|Year : 2018 | Volume
| Issue : 1 | Page : 35-49
A cross-sectional online evaluation of burnout risk factors among general surgical residents in Canada
Simon Timothy Adams1, Zeeshan Rana1, Rhonda Bryce2, Francis Christian1
1 Department of General Surgery, Royal University Hospital, Saskatoon, SK, Canada
2 Department of Community Health and Epidemiology, Clinical Research Support Unit, University of Saskatchewan, Saskatoon, SK, Canada
|Date of Submission||14-Oct-2017|
|Date of Acceptance||13-Dec-2017|
|Date of Web Publication||23-Apr-2018|
Dr. Simon Timothy Adams
Department of General Surgery, Royal. University Hospital, University of Saskatchewan, 103 Hospital Drive, Saskatoon, SK, S7N 0W8
Source of Support: None, Conflict of Interest: None
Background: Burnout is hallmarked by physical and psychological exhaustion, coupled with cynicism and disengagement. Evidence suggests that affected physicians not only suffer personally, but that patient safety and clinical outcomes are also negatively affected. This study aimed to identify potentially remediable risks for burnout among residents enrolled in Canadian general surgery programs.
Methods: In this cross-sectional design, a questionnaire was distributed to every general surgery resident in the 15 programs consenting to participate. Questions examined the following five domains: demographics, working patterns, attitudes toward residency, life experiences, and lifestyle/outlook. Respondents' risks of burnout were assessed using the Maslach Burnout Inventory™. Univariate analysis and then multiple logistic regression were used to assess predictors.
Results: A total of 114 completed questionnaires were received (22%). Of these residents, 39 (34%) met the criteria for high burnout risk. Inadequate personal/family time, a personal history of mental health or substance abuse-related issues, and moderately to poorly approachable staff/senior residents were all significantly associated with a high burnout risk (odds ratio [OR] =4.3, 95% confidence interval [CI] =1.6, 11.2, P = 0.003; OR =6.0, 95% CI = 1.6, 21.9, P = 0.007; and OR = 4.6, 95% CI = 1.7, 12.5, P = 0.003 respectively). Predicted high burnout risk probability with none of the above factors was 10%, increasing up to 40%, 75%, and 93% with one, two, or all of these risk factors present respectively.
Conclusion: One-third of general surgery residents in Canada are at high burnout risk. Residency programs may have considerable influence over factors associated with this outcome to the benefit of residents, staff, and patients.
The following core competencies are addressed in this article: Practice-based learning and improvement, Professionalism, Systems-based practice.
Keywords: Burnout, Canadian, general surgery, Maslach Burnout Inventory™, residency
|How to cite this article:|
Adams ST, Rana Z, Bryce R, Christian F. A cross-sectional online evaluation of burnout risk factors among general surgical residents in Canada. Int J Acad Med 2018;4:35-49
|How to cite this URL:|
Adams ST, Rana Z, Bryce R, Christian F. A cross-sectional online evaluation of burnout risk factors among general surgical residents in Canada. Int J Acad Med [serial online] 2018 [cited 2018 Jul 17];4:35-49. Available from: http://www.ijam-web.org/text.asp?2018/4/1/35/230857
| Introduction|| |
Burnout is a psychological condition characterized by physical and mental exhaustion, a lack of enthusiasm and motivation, and feelings of ineffectiveness, frustration, and cynicism., It is prevalent in healthcare, particularly among physicians in stressful specialties., Residents and students also experience high levels of burnout that can affect both career choice and progression, impeding abilities to learn and care for patients.,,,
Residency burnout has been evaluated in medical literature since the late 1980s and multiple risk factors have been identified (e.g. age, sex, marital status, culture, workload, autonomy, etc.), albeit with varying consistency. In general, no easily applied screening criteria are readily available for program directors to better support those most likely to be at risk.
