Methods to capture workflow
Workflows within healthcare settings are complex and diverse. They can be studied at levels ranging from the individual end-user, to a clinic, department, or beyond. In the process of implementing any new systems in a healthcare setting, workflow analysis is a crucial task to identify how users currently approach a given task or activity. This can drive insights for aligning the new work processes for users.
Any single approach to capturing a workflow will likely be incomplete to some degree. There are a range of methods to help study clinical workflows, any of which may be appropriate in the right context.
Contents
Introduction
Two methods commonly used by project teams to capture the workflow are the time-motion study and the work sampling study. Additional methods include qualitative questionnaires and quantitative measures to track movement. Since neither can provide a complete illustration on their own, using a combination of approaches (i.e., mixed methods) can provide the most comprehensive information for workflow assessment[1][2].
Time-motion study
Time-motion studies involve observation of end users as they perform tasks and record the time spent on each activity. This method has been considered the gold standard for workflow study.
The investigator follows the subject and records the temporal aspects of events (e.g. tasks) under evaluation. This method may be known as the "stopwatch method" due to use of stopwatches, however mobile apps and/or specialized software may be used to record observations.
Time-motion studies give detailed descriptions of workflow processes of the end user. However, it is a tedious and labor-intensive process, often involving 1:1 observation for long durations. It may be difficult to identify willing participants and coordinate schedules. Thus, this is a costly method of workflow capture.
In an article by Zheng et al.[3], the authors used time-motion methods to create a checklist called STAMP (Suggested Time and Motion Procedures). This study highlighted an additional issue of time-motion studies: complicated workflows and interruptions are frequent in healthcare and may impose limits on accurate analysis.
Audit log analysis
Audit logs capture detailed records of all user interactions with the EHR system. This includes actions such as accessing patient records, updating information, writing a note, or ordering a test.
Audit logging is a required function of all current EHR platforms. US regulations require tracking of at least four pieces of information for each access of patient records: who accessed which record at what time and the action performed in that record[4]. Many individual EHR vendors track additional information at high levels of granularity. The logs were originally designed for monitoring user access, tracking user behavior, and detecting unauthorized access. However, these logs are finding secondary uses.
A systematic review of literature on EHR audit logs identified common themes in their analyses ranging from studies of EHR use directly, clinical workflows extending beyond the EHR, and care team dynamics[4]. Others have used timestamps from audit logs to derive timing of workflow events[5].
Strengths
In contrast to time-motion studies, which generally require continuous observation by an external observer, audit logs are always active and capture data for every EHR user. Time-motion studies are difficult to carry out at a large scale due to being time-intensive and costly. After developing an initial algorithm, a workflow analysis with audit logs can be easily scaled up without incurring significant additional time or cost. Additionally, analysis of audit logs may reduce incidence of the Hawthorne effect that could otherwise affect direct observation methods. Audit logs also enable comparison between individual clinicians or roles.
Limitations
There are limitations, since the logs were not purpose-built to track workflows. It may be difficult to map logged actions to clinical activities since clinic/hospital workflows are not entirely EHR-based (e.g., a physician is not at the computer while examining a patient). Additionally, system configuration may affect which actions are captured in logs which could affect data points available for extraction and analysis. Similarly, analysis can be challenging due to the complex nature of the logs and noise in the data (e.g., user clicking aimlessly while on the phone to complete a prior authorization).
Examples
One such use is Signal, Epic's provider-efficiency tracking tool. This tool provides insights into how end users interact with the EHR. Its data is sources from aggregated audit log exports and reports are available for any provider in an institution. Importantly, a KLAS report indicates that this data is not predictive of dissatisfaction, provider burnout, or turnover[6]. Rather, it could be used to identify providers who might benefit from additional training.
Work sampling
Work sampling is a method used to observe and record the frequency of specific tasks or activities performed by users over a time interval. This method relies on periodic, random observation rather than continuous monitoring (such as in time-motion studies) to capture a representative picture of how time is spent during clinical work. For example, time on documentation or engaging in specific administrative, non-clinical tasks. This method can be directly observed, or the subject may keep a log on their own.
Work sampling techniques can improve the efficiency of data collection and reduce the required number of observers. It may also limit the effects of the Hawthorne Effect since a subject is not continuously observed.
This method can be unreliable, particularly if the data is self-reported. It may also be limited by the non-continuous observation, and may miss less frequent activities or fail to account for context.
Interviews and surveys
Interviewing and surveys of users in key roles in a workplace can also provide useful information for reconstructing workflows. They can provide useful feedback on their personal experiences. Subjects may participate in interviews (of varying structure), during which they detail their workflow, or complete surveys detailing their workflows[1][2]. These allow for data collection from a wide range of users, which can identify common workflow challenges.
However, there are limitations to this method. These are cross-sectional by nature, providing information from one point of view at any given time. Although this could be solved with more investigators collecting samples at one time, too many observers can lead to disruptions in the workflow[2].
Automatic movement tracking
Automatic movement tracking utilizes portable devices that are attached to subjects. The subjects’ movement throughout their workday is monitored. By identifying and tagging certain workstations before the tracking process, an investigator can deduce the type of work being done at time intervals preset in the devices. This method is most similar to a time motion study, yet allows for an investigator to monitor the time and movement of multiple subjects at one time. The investigator is only limited by the number of devices, and the operability of these devices.[1]
Once the data is compiled, it can be used in combination with data from other methods. This comprehensive approach can be used to create a 3D or virtual models of a workflow.[1].
References
- ↑ 1.0 1.1 1.2 1.3 Toward automated workflow analysis and visualization in clinical environments. Mithra Vankipuram, Kanav Kahol, Trevor Cohen, Vimla L. Patel. J Biomed Inform. 2011 Jun;44(3):442-440.
- ↑ 2.0 2.1 2.2 Fast, Formal, & Beautiful: Effectively Capture, Document, and Communicate User Workflow Information for Designing Complex Healthcare Software Systems. Jean M. R. Costa, Xianjun Sam Zheng, Roberto S. Silva Filho, Xiping Song. Proceedings of the Human Factors and Ergonomics Society Annual Meeting. 2012 Sep;56(1):526-530.
- ↑ Using the time and motion method to study clinical work processes and workflow: methodological inconsistencies and a call for standardized research. Kai Zheng, Michael Guo, David Hanauer. J Am Med Inform Assoc. 2011 September; 18(5): 704–710.
- ↑ 4.0 4.1 Rule A, Kannampallil T, Hribar MR, et al. Guidance for reporting analyses of metadata on electronic health record use. J Am Med Inform Assoc. 2023;31(3):784-789. doi:10.1093/jamia/ocad254
- ↑ Hribar MR, Read-Brown S, Goldstein IH, et al. Secondary use of electronic health record data for clinical workflow analysis. Journal of the American Medical Informatics Association. 2018;25(1):40-46. doi:10.1093/jamia/ocx098
- ↑ https://klasresearch.com/archcollaborative/report/epic-signal-data-2023/548. Accessed 2025-04-29.
Contributors: Jo Nie Sua, Cody Schindeldecker, Jessica S. Pierre, Nick Evans
