Error bars indicate standard error of the mean.
eTable 1. Demographic Characteristics of Responding Physicians Compared to All US Physicians
eTable 2. Comparison of Employed Physicians in the Sample Aged 29-65 Years With a Probability Based Sample of the Employed US Population Aged 29-65 Years
eTable 3. Multivariable Linear Regression Model of the Relationship Between Physician Resilience Score and Emotional Exhaustion Score
eTable 4. Multivariable Logistic Regression Model of the Relationship Between Physician Resilience Score and High Emotional Exhaustion
eTable 5. Multivariable Linear Regression Model of the Relationship Between Physician Resilience Score and Depersonalization Score
eTable 6. Multivariable Logistic Regression Model of the Relationship Between Physician Resilience Score and High Depersonalization
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
West CP, Dyrbye LN, Sinsky C, et al. Resilience and Burnout Among Physicians and the General US Working Population. JAMA Netw Open. 2020;3(7):e209385. doi:10.1001/jamanetworkopen.2020.9385
在这项对 30,456 名医生中的 5,445 名受访者进行的横断面调查研究中，这些医生的恢复力得分明显高于一般美国就业人口。较高的恢复力得分与较低的倦怠率相关，但即使是最有恢复力的医生也有相当高的倦怠率。
The prevalence of physician burnout is well documented, and resilience training has been proposed as an option to support physician well-being. However, the resilience of physicians compared with that of the US working population is not established, and the association between resilience and physician burnout is not well understood.
To evaluate resilience among physicians and US workers, and to determine the association between resilience and burnout among US physicians.
Design, Setting, and Participants
A cross-sectional national survey study of 5445 US physicians and a probability-based sample of 5198 individuals in the US working population was conducted between October 12, 2017, and March 15, 2018.
Main Outcomes and Measures
Resilience was measured using the 2-item Connor-Davidson Resilience Scale (total scores range from 0-8; higher scores indicate greater resilience); burnout was measured using the full Maslach Burnout Inventory with overall burnout indicated by a score of at least 27 on the 0 to 54 emotional exhaustion subscale and/or at least 10 on the depersonalization subscale (higher scores indicate greater burnout).
Of 30 456 physicians who received an invitation to participate, 5445 (17.9%) completed surveys (2995 men [62.1%]; median [IQR] age of 53 [42-62] years). In multivariable analysis, mean (SD) resilience scores were higher among physicians than the general employed population (6.49 [1.30] vs 6.25 [1.37]; adjusted mean difference, 0.25 points; 95% CI, 0.19-0.32; P < .001). Among physicians, resilience was associated with burnout. Physicians without overall burnout had higher mean (SD) resilience scores than physicians with burnout (6.82 [1.15] vs 6.13 [1.36]; adjusted mean difference, 0.68 points, 95% CI, 0.61-0.76; P < .001). Each 1-point increase in resilience score was associated with 36% lower odds of overall burnout (odds ratio, 0.64; 95% CI, 0.60-0.67; P < .001). However, 392 of 1350 physicians (29%) with the highest possible resilience score had burnout.
Conclusions and Relevance
The findings of this national survey study suggest that physicians exhibited higher levels of resilience than the general working population in the US. Resilience was inversely associated with burnout symptoms, but burnout rates were substantial even among the most resilient physicians. Additional solutions, including efforts to address system issues in the clinical care environment, are needed to reduce burnout and promote physician well-being.
The prevalence of physician distress has been well documented in recent years, with recent national data suggesting that 44% of US physicians experience symptoms of burnout, characterized by emotional exhaustion and/or depersonalization, at least weekly.1 Solutions to reduce distress and promote professional well-being have been broadly categorized into individual-focused and organization-oriented domains.2-4 Among individual-focused approaches, resilience training has been proposed as one means to support well-being.5,6
Resilience is the collection of personal qualities that enable a person to adapt well and even thrive in the face of adversity and stress.7,8 The physician training process is lengthy and rigorous. Given the intensity of this experience, resilience might be expected to be greater among practicing physicians than among workers in other careers. Among physicians, those with higher levels of resilience might be expected to navigate the demands of their professional life more effectively and experience lower levels of burnout. Preliminary evidence in support of the latter hypothesis has been reported in previous studies of 584 US9,10 and 247 UK11 physicians, although these studies included physicians in training and did not include concurrent population comparators. To our knowledge, no large-scale evaluation of resilience among physicians compared with the general working population or of the association between resilience and burnout among practicing physicians has been performed.
