Letter to the Editor: Secondary use of electronic health record data for clinical workflow analysis

To the Editor:

Hribar et al¹ provide us with a great summary of their work using electronic health record data for clinical workflow analysis.

At the heart of the matter is how Electronic Health Record (EHR) timing data is processed and validated for use in interpreting and improving clinical, operational, and EHR work flows. A principle question being to what degree should EHR timing data be independently validated for such studies at new sites, for different specialties, and after a process improvement is put in place. What form should this validation take if it is indicated—work sampling, full on time and motion studies, or something else entirely?

Expansion of this methodology to other ophthalmology practices and especially to other specialties presents unique challenges. We have seen great heterogeneity in both workflows and timestamp variability due to differences in initial EHR configuration or differences inherent in vendor products. Different medical or surgical specialties, practice environments with different levels of support, use of Midlevel Providers and medical scribes, and practitioner work styles add complexity not addressed in the authors’ present study.

The authors do not include granular EHR workflow observational data for this study, indicating the actual EHR tasks being done, but rather use the time duration for access to the medical record software. Granular detail of EHR-based work tasks would include a few minimal components such as reviewing existing records, entering new data, editing patient history, physical exam documentation, formulation of assessment and care planning, diagnostic study review and new test ordering, prescribing, referral and consult management, etc. We have found that the use of EHR timestamps to define EHR use can be problematic because of lags between clinician cognitive processing and actual data entry. Many clinicians have not fully integrated EHR related tasks into their present clinical work routines, many choosing alternative digital work styles as perhaps exemplified by Physician 2 in the Hribar et al study. We have seen a similar disagreement in our own work with clinicians due to the amount of time that is required for EHR data entry. This discrepancy, especially in high acuity settings, can become quite a problem when physicians cannot complete their documentation until hours or even days later due to the time constraints of their patient care requirements. This can make accurate data analysis in the short-term very difficult. Many times it will force a choice between either retrospective analysis or exclusion of incomplete data.

Understanding why this is the case is critical to improving EHR adoption, efficacy, and ultimately workflows built around them. As suggested by the author’s own research and other recent studies in ophthalmology, EHR based work still takes an inordinate amount of time.²

We agree, the above concerns notwithstanding, that EHR derived timing data can be used beneficially in modeling and simulation efforts and offer, as an example, our experience using a system dynamics analytic modeling and simulation as another approach in addition to discrete event techniques.³

References