The term “burnout” has come into common parlance in this day of economic unrest and job uncertainty. Since its coinage in the 1970s by Herbert Freudenberger,1,2 burnout has come to represent the negative result of work-life imbalance, high stress, job disengagement, and job dissatisfaction. Although Freudenberger was originally describing the plight of child mental health workers in free clinics in New York City, the phenomenon of burnout has been recognized and extensively studied in airline workers3–6 and in health care workers.
Burnout is not an isolated entity; it does not exist in a vacuum and probably modulates (and is modulated by) other factors in each burned-out person. The potential for burnout exists in everyone, but the same interactions between personal characteristics and situational characteristics that might lead to burnout in some people might leave others unaffected. Dr Christina Maslach,7,8 a social scientist and leading researcher on burnout, describes burnout as “an individual experience that is specific to the work context.” Careful consideration of Dr Maslach’s description of burnout leads to more questions. Is there an explanation for the individual differences in burnout? Are there differences in the coping abilities of individuals? Are there differences in social situations that alter one’s abilities to adapt to work stresses? Are there mental or physical processes that influence coping abilities? Do burned-out people tend to use or abuse controlled substances more?
Physicians, nurses, and other health care workers are not immune to the woes of burnout, health issues, social ills, or substance abuse. Numerous articles describe burnout in medical specialists,9–12 and there are reported associations with physical and mental health problems,13 social issues,14 and substance use.15,16 However, there are no firm data confirming such relationships.
For about 30 years, the Maslach Burnout Inventory-Human Services Survey (MBI-HSS) has been the most commonly used tool in quantifying burnout risk in health-related fields.7 Subjects answer 22 questions and from the results are calculated 3 aspects of burnout: emotional exhaustion (EE), depersonalization (DP), and personal accomplishment (PA). EE occurs when an individual becomes so stressed and psychologically depleted that he or she does not have the personal capacity to respond to a client’s needs. DP refers to a situation where the provider has a cynical dehumanizing approach to patients. PA refers to a personal sense of the value of one’s work.7,8
In a previous study,17 we used MBI-HSS to study burnout in a group of volunteers, including resident physicians, attending physicians, nurses, nurse anesthetists, and nonlicensed personnel at a single academic medical center. In addition to studying burnout, we were interested in how work and home environments affected its development. Therefore, we developed the Social Support and Personal Coping (SSPC) survey. The SSPC survey probes such areas as health, personal and professional support, work satisfaction, and outside activities. Both instruments were included in a single anonymous survey. Results showed that physicians (particularly residents) are more prone to burnout than are nurses and that health concerns, personal support, and workload have a more deleterious effect on residents than on attending physicians or nurse anesthetists.
In 2011, we participated in a webinar on the subject of burnout cosponsored by the American Society of Anesthesiologists (ASA) and the journal Anesthesiology. As part of the webinar experience, we were asked to administer our burnout survey to attendees. We planned to assess burnout risk, social support, and coping skills and to test the hypothesis that burnout is associated with physical health issues, mental health issues, and substance abuse. We enhanced the previous survey questions with the addition of the 12-item Short Form Health Survey (SF-12) and the National Institute on Drug Abuse Modified Alcohol, Smoking, and Substance Involvement Screening Test (NM ASSIST). This was done to test the relationships of burnout risk with mental health, physical health, and substance use. The SF-12 has a long history18 and allows physical and mental health to be estimated from a series of 12 simple questions. The NM ASSIST is a tool used by physicians primarily to help identify which of their patients might be at risk for substance abuse. We present here the results of our ASA webinar project.
Using an electronic survey of at least 60 questions, we addressed the magnitude of burnout risk, physical health problems, mental health problems, social support, and substance use among a sample of webinar participants. The number of survey questions varied because we used branching logic to eliminate questions that were not applicable to an individual participant or to add questions that were applicable to an individual participant. The survey included the 22-question MBI-HSS,7 the SF-12,18 26 questions that focused on the availability and impact of social support systems and coping strategies and approximately 8 (depending on branching logic) focusing on substance use. The SF-12 and the substance use questions were the only additions to our previous survey instrument.17
The Institutional Review Board Behavioral Sciences Committee at Vanderbilt University approved the study. To minimize potential bias, potential participants were not informed of the specific purpose of the study. The act of completing and submitting the web-based questionnaire implied consent. Given the intimate nature of some of the questions and the potential revelation of undesirable behaviors, it was important that individual participants not be identifiable. To assure complete anonymity, we did not ask any identifying questions. This article adheres to the applicable EQUATOR (Enhancing the Quality and Transparency of Health Research) guidelines.
