Physicians’ Well-Being Linked To In-Basket Messages Generated By Algorithms In Electronic Health Records

The electronic health record (EHR) potentially creates a 24/7 work environment for physicians. Its impact on physicians’ wellness has become a challenge for most health care delivery organizations. Understanding the relationships between physicians’ well-being and “desktop medicine”¹ work in the EHR and work environment is critical if burnout is to be addressed more effectively.

While various causes of burnout have been examined in prior research, including workload, efficiency, meaning in work, culture and values, control and flexibility, social support and community at work, and work-life integration,² only one previous effort has examined the role played by EHR workload.³ That study, which relied on physicians’ self-reported time using EHRs and computerized order entry, found that burnout was associated only with the latter.

A growing research literature suggests that time spent by physicians on the EHR has been linked to their reduced satisfaction with work.^4–6 An analysis of EHR log data on the distribution of work time of 471 primary care physicians (in family medicine, internal medicine, or pediatrics) suggests that over 50 percent of their time is spent on desktop medicine tasks.¹ A study of 142 family physicians reports that managing the EHR’s in-basket takes about 23 percent of their workday.⁷ Despite concerns about the amount of physicians’ time spent on EHR in-baskets,^7,8 the literature is relatively silent regarding the sources of in-basket messages and their relative volumes. Given the correlation between physician workload and desktop medicine¹ and the rising number of physicians who report burnout,⁹ it is important to carefully examine the relationship between pivotal aspects of desktop medicine and physicians’ well-being. As federal meaningful-use regulations continue to shape how health care organizations deliver care, the impact of desktop medicine workload on physicians’ well-being can have critical policy implications.¹⁰

Study Data And Methods

This observational study used survey, administrative, and EHR data from the Palo Alto Medical Foundation, a multispecialty health care delivery organization in California that was an early adopter of EHRs (it installed Epic in 1999). All physicians, whether in primary or specialty care, were invited to respond to a confidential one-page survey in 2016 (see the online appendix for the survey instrument).¹¹ A $10 gift card was provided with the survey, unconditional of response. Of the 1,292 physicians invited to take the survey, 934 (72 percent) responded.

Variable Definitions

Three dependent variables came from the one-item burnout measure (a validated five-point scale, with a score of 3 or higher indicating burnout),¹² intention to reduce clinical work time, and a three-item general life satisfaction scale used in previous research on well-being.¹⁰ For explanatory variables, we examined the volume of in-basket messages during a six-week period in late 2016. The in-basket messages included those from patients and other clinicians, as well as messages generated by the EHR system. (The appendix lists the types of messages in these categories of sources.)¹¹ We also measured the time each physician spent on entering progress notes in the EHR.¹ Additional variables included respondents’ level of agreement with the statement “physicians are highly valued”;¹³ their description of the atmosphere of their primary work area¹³ on a scale of 1 (calm) to 5 (hectic or chaotic), with a midpoint of 3 (busy but reasonable); and their control over their work schedule on a scale of 1 (poor) to 5 (optimal). Respondents further reported on sleep, exercise, and mindfulness practices, because previous research has documented their effects.^14–16 To account for the potential influence of sociodemographic factors, we obtained administrative data on specialty, sex, and leadership or administrative roles. Specialties were grouped into six categories: family medicine, internal medicine, pediatrics, surgical, nonsurgical procedural, and nonprocedural specialties focused on evaluation and management–oriented services (see the appendix for details).¹¹

Analytical Approaches

We used separate logistic regression analyses to examine the relationship between burnout or intention to reduce clinical work time and in-basket volume. We used linear regression analysis to examine the relationship between general life satisfaction and the explanatory variables. We included the variable that is potentially more modifiable by health system policy on physician work—that is, above-average numbers of system-generated in-basket messages—in the logistic and linear regression models. The time each physician spent on progress notes in the EHR was not significantly (statistically) associated with the dependent variables. In the interest of parsimony, we do not report the details of those results.

