Background
Implementation of the EMR and the increased ability for patients to view their records transparently and rapidly online have led to dramatic increases in direct access to clinicians through the EMR patient portal.(1) This has increased physicians’ administrative burden and also has been identified as a leading factor in physician burnout, which is characterized by a state of mental exhaustion, depersonalization, and a diminished sense of personal accomplishment.(2,3) The problem is exacerbated by the lack of support for managing clinically oriented messages, especially in the academic setting.(4) A minority of neurologists (23%) find the time spent on clerical tasks to be “reasonable,” but most (56.3%) report inadequate support to address messages,(5) a factor that has been identified as contributing to physician burnout.(6)
The Goal
The goal of our study was to decrease the number of in-basket messages to determine whether that would improve physician fulfillment and decrease burnout. If the intervention proved successful, it would be easily scalable and financially achievable. We developed a message triage algorithm and trained a non-clinical staff member, with the goals of decreasing the number of in-basket messages reaching participating clinicians and, as a result, improving neurologist career satisfaction. This study was supported by a BWell grant from the hospital physicians’ organization.
Pilot Execution
The Leadership Team
The authors are senior academic neurologists in leadership positions from four divisions: general neurology; headache; epilepsy; and multiple sclerosis. They all have active clinical panels as well.
Message Triage Algorithm
A message triage algorithm was developed, based on the authors’ expertise. The algorithm was revised after input from administrative staff and participating physician subjects. The goals were to decrease the time neurologists spent on in-basket management, improve job satisfaction, and maintain patient care quality, safety, and access (Figure 1).
Participant Recruitment and Baseline Data
Ten neurologists with four or more weekly clinical sessions were approached between September and October 2021. All ten agreed to participate in the study. The neurology participants represented five different divisions and consisted of five women and five men. Neurologist subjects completed the Maslach Burnout Inventory, a validated survey for measuring burnout.(7) One month of the number and types of EMR messages received by each physician was counted for the baseline. The message types that were tracked were patient calls, patient advice requests, and staff messages (Table 1).
Pilot Implementation
Staff Training
Administrative support staff and a medical assistant (MA) were trained on the message triage algorithm intervention by the principal study investigators (CB, MAO). An MA was chosen to participate because of their experience using the medical chart and also because of their lower cost compared with a nurse or physician’s assistant. Training for support staff and the initial MA took two hours each. The principal study neurologists directly trained the MA on how to best triage the in-basket messages. Every in-basket message for the participating neurologists was triaged according to the designed algorithm. All clinical messages went to the neurologists.
Assessment
Within three months of completion of the pilot, the neurologist participants were again given the Maslach Burnout Inventory, as well as a survey about the intervention and whether they perceived it to be helpful. Following one month of the intervention, the total number of patient calls, patient advice requests, and staff messages were tallied to compare with the baseline data.
Metrics and Hurdles
The pilot was delayed due to personnel turnover. One neurologist went on maternity leave before the intervention was completed, so data were acquired from only 90% of the original subjects for the analysis. The initial MA withdrew from the pilot after two weeks for personal reasons, but it became clear that further training was needed. It was not possible to recruit another MA for the pilot, so a research assistant was hired. The research assistant received more intensive training than the initial MA, with approximately six hours of direct training by the principal investigators. The research assistant was also offered the ability to contact those investigators at any point in the day to address any questions or concerns.
Results
At baseline, the nine academic clinical neurologists completed 797 total messages received between July 15 and August 15, 2022. Message types included 527 patient advice requests (59% of the total advice requests), 178 staff messages (20% of the total staff messages), and 192 patient calls (21% of the total number of patient calls) (Figure 2). The baseline Maslach Burnout Inventory scores were averaged after being divided into their three component scores: burnout; depersonalization; and personal achievement. The baseline average burnout component score was 15.75, with a range of 8 to 23 (a score of ≤17 indicates low burnout); the average depersonalization component score was 9.25, with a range of 7 to 15 (a score of 6 to 11 indicates moderate depersonalization/burnout); and the average personal achievement component score was 40, with a range of 36 to 42 (a score of >40 indicated a low level of burnout).(8)
The month following the intervention, the research assistant was able to complete 339 patient advice requests (45% of the total); 257 staff messages (34% of the total); and 156 patient calls (21% of the total) (Figure 3). A total of 752 messages (33% of the total) could be responded to by the research assistant alone or rerouted and did not require any physician input. Study participants completed 753 patient advice requests (55% of the total); 756 staff messages (66% of the total); and 743 patient calls (79% of the total). The intervention decreased the number of messages by an average of 85 messages per participating neurologist (33% of the total messages). All messages that required clinical information—a total of 2252—were completed solely by the neurologists.
When the physicians were asked whether the intervention had an impact on the number of in-basket messages they received, all nine replied that they did not notice any change in the number of in-basket messages. Three months after the intervention, the average Maslach burnout component score was 14.75, with a range of 6 to 24 (low burnout; not statistically different using a paired t-test). The average depersonalization component score was 9.75, with a range of 9 to 17 (moderate burnout; (not statistically different using a paired t-test). The average personal achievement component score was 30, with a range of 30 to 39 (moderate burnout; statistically different using a paired t-test).