Although ASA membership was not mandatory, potential webinar participants had to be registered on the ASA website to receive information regarding the webinar and survey. The ASA did all recruitment for the webinar, and beyond the initial invitation there were no additional recruitment efforts (eg, sequential invitations17,19). Once enrolled in the webinar, participants had the opportunity to complete the survey as part of their webinar experience and received a link to the survey hosted by Vanderbilt’s REDCap (Research Electronic Data Capture)20 server application. Survey questions are listed in Supplemental Digital Content 1, Appendix 1, http://links.lww.com/AA/B871.
Maslach Burnout Inventory-Human Services Survey (MBI-HSS).
The MBI-HSS8 has become the gold standard to assess burnout in health-related fields. As noted previously, it evaluates the 3 aspects of burnout—EE, DP, and PA. MBI-HSS consists of 22 questions, of which 9 evaluate EE, 5 evaluate DP, and 8 evaluate PA. Subjects give their answers with a 7-point Likert scale (encoded using the integers 0–6). The composite scores for the 3 aspects of burnout were computed by averaging the scores for the corresponding questions. We report PA as “LPA” (lack of personal accomplishment—the inverse of PA) so all 3 aspects of burnout get scored in the same direction, that is, greater values indicate greater risk of burnout.
Physical Health/Mental Health (SF-12).
The SF-12 is a validated subset of the Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36)18,21 used to assess physical and mental status when a lower response burden is desired. The SF-12, which can be completed in 2 minutes or less, has been used extensively to evaluate health-related quality of life issues,22 work-related stress,23 and response to disease24 in a variety of professions (including medical) and in patients. From the 12 questions, physical (PCS) and mental (MCS) composite scores are calculated, with scores ranging from 0 (worst health state) to 100 (best health state). Composite scores are compared to national norms such that positive differences indicate a better health status and negative differences a worse health status. Controls are not matched controls, but since values can change over a lifetime, age-specific controls are available.
Social Support and Personal Coping (SSPC-14).
The Social Support and Personal Coping (SSPC-25) survey, developed and used in our prior work,17 is intended to elucidate an individual’s coping strategies and social support system. With the addition of the SF-12 to the present survey, we eliminated the 11 health and wellness questions to create the SSPC-14 survey. These remaining 14 questions fall into 4 natural groupings—work satisfaction, workload and control, professional support, and personal support. Each question is scored with a 9-point scoring system where, for most questions, a higher score represents better coping/support.
Tobacco, Alcohol, and Cannabis Use.
Starting with the established NM ASSIST tool,25 we refined a list of 22 short questions to focus specifically on the use of tobacco, alcohol, and cannabis and their impact on respondents’ lives and work. There were no questions about legitimate prescription medicines or substances that might be available in the workplace. The questions included 3 response timescales: use in the last year, use in the last 5 years, and lifetime use. The option of “prefer not to answer” was also included. This tool can be completed quickly for initial assessment of substance use.
During the process of data analysis, we were concerned that survey participants would not answer substance abuse questions with the same honesty as the other questions. Even among our own research group, we were concerned that such sensitive questions would not garner the same honesty as less sensitive ones. In light of this, we sent a short institutional review board–approved questionnaire (see Supplemental Digital Content 2, Appendix 2, http://links.lww.com/AA/B872) to the perioperative service at our hospital. This included all physicians and nurses that work in the surgical flow loop (preoperative holding, operating room, and postanesthesia care unit). While these are not the same people who participated in the webinar, they do represent a large community of nurses and doctors who work in a very busy surgical unit in a large hospital where stress and burnout are not unknown.