Limitations

This study had several limitations. First, the use of the nonproprietary one-item questionnaire on burnout could have underestimated the level of burnout among physicians in this organization. A recent study¹⁷ measured burnout using multiple instruments, including the one-item questionnaire and the commonly used, proprietary, and significantly longer Maslach Burnout Inventory. That study found a lower burnout rate using the one-item measure compared to the inventory.

Second, while we surveyed physicians’ general satisfaction with life, burnout symptoms, and intention to reduce clinical work time, we did not measure other related aspects of well-being, such as feeling joy in practicing medicine.

Third, the measure of intention to reduce clinical work time was developed in response to the interests of Palo Alto Medical Foundation leaders. While the wording of the question and values for the answers had been vetted by multiple physicians and leaders, resulting in reasonable face validity, no additional validity tests had been done.

Fourth, we did not examine the percentage of in-basket messages that had been read or responded to. The results of our analyses should be interpreted in the context of the number of messages that had reached physicians’ in-baskets, rather than the number they had read or responded to. It has been reported that some physicians delete a large number of messages after glancing at them, without reading the body of the messages.⁸ The affective impact of in-basket messages may relate more to the number received than to the number of messages read.¹⁸

Lastly, the study’s results might not be generalizable because they came from one large delivery organization that adopted an EHR in 1999.

Study Results

Thirty-six percent of responding physicians reported burnout symptoms, and twenty-nine percent expressed an intention to reduce their clinical work time in the coming year (exhibit 1). The average score on life satisfaction was 78.22 out of 100 (see the exhibit 1 notes).

It is noteworthy that only 12 percent of respondents deemed the statement “physicians are highly valued” to be completely true regarding conditions in their primary practice setting (exhibit 1). Eighty-three percent reported having good or optimal control over their work schedule, and 60 percent reported having a calm to busy but reasonable primary work area.

The average number of weekly in-basket messages was 243, of which 114 (47 percent) were generated by the EHR system (exhibit 2). These included pending orders automatically sent to physicians according to algorithm-driven health maintenance reminders,¹⁹ requests for prior authorization, patient reminders, and many more. Only 30 messages per week were directly from patients. Fifty-three were from other physicians or care team members, and 31 were from the physicians themselves (for example, reports of laboratory tests they had ordered). Forty-two percent of physicians received above the average number of messages (114 per week) (data not shown).

SOURCE Authors’ analysis of data for 2016 from the Palo Alto Medical Foundation’s administrative and electronic health records. NOTES “System” refers to the electronic health record system. “Team” refers to other physicians or care team members. E&M is evaluation and management.

System-generated messages stood out as the largest source of messages for all specialties, especially for internal medicine (209) and family medicine (204), followed by pediatricians (102) (exhibit 2). The second-largest source across all specialties was other physicians or care team members. For primary care physicians (those in internal medicine, family medicine, or pediatrics), the third-largest source of messages was patients; for specialists, the third-largest source of messages was themselves.

Exhibit 3 shows the relationship between burnout symptoms and system-generated in-basket message volume, having a leadership role, perceptions of having control over one’s work schedule, having a calm to busy but reasonable primary work area, feeling valued, and being female. Notably, 45 percent of physicians with burnout symptoms received greater-than-average numbers of weekly system-generated in-basket messages, whereas 29 percent of physicians with burnout symptoms received only average or less-than-average numbers of the messages.

SOURCE Authors’ analysis of data for 2016 from a survey of physicians at the Palo Alto Medical Foundation and the foundation’s electronic health records. NOTE “System” refers to the electronic health record system.

The regression results suggest that factors associated with higher odds of burnout include an above-average number of system-generated in-basket messages and being female (exhibit 4). Factors associated with lower odds of burnout include feeling that physicians are highly valued; having good control over one’s work schedule; and having a calm or busy but reasonable work environment. Those factors were also associated with lower odds of intending to reduce clinical work hours and having higher life satisfaction. In contrast, high system-generated in-basket message volume; family medicine, internal medicine, and nonsurgical procedural specialties (compared to pediatricians) were associated with higher odds of intending to reduce clinical work hours. Female physicians had lower life satisfaction than males did.

