in usage of features rather than following the standards.
- Another was easy and immediate accessibility on screen for most important features to be placed in such a manner that there is smooth flow of sequence of patterns. 
Although the study highlighted the consistency in navigational patterns of different clinical tasks which was intended as per the designed UI/AF but the unexpected user behavior should be evaluated and any benefits if recognized, incorporated for the improvement of EHR features. Also deviations from recommended standards by clinicians to their actual clinical practice must be carefully addressed in the health IT design and implementation processes. 
I am attracted to the UI layout features a unique design in which all essential EHR features are placed in a single workspace in view of cognitive continuity when users navigate from one EHR feature to another. It could alleviate “fragmented displays” and “hidden information” problems. 
This article is a qualitative study that evaluates a research team’s initial implementation to a clinical decision support(CDS) system. The team includes researchers & practitioners at the Western Pennsylvania Hospital. User interface design also applies to mHealth domain(MHealth consumer apps is one source regarding this).
Electronic health record (EHR) data usage was analyzed for 10 months. The study included/ was measured by event sequences, pattern, chain analysis, usage rate and pattern encounters.
The “empirical data” showed popular pathways: Assessment and Plan, Order, Diagnosis and Medication Side Effects. Some of the most frequently assessed features were Retaking BP, Procedure, and Encounter Memo. Those features made up 70% to all EHR user interactions.
This research brought negatives to light. Poor user interface and application flow are the main reasons why many Health IT implementations fail. The results show hidden patterns such as those demonstrated by clinicians when they were navigating. Unanticipated patterns gave understanding to user behavior and recommendation standards. Overall, lessons were learned and the system was re-engineered with designs to fix the issues. See also User centered design. Another aspect of user interface to consider is whether or not to use single or dual monitors and what is the preferred dimensions of those monitors. Research on this topic was performed by Lee, K-H., Bae, W.K., Han, J.S., Yoo, S., Kim, J.S., Yun, J., Baek, H.Y., Baek, R-M., & Hwang, H. (2012) and was titled "Monitor Preference for Electronic Medical Record in Outpatient Clinic".
I think this was a great research article. The department I work in manages hospital owned ambulatory clinics and I have had exposure to more than 8 EHRs. They all behave differently. The workflow is similar but I hear complaints from the physicians and staff all the time. I feel this paper focused on the clinician factors of the system and not so much patterns or activity of the front staff, office managers and the technological analysts. Even though their input isn’t as important as the clinicians or the providers, their patters could have identified areas of improvement for registration, billing and scheduling.
To develop a clinical decision-support system for the improvement of the hospital’s internal medicine resident training program. At Western Pennsylvania Hospital they have design a homegrown EHR system that is used to schedule appointments, workflow management, pre-encounter assessments, and document clinical findings, prescribe medications, and enter orders.
Over a 10 month period, From October 01, 2005 to August 01, 2006. Analysis was performed on the hospital’s EHR data usage. The focused was to study EHR patterns. Several different analyses were performed during the 10 month.
- Construction of event sequences
- Sequential Pattern
- Within-session recurrence rates
- First-order Markov Chain
It identified several main features of the user interactions navigational patterns. The Features included assessment and plan, diagnosis, order and medication, and order and laboratory testing.
To have an effective user interface all systems most work together to provide the facility and its patient’s better quality.
- Zheng, K., Padman, R., Johnson, M., & Diamond, H. (2009). An Interface-driven Analysis of User Interactions with an Electronic Health Record System. http://jamia.oxfordjournals.org/content/16/2/228