Respondent characteristics were summarized using sample statistics, including the mean, standard deviation, count, proportion, and 95% confidence intervals, as appropriate. When comparing estimates to published control values, statistical significance was indicated when the 95% confidence interval for the estimated quantity failed to include the published quantity. Adjusted linear regression methods were used to test the associations between each aspect of burnout and reported substance use, components of the SF-12 and the SSPC-14, and respondent demographics. Graphical residual diagnostics were used to assess the validity of regression assumptions. Ninety-five percent confidence intervals and P values were used to summarize the results of regression. P values <.05 were considered evidence of statistical significance. Due to the observational nature of this study, no a priori sample size target or power analysis was implemented. However, a parametric bootstrap method was used to retrospectively assess the statistical power to detect the effect of MCS on the total burnout score, and the effect of tobacco use on total burnout score, adjusting for all other factors and assuming that the corresponding effects were identical to their estimates in this sample. Using this method, we found that the current study design had approximately 80% power to detect an effect of SF-12 MCS that was 65% smaller than the estimated effect. However, there was very little power to detect the estimated effect of tobacco use on total burnout. Indeed, to achieve 80% power to detect the estimated effect of tobacco use, a study approximately 5 times larger would be required. Thus, there was substantial power to detect some associations, while there was poor power to detect others. In addition, and in lieu of reporting similar power estimates for all other factors investigated, we refer readers to examine the widths of the reported confidence intervals (see Table 1) as a gauge of the statistical precision.
Respondent demographic data are presented in Table 2 and are similar to ASA membership data for attending anesthesiologists. Due to discrepancies in methods used to summarize demographic data, no formal statistical comparison was performed. While the current cohort was larger in the 51- to 60-year age group, the percentage over 51 years was similar to ASA attending members and similar to younger age attending anesthesiologists. Gender distribution was also similar. Twenty residents participated (9.6%), but the majority of respondents were attending anesthesiologists. Ethnic data and practice location data were not available from the ASA.
In total, 221 respondents began the survey, and 170 (76.9%) completed all questions. There were 266 registrants total (31 registrants for the live event and 235 for the archive event), yielding an 83% response rate. Among respondents who provided job titles, 206 (98.6%) were physicians and 2 (0.96%) were registered nurses. One participant (0.48%) was listed as “other.” Among physicians, 186 (89.0%) had completed their training, 181 (83.1%) were over 40 years of age, 160 (74.1%) were male, and 71.0% were in nonteaching practices.
Figure 1 illustrates the distribution of responses for the 3 burnout constructs of the MBI-HSS. Normative controls are matched only by profession and consist of 1104 physicians and nurses.7 Respondents overall scored higher (ie, worse) on the EE and DP than published controls. Scores for the LPA portion were not significantly different from published controls. One concludes that despite higher than normal EE and DP, respondents overall retained a sense of PA about their work. Nevertheless, a significant proportion of respondents was at high risk in each category (EE [n = 105, 59.0%], DP [n = 54, 30.0%], and LPA [n = 46, 26.0%]). Only 25 (14.8%) were at high risk in all 3 categories (Table 3), a figure well below published values in other medical specialties.9,17
The average mental composite score of the SF-12 (41.3 ± 11.5) was one standard deviation lower than that of the normative population (52.2; Table 4), suggesting that respondents might be at greater risk for depression. In the SSPC-14 (Table 5), respondents scored higher in the areas of work satisfaction and professional support than they did in the areas of personal support and work control.
Among the 221 survey starts, 171 respondents (77.4%) completed one or more of the substance use questions (Table 6). Only 3.5% chose the option “prefer not to answer,” and about 18% reported no nonmedical drug use during their lifetime. When asked about lifetime use, nearly one-quarter admitted to using tobacco and more than three-quarters admitted to using alcohol. We also asked about the frequency, within the last year, of substance use for stress management. Possible answers were “never,” “occasionally,” “frequently (once or twice a week),” and “daily.” To this question, 6% of respondents admitted daily substance use, 10% described frequent use, and almost 25% reported occasional use.
The adjusted associations between burnout constructs (MBI-HSS) and components of the SF-12, SSPC-14, substance use, and participant demographics are summarized in Table 1. Males had higher levels of DP, LPA, and total burnout and, relative to attending physicians, residents had higher DP. The MCS of the SF-12 was significantly associated with all 3 aspects of burnout, as well as with the total burnout score. The SF-12’s PCS was associated with EE, LPA, and total burnout score but not with DP (P = .08). In each case, unfavorable SF-12 scores were associated with unfavorable burnout scores.