To illustrate the association between burnout, intention to reduce clinical work hours, and the key explanatory variable—above-average volumes of system-generated in-basket messages—we calculated the variable’s marginal effects on burnout and intention to reduce clinical work hours. The results suggested that, at the margin, receiving an above-average number of system-generated messages was associated with a 40 percent increase in the probability of burnout and a 38 percent increase in the probability of intending to reduce clinical work hours (data not shown).

Discussion

Almost half of all weekly in-basket messages came from EHR algorithms, and they were significantly associated with physicians having burnout symptoms and intending to reduce their clinical work hours. These are new findings that deserve careful consideration. The finding that family physicians and internists receive disproportionately high numbers of system-generated messages (greater than 2.5 times the volume of surgeons, 4.0 times that of nonprocedural evaluation and management–oriented specialists, and 5.0 times that of nonsurgical proceduralists) also calls for our attention. Keeping up with system-generated messages can be overwhelming, particularly for internists and family physicians. Some of those messages are generated by population health management algorithms that remind physicians to perform work that might have otherwise been overlooked. These were in addition to messages coming directly from patients and from other physicians or care team members. Therefore, both perceived and realized loss of autonomy over their work schedules could leave physicians feeling defeated,²⁰ even though some of these system-generated messages have been shown to improve certain processes of care for patients with chronic illnesses.¹⁹

Health care organizations need to reconsider some of their approaches to improving the quality of care and population health. Physicians might not be the most appropriate recipients of some system-generated messages. Payers and government regulators may need to be part of the solution in enabling physicians to practice at the top of their license. EHR design engineers also need to reconsider whether system-generated automatic messages are the best way to ensure quality of care. It may be time to examine whether every reminder to order routine chronic disease management lab tests (for example, periodic glycosylated hemoglobin A1c tests) must be signed and placed by a physician.

Health care organizations may benefit from engaging with their physicians in creating optimal policies on email work, in addition to helping them with such work. Limiting desktop medicine work during evenings, weekends, and holidays, unless the physician is on call, could reduce burnout. Health care organizations could allow messages to reach physicians’ in-baskets only during work hours, conveying the message that physicians are so highly valued that the organization wants to protect their private time. To reduce the number of messages reaching physicians’ in-baskets, some messages (such as prescription refill requests) may be delegated to nonphysician clinicians (for example, registered nurses or clinical pharmacists) on the care team.

Another finding also calls for attention. Compared to pediatricians, physicians in four specialties (surgeons were the only exception) had higher odds of intending to reduce clinical work hours in the coming year. Inaction or ineffective actions in addressing the mounting in-basket workload and issues with physicians’ work environment could exacerbate their burnout and result in more reduction in work hours,²¹ which could negatively affect patients’ access to care and organizations’ workforce stability.

Female physicians had 1.4 times higher odds of burnout and 3 points lower satisfaction with life, compared to male physicians. While these findings have been reported previously,²² actions are still needed to support female physicians. It is necessary to recognize the different manifestations of burnout symptoms across physician genders. Female physicians have been documented to first suffer from emotional exhaustion, whereas male physicians first experience depersonalization.²² Female physicians are also more likely to experience burnout resulting from work-home conflict, whereas workload was a key predictor of burnout for male physicians.²³ More flexibility in work schedules may enable female physicians to better harmonize work-home responsibilities and achieve joy in work and home life.²⁴

The EHR has become the symbol of physician burnout.²⁵ The reality of physician wellness is complex, however. Physicians feel that they are being consumed not only by cumbersome EHR work but also by fee-for-service payment—among other detractors to their wellness.²⁴ Burnout risk factors that are modifiable by organizational changes include the workplace environment, physicians’ perception of being valued, and control over their work schedules. Organizational leaders can address these factors through changes in their communication with physicians, staffing, and scheduling. Policy makers and organizations also need to effectively address desktop medicine work, including redesigning in-basket work responsibilities. Using computer algorithms to send messages to physicians may need to be balanced with considerations of what brings joy and which tasks are truly physicians’ work. As noted in the “Study Data And Methods” section above, we did not observe that the amount of time physicians spent in EHR progress notes was significantly associated with burnout (results not shown). When we discussed the absence of a significant relationship between burnout and progress notes work with physicians, we learned that they considered work in progress notes to be work that they perform with more autonomy. They derived professional satisfaction from writing good notes.

