Metrics for assessing physician activity using electronic health record log data | Literature Review

1.       Article

• Title of article: Metrics for assessing physician activity using electronic health record log data

• Authors (full name): Sinsky, Christine A; Rule, Adam; Cohen, Genna; Arndt, Brian G; Tait, D; Sharp, Christopher D; Baxter, Sally L; Tai-seale, Ming; Yan, Sherry; Chen, You; Adler-milstein, Julia; Hribar, Michelle

• Publisher’s name: Oxford University Press

• Publication city: Oxford, United Kingdom

• Date of publication: February 6, 2020

• Number of pages: 5

2.       Summary

The article primarily discusses seven proposed core electronic health record (EHR) measures using EHR log data. Given the requirement of all certified EHRs in the United States to have EHR logs that track the activities of a clinician (e.g. accessing a patient record, actions within the record), the authors point out that there is a large opportunity here to leverage EHR log data to help quantify and better understand the clinical environment. More importantly, they noted that a systematic review of 85 studies using EHR log data did not detail the methodology used in how measures were calculated. Thus, the seven proposed core EHR measures will help promote comparative research analysis.

In selection of the seven measures, there were a few guiding considerations. The first of which is the use of time as a unit of measure given the limited supply of it within healthcare professionals. Secondly, normalization to an 8-hour period was important to help compare across different specialties. Third, the measures were designed towards simplicity rather than accounting for all the details that may complicate these measures and make them unwieldy. Lastly, the authors note the limitation that these measures may not be applicable to inpatient units and are currently only focused on physicians. The seven proposed core measures are: 1) total EHR time, work outside of work (WOW), time on documentation, time on prescriptions, inbox time, teamwork for orders, and undivided attention patients receive from their physicians (ATTN).

3.       Discussion of authors

All the authors have advanced degrees in medicine, bioinformatics, epidemiology, computer science or statistics and all of them have previous work related to electronic health records. Given their qualifications, each author would appear to be qualified as a primary author or solo author on this topic. The lead author, Dr. Christine A. Sinsky, is the Vice President of Professional Satisfaction at the American Medical Association and has been working on physician well-being since 2011 at the national level. While each author is well-qualified on this topic, the diversity of their experiences as evidenced by their advanced degrees, provide a synergistic approach to tackling the problem of quantifying EHR burden.

4.       Summarize strong points of the article

The article has a few compelling points that stood out. The first of which is the case the authors make about the current landscape of measures used to quantify the time clinicians spend on the EHR. Essentially, despite 85 studies on EHR log data, definitions are not standard and the ability to benchmark and compare results across organizations are not feasible. Moreover, the ubiquitous nature of EHRs in the United States coupled with the regulatory requirement of maintaining EHR log data for certified EHRs makes a great case for capitalizing on this opportunity. Secondly, the article goes into detail about their seven proposed core EHR measures, their rationale for choosing it, and how it can be applied to practice decisions. Lastly, I thought their points about the gold standard for validating EHR log data (e.g. direct observation time-motion analysis) was an important one as they noted how some of the EHR log studies did not include validation.

5.       Summarize weak points of the article

While the authors did have some convincing points about using their proposed EHR log data measures, there were quite a few limitations as well. The first issue is the applicability of these core measures toward inpatient care. The authors noted that the measures primarily apply to ambulatory settings and shouldn’t be used, without modification, for inpatient care. However, despite the impetus of this article (i.e. convergence on the definitions), the authors make no suggestion on how these measures should be modified to account for inpatient settings. Secondly, while simplicity is important and the proposed measures would likely evolve over time, the nuances in clinical care across specialties would more than likely result in objective differences in the EHR log data. Certain specialties may spend more time on inbox messages or time on prescriptions. This ties into the third point which is arguably the most important: not normalizing to patient volumes. Workflows commonly differ from one implementation of an EHR to another, even if it’s the same EHR vendor. Thus, simply measuring time and normalizing against time does not account for the implementation differences across institutions, which certainly exists. This introduces the question of whether the time spent on EHR tasks are due to EHR burden vs. patient volume. Albeit, the authors themselves did point this out and provided a solution of also reporting average clinic volumes along with other measures of workflow.

6.       Share who should read this article?

The targeted audience for this article is likely to be EHR researchers primarily given the need to standardize the definitions of EHR log data measures. However, the authors likely expected other individuals in the areas of clinical informatics and health IT policy to also take interest in this article. The former is intuitive as individuals that configure EHRs would be interested. The latter is also important as these types of articles have the potential to influence reimbursement policy in the form of incentives or other alternative payments. Background required to understand the material in this article would likely include knowledge in informatics, EHRs, and basic research and clinical workflows. 

7.       Why should someone read or not read this article?

Considering the clinician burnout crisis and its potential association with EHRs, solutions that help to quantify burnout attributable to EHRs and allow researchers to compare effectiveness of interventions across organizations, is absolutely welcomed. As noted above, the authors point out the importance of standardizing definitions used and presents 7 core EHR log data measures that everyone can adopt and use. Anyone involved in tackling the clinician burnout crisis could gain a lot of great ideas for implementation at their own institutions.

8.       How would you improve the article

There are two major things I would have liked to see in the article that was missing. The first of which is a table or list of all the EHR log measures used by the other 85 studies. While compilation of that list may be quite tedious, an article that proposes new EHR log measures should provide the audience details into ones that have already been published. Secondly, I would have liked to see some preliminary validation of these seven proposed measures. Despite the authors criticism of measures used in other studies, I found their argument to use these measures less compelling when they haven’t validated the proposed measures themselves.

9.       Quotable Quote

“While EHR log data has shown promise in measuring time spent on these clinical activities, the use of EHR log data to further under- stand the clinical environment is a nascent science”