Stronger personal and professional support scores from the SSPC-14 were associated with decreased levels of EE (Table 1), but work satisfaction was not associated with decreased levels of DP. Personal support, professional support, and work satisfaction were each associated with perceived LPA, but not in a consistent direction. The association of overall burnout score with work satisfaction (P = .06) and professional support (P = .08) failed to reach statistical significance. The association of alcohol use with decreased LPA was likewise near the threshold for statistical significance. Thus, there was no significant evidence of associations between burnout and alcohol, tobacco, or cannabis use.
We compared the results of the SSPC-14 questions in this study to the same 14 questions from our previous study (Figure 2). The differences in mean responses between studies (mean values from the current study minus previous study mean values) were small (<1 point) for 11 of the questions. In 3 questions, however, the difference was greater than 1 point in either direction. Compared with the prior cohort, the ASA webinar-associated respondents were more likely to feel that their job satisfied them economically but also that they had less control at work, and they were more likely to report that their work kept them from friends and family.
Five hundred twenty-eight people answered the Honesty Survey. Most respondents (97.1%) said they would not be offended by such questions, and 73% said they would not be concerned about what would be done with the information. When asked about what safeguards would increase the likelihood of honest answers, anonymity was the most common answer. Similarly, most people said they would answer honestly if anonymity were assured. There was a nearly even split as to whether an opt-out feature would improve participation and a nearly even split as to whether paper or electronic forms assured anonymity more effectively.
We studied 170 people participating in a webinar on the topic of burnout. Subjects were predominantly physicians, 74% male, 83% over 40 years old, and 71% in nonteaching practices. Twenty-six to 59% of the respondents had a Maslach score—EE, DP, or LPA—suggestive of burnout, but only 15% had all 3. All burnout scores were independently associated with the SF-12 MCS, while only EE and LPA were associated with the SF-12 PCS. Burnout scores were associated with some elements of the SSPC-14, but there was no evidence of association between burnout scores and alcohol, cannabis, or tobacco use.
Each person has the required elements to develop burnout since, to some degree, we all have a certain amount of EE, DP, and LPA. Most people have occasional imbalance from either personal or situational characteristics (or both) but are still able to cope. For the remainder, if these situational and personal characteristics get out of balance with their life and work, they are unable to cope, they become disengaged, and burnout results (Figure 3). Certain people at increased risk for burnout include those who are younger, highly educated, unmarried, with high expectations and low self-esteem.7,8,17,26 Previously, we found that residents had more risk characteristics than did attending physicians or nurses,7,8,17 but even mature physicians12 and department chairs10,27,28 are not immune. In addition, not all specialists have the same risk of burnout; interspecialty variation in the incidence of burnout is common among physicians.17
The gender effect on burnout has been reported but is not firmly established. Balch et al9 reported male surgeons were at lower risk for burnout than their female colleagues. However, not all studies have found females to be at higher risk. We previously reported no gender association with burnout in a cross-section of perioperative providers.17 Completely different results were found by Merlani et al,29 who found that being male was associated with a high risk of burnout in intensive care unit providers. Our present results are most like Merlani’s in that males had a greater risk for DP, LPA, and total burnout score, but not for EE. Each of these studies has a good distribution of men and women, but each study is looking at a different set of influences on its specific cohort. Perhaps gender alone is not an independent risk factor for the development of burnout. Perhaps there is some other confounding interaction. Until future studies examine and clarify these questions, it would be prudent not to make any generalizations about gender effects on burnout.
Some have suggested30–32 that illness-related absences are common in burned out employees. A review by Kuhn and Flanagan,33 looking at literature back to 1986 dealing with the topic of burnout, depression, and suicide, suggests that the incidence of all these findings is on the rise. The number of publications dealing with physician mental health topics, in either the lay or the professional press, would seem to confirm that observation. We know of no previous studies that directly measure the associations between burnout indices and the MCS or PCS of the SF-12. With the exception of DP, we found that poor MCS and PCS health were independently associated with greater burnout risk in each MBI-HSS category. While these data do not establish whether the incidence is changing or which came first, they do suggest that burnout, physical problems, and mental health problems likely coexist and could affect overall health-related quality of life. Although it is reasonable to expect that certain aspects of burnout (particularly EE) would overlap with depression, further work is necessary to confirm these findings. In Maslach et al’s8 words, “…there are important distinctions between ‘burnout’ and ‘depression.’ Depression is a clinical syndrome, whereas, burnout describes a crisis in one’s relationship with work.”