To evaluate resilience among physicians and how it compares with resilience among other US workers, we conducted a national survey in 2017. This study also measured burnout symptoms to allow analysis of the association between resilience and burnout among physicians.
We conducted a national survey of US physicians as well as US workers in other career fields in 2017. Complete details of the 2017 survey methodology have been previously reported.1 The 2017 survey used methods similar to the previous 2011 and 2014 studies.12,13 At all 3 time points, we assessed a range of personal and professional characteristics as well as several dimensions of well-being. The institutional review boards of Stanford University, Palo Alto, California, and the Mayo Clinic, Rochester, Minnesota, reviewed and approved this study. Informed consent of study participants was indicated by voluntary completion of the survey. This study followed the American Association for Public Opinion Research (AAPOR) reporting guideline.
A sample of physicians from all specialty disciplines was developed from the American Medical Association Physician Masterfile, a nearly complete record of all US physicians independent of membership with the American Medical Association. A greater proportion of participants were sampled from specialties other than primary care to provide representation across specialties. Email correspondence stating the purpose of the study (ie, to better understand the factors that contribute to satisfaction among US physicians), along with an invitation to participate and a link to the survey, were sent to 83 291 physicians on October 12, 2017, with 4 reminder requests sent during the ensuing 6 weeks. A total of 27 071 physicians opened at least 1 invitation email. After these 6 weeks, a random sample of 5000 physicians who did not respond to the electronic survey were mailed a paper version of the survey on December 6, 2017 (1426 physicians had opened an email invitation and 3574 physicians had not). Of these, 269 surveys were returned as undeliverable (80 physicians had opened an email invitation and 189 physicians had not).
To evaluate for response bias, we also conducted a secondary survey with intensive follow-up in a random sample of 500 physicians who did not respond to the electronic survey. These individuals were mailed a paper copy of the survey with a $20 incentive to participate. Individuals in the secondary survey who did not respond to the first mailing were sent a second mailing 3 weeks later (without additional compensation). Twenty-four mailed surveys were returned as undeliverable, yielding a final sample of 476 surveys. Those who did not respond to the second mailing within 3 weeks were mailed a brief postcard survey requesting basic demographic characteristics and measures of well-being. Completed surveys returned by March 15, 2018, were included in the analysis. The 30 456 physicians who opened at least 1 invitation email and/or received a paper mailing of the survey were considered to have received an invitation to participate in the study.14 Participation was voluntary and all responses were anonymous.
For comparison with physicians, we surveyed a probability-based sample of individuals in the general US population from October 13 through October 21, 2017. The population survey was conducted using the KnowledgePanel (Ipsos),15 a probability-based panel designed to be representative of the US population. Based on the intent to compare workers in other career fields to physicians, only employed individuals were surveyed.
Both the physician and population samples provided information on demographic characteristics (age, sex, and relationship status), hours worked per week, resilience, and symptoms of burnout. Physician professional characteristics were ascertained by asking physicians about their medical practice.
Resilience among both physicians and other US workers was assessed using the 2-item Connor-Davidson Resilience Scale (CD-RISC), a standardized and validated instrument measuring “bounce-back” and adaptability aspects of resilience.16,17 This scale has been studied and applied across diverse populations, including physicians and medical students, with consistently strong psychometric properties.16 Scoring of this scale is based on the sum of scores from 0 to 4 for each item (0 indicates the characteristic is not true at all; 4, it is true nearly all the time), for a total score range of 0 to 8 (0 indicates the lowest resilience level; 8, the highest resilience level).