Only a small proportion of in-basket messages came from patients and other clinicians, whereas almost half were generated by system algorithms. Receiving an above-average number of system-generated messages was associated with 40 percent higher probability of physicians having burnout symptoms and 38 percent higher probability of intending to reduce their clinical work hours. Collective actions are needed at the national and local organizational levels to bring about systematic changes to this previously little-known yet important source of friction.

ACKNOWLEDGMENTS

A previous version of this article was presented at the 2017 American Conference on Physician Health in San Francisco, California, October 13, 2017. The authors thank the Palo Alto Medical Foundation Philanthropy Department for funding the study. They also thank Kevin Chen, Jin Chang, Susan Connolly, Cliff Olson, Angela Scioscia, and the physicians who participated in the survey for their contributions.

NOTES

• 1 Tai-Seale M, Olson CW, Li J, Chan AS, Morikawa C, Durbin Met al. Electronic health record logs indicate that physicians split time evenly between seeing patients and desktop medicine. Health Aff (Millwood). 2017;36(4):655–62. Go to the article, Google Scholar
• 2 Shanafelt TD, Noseworthy JH. Executive leadership and physician well-being: nine organizational strategies to promote engagement and reduce burnout. Mayo Clin Proc. 2017;92(1):129–46. Crossref, Medline, Google Scholar
• 3 Shanafelt TD, Dyrbye LN, Sinsky C, Hasan O, Satele D, Sloan Jet al. Relationship between clerical burden and characteristics of the electronic environment with physician burnout and professional satisfaction. Mayo Clin Proc. 2016;91(7):836–48. Crossref, Medline, Google Scholar
• 4 Friedberg MW, Chen PG, Van Busum KR, Aunon F, Pham C, Caloyeras Jet al. Factors affecting physician professional satisfaction and their implications for patient care, health systems, and health policy. Rand Health Q. 2014;3(4):1. Medline, Google Scholar
• 5 Kim LY, Rose DE, Soban LM, Stockdale SE, Meredith LS, Edwards STet al. Primary care tasks associated with provider burnout: findings from a Veterans Health Administration survey. J Gen Intern Med. 2018;33(1):50–6. Crossref, Medline, Google Scholar
• 6 Babbott S, Manwell LB, Brown R, Montague E, Williams E, Schwartz Met al. Electronic medical records and physician stress in primary care: results from the MEMO Study. J Am Med Inform Assoc. 2014;21(e1):e100–6. Crossref, Medline, Google Scholar
• 7 Arndt BG, Beasley JW, Watkinson MD, Temte JL, Tuan WJ, Sinsky CAet al. Tethered to the EHR: primary care physician workload assessment using EHR event log data and time-motion observations. Ann Fam Med. 2017;15(5):419–26. Crossref, Medline, Google Scholar
• 8 Gawande A. Why doctors hate their computers. New Yorker [serial on the Internet]. 2018 Nov 5 [cited 2019 May 13]. Available from: https://www.newyorker.com/magazine/2018/11/12/why-doctors-hate-their-computers Google Scholar
• 9 Shanafelt TD, Hasan O, Dyrbye LN, Sinsky C, Satele D, Sloan Jet al. Changes in burnout and satisfaction with work-life balance in physicians and the general US working population between 2011 and 2014. Mayo Clin Proc. 2015;90(12):1600–13. Crossref, Medline, Google Scholar
• 10 Mochon D, Norton MI, Ariely D. Getting off the hedonic treadmill, one step at a time: the impact of regular religious practice and exercise on well-being. J Econ Psychol. 2008;29(5):632–42. Crossref, Google Scholar