There were few overarching trends associating SSPC with individual burnout elements (EE, DP, LPA), but we did observe a relationship with total burnout score. To tease more information from our results, we compared them with those of our 2011 study,17 where we utilized SSPC and MBI-HSS to evaluate providers from a single surgical unit. Compared to that study, more respondents felt economically satisfied, but at the same time, they felt less in control of their daily work. They also felt their jobs kept them from family and friends, that is, work-life imbalance.
In this light, it is strange that the same group felt less in control and had a less favorable work–life balance. Demographics again help explain this finding. There are only 2 nurses in the current study, whereas 63 (43%) of respondents in the prior study were nurses or nurse anesthetists. Nurses generally were better off in the areas of burnout, social support, and personal coping than were physicians.17 In most hospitals, whether teaching or community-based, nurses are for the most part removed from the stressful business aspects of patient care. Physicians, particularly in community settings, are often faced with nonpatient-related administrative duties and workforce demands33 and are thus susceptible to the uncertainties and lack of control that cause stress.12 While many anesthesia groups have been acquired by large management companies, many remain independent. The leaders of these independent groups remain responsible for the fiscal well-being of these practices. We cannot say with certainty whether being in a teaching or community hospital would change the incidence of burnout and stress for anesthesiologists, but it has been shown that private practice surgeons have more burnout than do academic surgeons.9
Substance use in physicians is not new,35 but identifying precipitating factors can help with understanding the problem and creating interventions. O’Connor and Spickard36 think it reasonable that substance use habits of practicing physicians are based on “behaviors during and before medical training.” In 1988, Juntunen et al16 studied alcohol consumption habits of Finnish doctors and reported that they drank more than the general population and that their drinking was associated with burnout. Unfortunately, they did not use a specific burnout measuring tool to reach this conclusion. Studies of physician substance abuse are fraught with problems of bias, and underreporting maybe more common than overreporting. Brooke et al15 point out that the stigma associated with addiction “forms a barrier” that prevents physicians from getting necessary help, as well as potentially biasing study results. We were unable to detect a relationship between substance use (tobacco, cannabis, alcohol) and any element of burnout after adjusting for other factors, although the association between alcohol use and LPA was near the threshold of statistical significance (P = .06; see Table 1). However, there is substantial statistical uncertainty regarding substance use, and our findings do not exclude the possibility of association with burnout (eg, the upper 95% confidence limit for the effect of tobacco use on total burnout is larger than the estimated effect of gender).
As with all survey-based studies, the results are biased. Participants were initially recruited because of an a priori interest in burnout and the burnout webinar, but we do not know why they were interested in the topic. Regardless of participants’ reasons for opting in or out of the survey, the final cohort is demographically similar to ASA-attending members, albeit somewhat older (see Table 2).
We also tried to maximize response to potentially “more sensitive” questions (ie, questions on substance usage), primarily by placing them at the end of the survey and providing assurances of anonymity.19,37 The percentage of participants completing all survey questions was nearly 77%, more than twice the response rate of some other studies.9 Although 96.5% of respondents answered the substance use questions, we wondered whether they answered them with the same honesty as with the remainder. In our short Honesty Survey, we found that most people would answer honestly if anonymity were assured. There was a nearly even split as to whether a “prefer not to answer” option improved participation or whether paper or electronic forms more effectively assured anonymity. We feel confident that people answered all questions honestly.
Studies on effectively identifying burned out individuals remain necessary; however, studies describing effective treatment options are also needed.33 A Cochrane database review, published in 2014, looked at controlled trials up to November 2013 on the subject of treating job-related stress in health care professionals.38 Of thousands of studies, only 58 studies met the inclusion criteria for the review, and even in these, the evidence was of low quality at best. Future studies will need to confirm the salutary effects of such things as personal and professional support and coping so that special treatments with measurable outcomes can be developed.