Burnout among physicians was measured using the full emotional exhaustion and depersonalization scales of the Maslach Burnout Inventory (MBI), a validated 22-item questionnaire considered the criterion standard tool for measuring burnout.18-21 Consistent with convention,22-24 we considered physicians with a high score on the emotional exhaustion subscale (≥27 on a 0-54 scale, with 0 indicating no emotional exhaustion and 54 indicating the greatest possible emotional exhaustion) and/or depersonalization subscale (≥10 on a 0-30 scale, with 0 indicating no depersonalization and 30 indicating the greatest possible depersonalization) of the MBI as having at least 1 manifestation of professional burnout.18
To minimize survey burden for general population respondents, we measured burnout in analyses comparing physicians with the general working population using 2 single-item measures adapted from the full MBI. These 2 items have been shown to be associated with the emotional exhaustion and depersonalization domains of burnout measured by the full MBI in a sample of more than 10 000 individuals. The area under the receiver operating characteristic curve is 0.94 for emotional exhaustion and 0.93 for depersonalization for these single items relative to the full MBI.25,26
Standard descriptive summary statistics were used to characterize the physician and comparison samples. Associations between resilience and demographic and professional factors were evaluated using Kruskal-Wallis tests. Multivariable models comparing resilience scores of physicians with those of the general population used multiple linear regression adjusted for sex, age, relationship status, hours worked per week, and burnout status. For all comparisons with population comparators, physician data were restricted to responders who were between the ages of 29 and 65 years and not retired to match the population sample. Multivariable models examining the association between resilience and physician burnout used multiple logistic regression adjusted for sex, age, hours worked per week, practice setting, and specialty. All tests were 2-sided, with type I error rates of .05. All analyses were completed using R, version 3.4.2 (R Project for Statistical Computing).27
Of the 30 456 physicians who received an invitation to participate either electronically and/or by mail, 5445 (17.9%) completed a survey (2995 were men [62.1%]; median [IQR] age was 53 [42-62] years). Previous analysis comparing these responders with the participants in the secondary survey of nonresponders, which achieved a more than 50% response rate, supported participant representativeness of US physicians across domains of well-being.1 As previously reported, the demographic characteristics of participants relative to all 890 083 practicing US physicians were generally similar, although participants were slightly older (eTable 1 in the Supplement).1
The mean (SD) resilience score among the 4705 physicians who completed the CD-RISC was 6.51 (1.29) (Table 1). Resilience scores varied modestly across demographic and professional factors, with slightly higher resilience among male and older physicians. Across specialties, resilience scores were highest in emergency medicine, neurosurgery, and preventive and occupational medicine, and lowest in general pediatrics, neurology, and obstetrics and gynecology (Table 1).
Next, we compared resilience scores among physicians aged 29 to 65 years with those of the general US working population of the same age range (Table 2). Demographic differences between the physician and general population samples in 2017 have been published previously1 and are summarized in eTable 2 in the Supplement. Briefly, physicians were more likely to be male, younger, and married and reported working longer hours. Among the 3971 responding nonretired physicians aged 29 to 65 years, the mean (SD) resilience score was 6.49 (1.30) compared with 6.25 (1.37) for the 5198 nonretired individuals aged 29 to 65 years from the general US working population (mean difference, 0.24; 95% CI, 0.19-0.29; P < .001) (Table 2). After adjustment for sex, age, relationship status, hours worked per week, and burnout status, the higher resilience score among physicians persisted (mean difference, 0.25; 95% CI, 0.19-0.32; P < .001).
Among physicians, resilience was associated with burnout symptoms (Table 3 and Figure). Mean (SD) resilience was 6.82 (1.15) among physicians without burnout symptoms, and 6.13 (1.36) among those with burnout symptoms (mean difference, 0.68; 95% CI, 0.61-0.76; P < .001). On multivariable analysis adjusted for sex, age, hours worked per week, practice setting, and specialty, each 1-point increase in resilience score was associated with 36% lower odds of burnout (OR, 0.64; 95% CI, 0.60-0.67; P < .001) (Table 4). Forty-nine of 60 (82%) physicians with resilience scores of 3 or less had burnout symptoms, and 392 of 1350 (29%) with the highest possible resilience score of 8 had burnout symptoms (Figure).
When the individual domains of burnout were examined separately, physicians with higher resilience scores had lower emotional exhaustion scores. Each 1-point increase in resilience score was associated with a 3.18-point decrease in emotional exhaustion score (95% CI, 2.90-3.45; P < .001) (eTable 3 in the Supplement) and 36% lower odds of high emotional exhaustion (OR, 0.64; 95% CI, 0.61-0.68; P < .001) (eTable 4 in the Supplement). Physicians with higher resilience scores also had lower depersonalization scores. Each 1-point increase in resilience score was associated with a 1.43-point decrease in depersonalization score (95% CI, 1.29-1.57; P < .001) (eTable 5 in the Supplement) and 35% lower odds of high depersonalization (OR, 0.65; 95% CI, 0.61-0.68; P < .001) (eTable 6 in the Supplement).
