Interruptions, Usability and EMR Workflow

Short Link: http://j.mp/84ujmB

An interruption–is there anything more dreaded than, just when you are beginning to experience optimal flow, a higher priority task interrupts your concentration. This is ironic, since so much of work-a-day ambulatory medicine is essentially interrupt-driven (to borrow from computer terminology, see also). Unexpected higher priority tasks and emergencies *should* interrupt lower priority scheduled tasks. Though at the end of the day, ideally, you’ve accomplished all your tasks.

In my last post on aviation human factors and EMR workflow and usability I described an interruption-based action loop in which staff routinely by step away from one sequence of tasks to work on another higher priority sequence of tasks. Staff eventually do accomplish lower priority interrupted tasks through reliance on colorful information-packed items in what’s called a “radar view” by usability engineers). A continually updated number, the number of minutes since the task began, helps prioritize order of task accomplishment.

Example of a Radar View in an EMR Workflow System

Interruption Theory is an important area of research and the medical domain is well represented there. For example, here are over 600 research papers on the topic (many of which are aviation or medically related). Interruptions can lead to failures to complete medical tasks, which may lead to errors that affect patient safety.

1. “Rothschild et al….found that over 50% of all healthcare errors noted were slips and lapses, such as not executing an intended action.” (see below)

Interruption theory and research in healthcare primarily addresses medical errors and patient safety. However, much about what we understand about interruptions also applies to achieving fast, accurate data and order entry in EMR workflow systems. Simply put, the faster and more accurately one can enter data and orders into an EMR, in the face of interruptions that threaten this speed and accuracy, the fewer errors. Whatever contributes to fast, accurate use on one hand and fewer errors on the other usually serves both goals.

Interruptions in Healthcare: Theoretical Views

A good place to start is “Interruptions in Healthcare: Theoretical Views,” by Grundgeiger and Sanderson (by the way, I took an excellent class on human factors from Dr. Sanderson when I was a graduate student in industrial engineering). The results of interruption research support an especially compelling argument for EMR workflow systems.

First G&S describe prospective memory (PM), which occurs when “a person must recall an intention or plan in the future without an agent telling them to do so.” There are two dominant views of PM:

2. “The monitoring view proposes that when a person forms an intention…to monitor the environment for a specific cue that will remind them to act….”

3. “[T]he automatic associ­ation view proposes that when a person forms an intention, an association is formed automatically between the intention and the reminding cue….When the cue is later encoun­tered, a spontaneous retrieval process brings the intention into mind.”

And:

4. “[N]urses ‘stack’ activities in memory—moving on to a next activity and coming back to stacked activity later—to prevent down times”

From a section entitled Interruptions in the medical and aviation domain G&S write:

5. “Similarities between the domains have been shown in tasks consisting of multiple well defined steps.”

6. “Pilots have a highly structured workplace that supports the use of workflow tools and check­lists”

(I am reminded of this blog post by a physician reflected on the checklists used by her pilot husband.)

In their section entitled Informatics implications of prospective memory G&S advise:

7. “IT systems could remind people what they were doing before the interruption by providing cues on the display.”

8. “IT systems either should be designed in a way that makes them sensitive to the possibility of interfering with working memory processes or they should use non-interfering means of output.”

9. “[O]ffer solutions that help healthcare workers execute plans. For example, electronically accessible work lists that integrate system inputs from different healthcare workers in a timely fashion would provide non-disturbing reminders.”

10. “[S]upport tailoring—the possibility for the user to make modifications that preserve awareness of intended actions or that produce reminders on demand”

11. “[M]itigate memory limitations by providing cues or explicit reminders.”

12. “Distributed cognition is a promising approach in understanding and designing support for many kinds of tasks involving PM.”

13. “[N]otify a person without interrupting him, so removing the load of remembering from the sender and creating a reminder for the recipient.”

14. “Currently, attention aware systems may not be as helpful in healthcare as context-mediated social awareness.”

15. “[H]ealthcare IT systems should be designed and evaluated with regard to their potential disruptiveness (especially on working memory processes) and their potential to provide cues and non-interruptive reminders for intended tasks.”

16. “[A]rtifacts used to manage the effect of interruptions might be useful for designing new technological products. For example, IT products could support cognitive processes disrupted by interruptions or the results could be used for interruption management”

And finally:

17. “[M]ake the workplace resilient to the effects of interruptions. This recommendation has the advantage of preserving the potential positive effects of interruptions, because it does not prevent interruptions. Resilience is enhanced if the burden of resuming an interrupted task is not the PM task of just a single person (e.g., a nurse), but of the interrupter as well, or of the unit as a whole. The PM task needs to be a distributed prospective memory task in the sense that multiple agents (other nurses, equipment, IT) remind the nurse of the intended task. “

Interruption Management Using an EMR Workflow System

Consistent with the ideas and advice in G&S (numbers in parentheses correspond to previous quotes), an EMR workflow system:

Provides environmental cues to remind automatically (via the radar view) pediatric and primary care staff to resume interrupted tasks (2, 3).