Despite the critical need for research on therapy, there is still a need for studies of burnout itself. For example, how does burnout compare across practice types (eg, fee-for-service versus salaried, academic practice versus private practice)? Are there regional and national trends in burnout incidence? Does a fellowship or a subspecialty practice change the risk of burnout? How does burnout in anesthesia providers compare with burnout in other health care providers (including both physician and nursing specialties)? What is the definitive answer on the effects of gender on burnout? These studies are necessary to identify groups most in need of help so that necessary resources are utilized in the best places to do the most good.
Because burnout tends to affect younger people more,7,8 and residents were underrepresented compared to the ASA-attending membership, the actual risk of burnout may be higher. Even though a large number still exhibit one or more high-risk characteristics, anesthesiologists as a group may have a lower risk of burnout than some other medical specialists.17 The burned-out individual is likely to have physical or mental health issues or both. Although mostly economically satisfied, these participants felt less in control at work and felt work kept them away from friends or family. We were unable to detect a relationship between burnout and substance abuse with the questions we asked. However, the continued inclusion of questions of this type—perhaps a different questionnaire—may be of importance in identifying potential at-risk groups.
Name: Steve Alan Hyman, MD.
Contribution: This author helped in literature search, design the study, collect the data, analyze and interpret the data, and write and edit the manuscript.
Name: Matthew S. Shotwell, PhD.
Contribution: This author helped design the study, analyze and interpret the data and figures, and write and edit the manuscript.
Name: Damon R. Michaels, MMHC, BS.
Contribution: This author helped design the study, collect the data, and edit the manuscript.
Name: Xue Han, MPH, MS.
Contribution: This author helped analyze and interpret the data and figures.
Name: Elizabeth Borg Card, MSN, APN, FNP-BC, CPAN, CCRP.
Contribution: This author helped design the study, collect and interpret the data, and write and edit the manuscript.
Name: Jennifer L. Morse, MS.
Contribution: This author helped collect and analyze the data.
Name: Matthew B. Weinger, MD.
Contribution: This author helped design the study and edit the manuscript.
This manuscript was handled by: Nancy Borkowski, DBA, CPA, FACHE, FHFMA.
1. Freudenberger HJThe staff burn-out syndrome in alternative institutions. Psychother Theo Res Prac. 1975;12:73–82.
2. Freudenberger HJBurn-out: occupational hazard of the child care worker. Child Care Q. 1977;6:90–99.
3. Dell’Erba G, Venturi P, Rizzo F, Porcù S, Pancheri PBurnout and health status in Italian air traffic controllers. Aviat Space Environ Med. 1994;65:315–322.
4. Girodo MThe psychological health and stress of pilots in a labor dispute. Aviat Space Environ Med. 1988;59:505–510.
5. Little LF, Gaffney IC, Rosen KH, Bender MMCorporate instability is related to airline pilots’ stress symptoms. Aviat Space Environ Med. 1990;61:977–982.
6. Singh RGRelationship between occupational stress and social support in flight nurses. Aviat Space Environ Med. 1990;61:349–352.
7. Maslach C, Jackson SE, Leiter MPThe Maslach Burnout Inventory
. 3rd ed. Palo Alto, CA: Consulting Psychologists Press, 1996.
8. Maslach C, Schaufeli WB, Leiter MPJob burnout. Annu Rev Psychol. 2001;52:397–422.
9. Balch CM, Shanafelt TD, Sloan JA, Satele DV, Freischlag JADistress and career satisfaction among 14 surgical specialties, comparing academic and private practice settings. Ann Surg. 2011;254:558–568.
10. De Oliveira GS Jr, Ahmad S, Stock MC, et al.High incidence of burnout in academic chairpersons of anesthesiology: should we be taking better care of our leaders? Anesthesiology. 2011;114:181–193.
11. De Oliveira GS Jr, Almeida MD, Ahmad S, Fitzgerald PC, McCarthy RJAnesthesiology residency program director burnout. J Clin Anesth. 2011;23:176–182.
12. Spickard A Jr, Gabbe SG, Christensen JFMid-career burnout in generalist and specialist physicians. JAMA. 2002;288:1447–1450.
13. Devi SDoctors in distress. Lancet. 2011;377:454–455.
14. Brown SD, Goske MJ, Johnson CMBeyond substance abuse: stress, burnout, and depression as causes of physician impairment and disruptive behavior. J Am Coll Radiol. 2009;6:479–485.