Discussion
In-basket messages are used by a subset of patients who may enjoy the advantage of “on-demand” care. However, physicians report higher rates of burnout and feeling overwhelmed by the volume of messages and the number of hours spent answering them, even on days with no direct patient care.(2,9) Changes in the practice of medicine have led to a new focus on transparency and improved availability of physicians to answer questions from patients that arise outside of the appointment. This shift in nonreimbursable care has increased demands on physicians, heightened by the decreased infrastructure in academic centers since the pandemic. Patients may send messages that are not appropriate for a neurologist; ask for decision-making where a chart review, clinical history or exam, and documentation into the record is needed; or require an urgent clinical response, new medication, or test orders. The time needed to address these messages detracts from time scheduled for academic pursuits. Furthermore, it adds to physician administrative burden and contributes to burnout.(10) Similar to published data, 50% of our neurologists had moderate burnout, which contributes to both decreased safety and decreased quality of care.(11)
The intervention did not address support for clinically oriented messages, which are the ones that require the most time and require a clinician to address.
Despite the reduction in the numbers of messages received by the neurologists during the intervention, they did not perceive improvement in clinical message burden. One possible reason for the lack of perceived benefit of NeuroMessage is that our algorithm and trained research assistant support did not provide enough reduction to the factors that contribute most to physician administrative burden. The intervention did not address support for clinically oriented messages, which are the ones that require the most time and require a clinician to address. A potential future intervention could exempt the physician from responding to (or even seeing) the message at all by having a nurse or medical assistant decide whether the patient requires an appointment and to provide clinical support for messages and calls that do not require the physician.
Interestingly, the Maslach Burnout Inventory survey pre- and post-intervention did not show major changes in measures of burnout and depersonalization, but there was a significant decline in the average participant’s sense of personal achievement. The many potential reasons for this drop may be unrelated to in-basket management; however, having less time to pursue teaching and research may have had an impact. The daily interruptions required to check in-basket messages for urgent ones, and feeling that clinical work must be done daily rather than being restricted to the days physicians are seeing patients could play a role. Further study is indicated to identify the main drivers of neurologist burnout. For example, we noted marked variability in how female versus male physicians responded to in-basket messages.(12) Studying such factors may predict who is at risk for burnout. Offering clinical support to triage messages, allowing only rare urgent messages through to the neurologist, also should be studied to see whether aligning duties that are uniquely suited for a neurologist improves personal achievement. Although this type of response may require more clinical support, an analysis of the cost of physician turnover and burnout, as well as allowing the physician to focus on visits that are billable, may decrease burnout while not affecting the bottom line for institutions.
Conclusion
Dealing with in-basket messages has become a substantial contributor to burnout for clinical academic neurologists. We found that use of a streamlined algorithm and research assistant training designed to decrease non-clinical messages did not affect the wellness of academic neurologists. Further study of the factors contributing to neurologist burnout within academic centers is needed, and one recommendation may be to provide more support for clinical messages.
References
Hefner JL, MacEwan SR, Biltz A, Sieck CJ. Patient portal messaging for care coordination: a qualitative study of perspectives of experienced users with chronic conditions. BMC Fam Pract. 2019;20(1):57.
Holmgren AJ, Downing NL, Bates DW, et al. Assessment of electronic health record use between US and non-US health systems. JAMA Intern Med. 2021;181:251-259.
Lacy BE, Chan JL. Physician burnout: the hidden health care crisis. Clin Gastroenterol Hepatol. 2018;16:311-317.
Akbar F, Mark G, Warton EM, et al. Physicians’ electronic inbox work patterns and factors associated with high inbox work duration. J Am Med Inform Assoc. 2021;28:923-930.
Busis NA, Shanafelt TD, Keran CM, et al. Burnout, career satisfaction, and well-being among US neurologists in 2016. Neurology. 2017;88:797-808.
West CP, Dyrbye LN, Shanafelt TD. Physician burnout: contributors, consequences and solutions. J Intern Med. 2018;283:516-529.
Maslach C, Jackson SE, Leiter MP. Maslach burnout inventory manual. 3. Palo Alto, CA: Consulting Psychologists Press; 1996.
Maslach-Burnout-Inventory.pdf https://www.healthlinkscertified.org/uploads/files/2022_11_16_19_56_57_Maslach-burnout-inventory-english.pdf
Holmgren AJ, Downing NL, Tang M, et al. Assessing the impact of the COVID-19 pandemic on clinician ambulatory electronic health record use. J Am Med Inform Assoc. 2022;29:453-460.
Deb A. Practical Considerations in Addressing Physician Burnout. Minneapolis: Continuum; 2017 Apr;23(2, Selected Topics in Outpatient Neurology):557-562.
Sigsbee B, Bernat JL. Physician burnout: a neurologic crisis. Neurology. 2014;83:2302–2306.
LaFaver K, Miyasaki JM, Keran CM, et al. Age and sex differences in burnout, career satisfaction, and well-being in US neurologists. Neurology. 2018 Nov 13;91(20):e1928-e1941.