Represents “stacks” of nursing tasks so the entire care team works together to make sure that interrupted tasks are eventually and appropriately resumed (4).

Transforms the pediatric and primary care domain into one that more closely resembles the aviation domain, with its well-defined steps, checklists, workflow tools (5, 6).

“[Re]mind[s] people what they were doing before the interruption by providing cues on the display” (7)

“[U]se[s] non-interfering means of output” (8)

“Help[s] healthcare workers execute plans” (through use of what are called workflow or process definitions in the workflow and business process management industry) (9)

“[S]upport[s] tailoring—the possibility for the user to make modifications that preserve awareness of intended actions or that produce reminders on demand” (an EMR workflow system allows workplan customization as well as user actions affecting workplan execution: postpone, cancel, forward, etc.) (10)

Exploits distributed cognitive processes (see below) (12)

Notifies without interrupting (via radar view and To-Do lists) (13)

Promotes “context-mediated social awareness” (through use of the radar view and user awareness of consistently executed workflow process definitions) (14)

Supports cognitive processes with non-interruptive reminders (15, 16)

Enhances the resilience of the entire pediatric and primary care human-computer cognitive system by “preserving the potential positive effects of interruptions.” An EMR workflow system converts a prospective memory task into a “distributed prospective memory task.” Each member of the healthcare team efficiently and effectively resumes and accomplishes previously interrupted tasks. (17)

I am reminded of what I wrote (in the proceedings of the 2004 TEPR conference, and exercising the blogger’s prerogative to quote myself) about the important connection between distributed cognition and EHR workflow management systems:

[begin 2004 TEPR quote]

EHR Workflow Management Systems: Essentials, History, Healthcare

…

Workflow Management and EHR Usability

EHR workflow management concepts mesh with research initiatives to improve EHR usability. For example, Human-Centered Distributed Information Design [6] (there applied to EHR usability issues) distinguishes four levels of distributed analysis: user, function, task, and representation, which correspond well to workflow management architectural distinctions. Distributed user analysis can be interpreted to include allocation of tasks, relationship between roles, and task-related messaging, all of which are important workflow management concepts.

Distributed function analysis involves high-level relationships among users and system resources. From a workflow management perspective, this includes who reports to whom and who is allowed to accomplish what.

Distributed task analysis roughly corresponds to the creation of process definitions that in turn drive EHR behavior: What is to be accomplished by whom, in what order, and what needs to happen automatically.

Distributed representational analysis corresponds to something that workflow management systems intentionally do not address. Workflow management system design tends to be agnostic about how information is displayed to, transformed, or collected from the user. The underlying workflow engine is intended to be a general purpose tool that can be used to sequentially launch whatever screen or initiate whatever behind the scenes action that the implementer of the workflow system deems most apt as part of workflow analysis and design. However, by remaining orthogonal to the choice of screen, by not mandating or hard coding, the designer/implementer is free to bring to bear the powers of representational analysis to use whatever screen and attendant representation is most appropriate.

Thus, workflow management concepts are consistent with human-centered distributed information design, an important emerging area of medical informatics research. “Task-specific, context-sensitive, and event-related displays are basic elements for implementing HCC [human-centered computing] systems,” (p. 46 [6]) and they are the basic elements provided by EHR workflow management systems, too.

…

References

…

[6] Zhang J, Patel V, Johnson K, Smith J, Malin J. Designing human-centered distributed information systems. IEEE Intelligent Systems 2002: Sept/Oct: 42-47.

[end 2004 TEPR quote]

Conclusion

I don’t think you will be able to find a more compelling, research-based argument that ‘anticipatory’ EMR workflow systems are superior to traditional ‘hunt-and-peck’ EMRs. Traditional EMRs soak up scarce attention and memory and interfere with prospective memory processing. EMR workflow systems free up attention for prospective memory processing while reducing interruptions and providing unobtrusive future cues to resume previously interrupted tasks efficiently and effectively.

You can retrieve the complete PDF for the “Designing human-centered distributed information systems” article here.

The complete reference for Grundgeiger and Sanderson paper that triggered this post is:

Grundgeiger T. & Sanderson P. (2009) Interruptions in healthcare: Theoretical views, International Journal of Medical Informatics, 78 (5), 293-307.

And their complete PDF article is here.