Tethered to the EHR: Primary Care Physician Workload Assessment Using EHR Event Log Data and Time-Motion Observations

This article by Arndt et al was designed to look at the amount of time spent on Electronic Medical Records (EMRs) in family practice, which has concerned physicians as many healthcare systems abandon paper charts in lieu of this emerging technology. They addressed this concern by examining the EMR experience within one academic health system - University of Wisconsin. They obtained quantitative metadata from the EMR backend, validated by an additional study of observed physician interactions, in order to assess how much time physicians spent on EMR tasks. The authors determined that there was an increase in non-face-to- face communication, to the point that physicians spend more than half their day interacting with the EMR. They point to this as a potential cause of increased physician burnout.

Our discussion shared the concerns of the study authors over the amount of time spent on EMR and how that could lead to burnout. To us, it was clear that physicians would generally rather spend time interacting with patient directly, as medicine has traditionally been a field centered on the physician-patient relationship. We are also aware that EMR use is only going to become more widespread, so it is useful and informative that the study was able to identify inefficiencies in EMR use and ways to improve the process for future practice.

In looking at the study design, we discussed the retrospective nature of the study. The study looked at EMR interactions in a single healthcare system. We do not have direct knowledge of the EMR interface used by this health system, but we expect that, given the current pace of technological innovation, there could have been several changes and updates to the EMR in those three years. Any changes could alter the amount of time spent on EMR tasks as some workflows are improved and new workflows are introduced and re-learned. This study did not consider these confounding variables. The time study used to validate the results served as an attempt to resolve these variables by attempting to show that their retrospective results correlate with prospective observational data. While this does introduce the potential for observer bias, our discussion did think that this was necessary to accurately classify workflow tasks, as the authors did. Indeed, it would be difficult to design a prospective study that could blind physicians to their participation while being observed. Even without blinding, extending the time study into a larger prospective study over the same three years as the retrospective study would require many more resources, especially an increased number of direct observers to accurately classify time spent on both patient and EMR tasks.

Our concerns around physician recruitment revolved around the idea that physicians might be more willing to participate in the study if they already have strong feelings about EMR use. These strong feelings tend to be negative from anecdotal experience, as those who have trouble with the switch to EMR would be more willing to look into EMR use, which could introduce bias into study recruitment. We also surmised that physicians who had practiced longer and were more accustomed to paper charts would have more difficulty with the transition than physicians who have practiced for a shorter period of time and had known only electronic charting in their practice. The familiarity and comfort with using EMR would have implications on the study's applicability for future practice with EMR. While the authors did recruit family physicians with a wide variety of clinical experience and previous years of EMR use, as evidenced in Table 2, there was no attempt to stratify the results to see if less years of EMR experience or more years of paper chart experience correlated with increased time on EMR tasks. We acknowledge that there was not enough data collected to provide meaningful subtype analysis, but believe that prior experience within and without this EMR system was a factor that should be looked at. Such analysis could help elucidate whether younger physicians, accustomed to only EMR use, can navigate EMR more efficiently and reduce time spent on EMR tasks, or if the problem will persist irrespective of this factor. The group also discussed the implications of physician multitasking while on EMR. We are aware that many physicians have electronic charts open during patient visits, often typing notes from the patient interview in front of patients. If a physician stops typing to ask the patient more information, this could prolong the amount of time spent typing that note as determined by metadata. While the study did accommodate this by setting a cutoff of maximum of 90 seconds between software interactions for classification of time as "time spent on EMR," we thought this would still capture instances of physician multitasking with the patient. To us, this is not a problem unique to EMR use, as paper notes were written during patient interviews before EMRs. This line of reasoning led us to wonder which tasks on Table 3 are inherent to EMR use and which are not. In fact, some current EMR tasks are the result of increased documentation requirements, which could account for the 23% of time listed under "Documentation." Other tasks are obviously inherent to EMRs (such as "system security"). Many, however, are the same tasks that had to be done before, including much of the 32.1% of time classified as "Patient Care." Physicians have reviewed charts, laboratory results, and imaging reports since before EMR, and these tasks certainly took away from patient contact. Considering the total time captured on EMR as completely separate from patient contact can cause problems, as there is overlap that is not taken into account.

We found Figure 2 interesting in that EMR use was lower on weekends, with the exception of the 18-24 hour time period. The authors did not discuss any potential reasons for this weekend spike, yet our group speculated on the reasons for this. We figured that chart review in preparation for the upcoming workweek and the next day's patients could account for this. Had the authors looked into this, it could have strengthened the argument that, as physicians try to preserve time spent on patient interaction, EMR use takes away from physician's time outside the office. Information pertaining to whether the EMR access was from an alternate computer or an office computer would have been helpful here.

Our group's ultimate concern was the generalizability of this study. We noted that there are several EMR companies in use, and that there is variability between EMR's from the same company. Given this, we thought that one healthcare system could have improved, efficient workflows compared to another. This study can serve as a case study for why different systems should look into their own EMR use to look for where efficiencies can be made, and makes the case for an internal audit. This study provides a model for other health systems to use when identifying physicians who struggle with EMRs or to improve workflows. However, we felt that this study alone does not make the case that EMR use is significantly detracting from direct patient care and leading to physician burnout. There are a variety of other factors that need to be taken into account, such as the need to see (and document for) more patients in less time for economic reasons, something especially prevalent in primary care settings. These factors, in addition to the previously mentioned factors, led our group to conclude that this study does not make a strong case for increased EMR use leading to burnout.

Competing interests: None declared