15. Brooke D, Edwards G, Taylor CAddiction as an occupational hazard: 144 doctors with drug and alcohol problems. Br J Addict. 1991;86:1011–1016.
16. Juntunen J, Asp S, Olkinuora M, Aärimaa M, Strid L, Kauttu KDoctors’ drinking habits and consumption of alcohol. BMJ. 1988;297:951–954.
17. Hyman SA, Michaels DR, Berry JM, Schildcrout JS, Mercaldo ND, Weinger MBRisk of burnout in perioperative clinicians: a survey study and literature review. Anesthesiology. 2011;114:194–204.
18. Ware J Jr, Kosinski M, Keller SDA 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity. Med Care. 1996;34:220–233.
19. Dillman DA, Smyth JD, Christian LM, Dillman DAInternet, Mail, and Mixed-Mode Surveys: The Tailored Design Method. 2009.3rd ed. Hoboken, NJ: Wiley & Sons.
20. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JGResearch electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–381.
21. Ware JE JrSF-36 health survey update. Spine (Phila Pa 1976). 2000;25:3130–3139.
22. Kudielka BM, Hanebuth D, von Känel R, Gander ML, Grande G, Fischer JEHealth-related quality of life measured by the SF12 in working populations: associations with psychosocial work characteristics. J Occup Health Psychol. 2005;10:429–440.
23. Voltmer E, Schwappach DL, Frank E, Wirsching M, Spahn CWork-related behavior and experience patterns and predictors of mental health in German physicians in medical practice. Fam Med. 2010;42:433–439.
24. Walt JG, Rendas-Baum R, Kosinski M, Patel VPsychometric evaluation of the Glaucoma Symptom Identifier. J Glaucoma. 2011;20:148–159.
25. National Institute on Drug Abuse. Available at: http://www.drugabuse.gov/nmassist
. Accessed November 29, 2016.
26. De Oliveira GS Jr, Rahmani R, Fitzgerald PC, Chang R, McCarthy RJThe association between frequency of self-reported medical errors and anesthesia trainee supervision: a survey of United States anesthesiology residents-in-training. Anesth Analg. 2013;116:892–897.
27. Johns MM III, Ossoff RHBurnout in academic chairs of otolaryngology: head and neck surgery. Laryngoscope. 2005;115:2056–2061.
28. Gabbe SG, Melville J, Mandel L, Walker EBurnout in chairs of obstetrics and gynecology: diagnosis, treatment, and prevention. Am J Obstet Gynecol. 2002;186:601–612.
29. Merlani P, Verdon M, Businger A, Domenighetti G, Pargger H, Ricou BSTRESI+ Group. Burnout in ICU caregivers: a multicenter study of factors associated to centers. Am J Respir Crit Care Med. 2011;184:1140–1146.
30. Ahola K, Kivimäki M, Honkonen T, et al.Occupational burnout and medically certified sickness absence: a population-based study of Finnish employees. J Psychosom Res. 2008;64:185–193.
31. Borritz M, Rugulies R, Christensen KB, Villadsen E, Kristensen TSBurnout as a predictor of self-reported sickness absence among human service workers: prospective findings from three year follow up of the PUMA study. Occup Environ Med. 2006;63:98–106.
32. Toppinen-Tanner S, Ojajärvi A, Väänänen A, Kalimo R, Jäppinen PBurnout as a predictor of medically certified sick-leave absences and their diagnosed causes. Behav Med. 2005;31:18–27.
33. Kuhn CM, Flanagan EMSelf-care as a professional imperative: physician burnout, depression, and suicide. Can J Anaesth. 2017;64:158–168.
34. Kumar SBurnout and doctors: prevalence, prevention and intervention. Healthcare (Basel). 2016;4:37.
35. Gundersen LPhysician burnout. Ann Intern Med. 2001;135:145–148.
36. O’Connor PG, Spickard A Jr.Physician impairment by substance abuse. Med Clin North Am. 1997;81:1037–1052.
37. Siebert DMaximizing response rates in survey research: Issues and methods. Advances in Social Work. 2006;7:1–11.
38. Ruotsalainen JH, Verbeek JH, Marine A, Serra CPreventing occupational stress in healthcare workers. Cochrane Database Syst Rev. 2014:CD002892.