• 11 To access the appendix, click on the Details tab of the article online.
• 12 Rohland BM, Kruse GR, Rohrer JE. Validation of a single-item measure of burnout against the Maslach Burnout Inventory among physicians. Stress Health. 2004;20(2):75–9. Crossref, Google Scholar
• 13 Linzer M, Manwell LB, Williams ES, Bobula JA, Brown RL, Varkey ABet al. Working conditions in primary care: physician reactions and care quality. Ann Intern Med. 2009;151(1):28–36, W6–9. Crossref, Medline, Google Scholar
• 14 Rosen IM, Gimotty PA, Shea JA, Bellini LM. Evolution of sleep quantity, sleep deprivation, mood disturbances, empathy, and burnout among interns. Acad Med. 2006;81(1):82–5. Crossref, Medline, Google Scholar
• 15 Krasner MS, Epstein RM, Beckman H, Suchman AL, Chapman B, Mooney CJet al. Association of an educational program in mindful communication with burnout, empathy, and attitudes among primary care physicians. JAMA. 2009;302(12):1284–93. Crossref, Medline, Google Scholar
• 16 Weight CJ, Sellon JL, Lessard-Anderson CR, Shanafelt TD, Olsen KD, Laskowski ER. Physical activity, quality of life, and burnout among physician trainees: the effect of a team-based, incentivized exercise program. Mayo Clin Proc. 2013;88(12):1435–42. Crossref, Medline, Google Scholar
• 17 Trockel M, Bohman B, Lesure E, Hamidi MS, Welle D, Roberts Let al. A brief instrument to assess both burnout and professional fulfillment in physicians: reliability and validity, including correlation with self-reported medical errors, in a sample of resident and practicing physicians. Acad Psychiatry. 2018;42(1):11–24. Crossref, Medline, Google Scholar
• 18 Gregory ME, Russo E, Singh H. Electronic health record alert–related workload as a predictor of burnout in primary care providers. Appl Clin Inform. 2017;8(3):686–97. Crossref, Medline, Google Scholar
• 19 Panattoni L, Chan A, Yang Y, Olson C, Tai-Seale M. Nudging physicians and patients with autopend clinical decision support to improve diabetes management. Am J Manag Care. 2018;24(10):479–83. Medline, Google Scholar
• 20 Ariely D, Lanier WL. Disturbing trends in physician burnout and satisfaction with work-life balance: dealing with malady among the nation’s healers. Mayo Clin Proc. 2015;90(12):1593–6. Crossref, Medline, Google Scholar
• 21 Shanafelt TD, Mungo M, Schmitgen J, Storz KA, Reeves D, Hayes SNet al. Longitudinal study evaluating the association between physician burnout and changes in professional work effort. Mayo Clin Proc. 2016;91(4):422–31. Crossref, Medline, Google Scholar
• 22 Houkes I, Winants Y, Twellaar M, Verdonk P. Development of burnout over time and the causal order of the three dimensions of burnout among male and female GPs. A three-wave panel study. BMC Public Health. 2011;11:240. Crossref, Medline, Google Scholar
• 23 Innstrand ST, Langballe EM, Falkum E, Aasland OG. Exploring within- and between-gender differences in burnout: 8 different occupational groups. Int Arch Occup Environ Health. 2011;84(7):813–24. Crossref, Medline, Google Scholar
• 24 Humikowski CA. Beyond burnout. JAMA. 2018;320(4):343–4. Crossref, Medline, Google Scholar
• 25 Verghese A. How tech can turn doctors into clerical workers: the threat that electronic health records and machine learning pose to physicians’ clinical judgment—and their well-being. New York Times Magazine [serial on the Internet]. 2018 May 16 [cited 2019 May 13]. Available from: https://www.nytimes.com/interactive/2018/05/16/magazine/health-issue-what-we-lose-with-data-driven-medicine.html Google Scholar