In this national survey study in the US, levels of resilience were greater among physicians than among the general working population. In addition, physician resilience was inversely associated with burnout symptoms, and symptoms of burnout were common even among physicians with the highest possible resilience score.
These results suggest that, although higher levels of resilience might protect against burnout to a degree, physicians are not collectively deficient in resilience and even the most resilient physicians are at substantial risk of burnout. Therefore, although efforts to maintain or strengthen resilience are appropriate, equal or greater emphasis should be placed on alternative and complementary efforts, especially those addressing characteristics of the practice and external environments (eg, regulatory requirements) that contribute to burnout.4,28-30 For example, targets for improvement include inefficient workplace processes, excessive workloads, and negative leadership behaviors.30 This approach aligns with evidence to date supporting equal or greater effectiveness of organizational solutions to reduce burnout and promote well-being relative to individual-focused solutions such as those oriented around personal resilience.2,3
Although many of the specialties with the highest resilience scores in this study have shown lower burnout rates and many of the specialties with the lowest resilience scores in this study have shown higher burnout rates,1 there were notable exceptions. For example, the specialty with the highest adjusted mean resiliency score in the present study—emergency medicine—has had the highest burnout rate in previous research.1 The set of disciplines with the lowest burnout rate in previous research—pediatric subspecialties1—also had below-average resilience in the present study. The observed differences in resilience across specialties and the association of resilience with burnout within each specialty are intriguing and merit further study.
Our study had several limitations. First, the participation rate among physicians who opened the invitation email was only 18%, raising concern for nonresponse bias. Although in line with response rates of other national survey studies of physicians,31-33 this rate was lower than response rates of some physician surveys.34 To address the concern for nonresponse bias, as previously reported1 we used a robust double survey approach using incentives to compare participants with nonresponders.35 The results revealed no statistically significant differences with respect to age, years in practice, burnout, or satisfaction with work-life integration, suggesting that the responders were representative of US physicians for at least these variables. Second, more detailed resilience instruments exist, including 10- and 25-item versions of the CD-RISC.16 These versions were not applied in this study to limit participant survey burden but could provide more nuanced insight into physician resilience. Third, the cross-sectional survey method does not allow assessment of the direction of effect for the associations described in this study.
In summary, in this national cross-sectional survey study in the US, physicians exhibited greater resilience than the general working population. Resilience was inversely associated with burnout symptoms. Although maintaining and strengthening resilience is important, physicians are not generally resilience-deficient and burnout rates are substantial even among the most resilient physicians. Additional solutions, including efforts to address system issues in the clinical care environment, are needed to reduce burnout and promote physician well-being.
Accepted for Publication: April 23, 2020.
Published: July 2, 2020. doi:10.1001/jamanetworkopen.2020.9385
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 West CP et al. JAMA Network Open.
Corresponding Author: Colin P. West, MD, PhD, Division of General Internal Medicine, Department of Medicine, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (email@example.com).
Author Contributions: Drs West and Nedelec had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: West, Dyrbye, Sinsky, Shanafelt.
Acquisition, analysis, or interpretation of data: West, Sinsky, Trockel, Tutty, Nedelec, Carlasare, Shanafelt.
Drafting of the manuscript: West, Tutty, Shanafelt.
Critical revision of the manuscript for important intellectual content: West, Dyrbye, Sinsky, Trockel, Nedelec, Carlasare, Shanafelt.
Statistical analysis: Nedelec.
Obtained funding: Shanafelt.
Administrative, technical, or material support: Dyrbye, Tutty, Carlasare, Shanafelt.
Supervision: West, Shanafelt.
Conflict of Interest Disclosures: Dr Dyrbye reported receiving royalties from the Well-Being Index from Med Ed Solutions outside the submitted work. Dr Shanafelt reported having a patent to the Physician Well-being Index with royalties paid. No other disclosures were reported.
Funding/Support: Funding for this study was provided by Stanford WellMD Center, Palo Alto, California; the American Medical Association, Chicago, Illinois; and the Mayo Clinic Department of Medicine Program on Physician Well-being, Rochester, Minnesota.
Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.