The aims of this study were threefold: to determine the prevalence of high burnout risk among Canadian general surgery residents, to identify remediable factors, and to quantify risk at differing combinations of predictive characteristics.
| Methods|| |
To evaluate potential predictors of burnout a 25-item questionnaire was devised by the authors containing questions covering the following five broad domains: resident demographics, working patterns, attitudes toward residency, life experiences, and lifestyle/outlook [see [Figure 1] and [Figure 2] for complete questionnaire]. The questions and domains were derived partly from the existing literature [Appendix A] and partly from factors felt to be worthy of investigation by the authors. The questionnaire required respondents to give answers in a variety of formats including dichotomous responses, Likert scales, and open text.
|Figure 1: Questionnaire (demographics and working patterns) with response options shown in brackets|
Click here to view
|Figure 2: Questionnaire (attitudes toward residency, life experiences, and lifestyle/outlook) with response options shown in brackets|
Click here to view
Although several tools exist for the measurement of burnout more than 90% of studies utilize the Maslach Burnout Inventory ™ (MBI ™)., This well-validated tool has been refined over the years and developed into several career/role-specific versions, including one for healthcare professionals. Respondents are asked to indicate on a 7-option Likert scale how frequently 22 work-related questions are true in their experience. These 22 questions measure three overall domains: emotional exhaustion, depersonalization, and personal achievement. As per the tool's scoring instructions, a respondent is considered to be at high risk of burnout if he/she scores highly for emotional exhaustion(≥27) in addition to having a high score for depersonalization (≥13) and/or a low personal achievement score (≤31).,
Following the development of the questionnaire a license to distribute the healthcare provider version of the MBI ™ was obtained. Both the questionnaire and the MBI ™ were translated into French and each language version was incorporated into an anonymized online survey tool. An E-mail containing a link to the survey was then composed, both in English and French, inviting residents to participate in the study. Approval was obtained from our institutional research ethics board prior to distributing the survey.
The study was based at the University of Saskatchewan with data collection occurring between October 2015 and December 2015. Surveying was initiated by contacting all 17 Canadian general surgery residency program directors and administrators via information taken from the Canadian Resident Matching Service website. The program directors were approached via E-mail and asked to distribute the invitation E-mail to their residents. In addition, the Canadian Association of General Surgeons published the link to the survey on their weekly members E-mail at the start of the 3-month data collection to period. Several reminding E-mails were sent out during this time. The maximum number of responses possible was calculated using information from the Canadian Post-MD Education Registry (CAPER) website, detailing the total number of residents enrolled in each program at the beginning of the 2015–2016 academic year.
The inclusion criteria included all residents enrolled in general surgery programs in Canada at the start of the 2015–2016 academic year. The exclusion criteria were <100% completion of the burnout assessment portion of the survey (i.e. outcome status could not be determined) or complete absence of information regarding any of the covariates (i.e. did not contribute to the evaluation of the association between any covariate and outcome).
The primary end point of the study was burnout risk (dichotomized as high risk versus low-to-moderate risk) as calculated using the MBI ™ which was then compared to the questionnaire responses using Chi-square testing, or Fisher's exact testing if more than 20% of expected cell counts were <5. Categories with low counts were collapsed where clinically relevant and associations with the outcome were similar. Those factors showing a significant individual Chi-square association with burnout risk (P< 0.05 being considered significant) were initially included in the multiple logistic regression analysis as covariates; a backward step-wise approach was subsequently utilized to determine which factors remained significant as independent predictors of high burnout risk when the potential confounding effects of the other significant factors were controlled for. The empirical form of the model was also used to examine predicted probabilities of high burnout risk for specific combinations of the included predictors to quantify the burnout probabilities suggested by these individual factors. The statistical analyses were performed using SPSS, version 22 (IBM Corp., Armonk, NY, USA, 2013).
As the study goal was to evaluate risk factors comprehensively, rather than to focus on one key predictor, an a priori sample size was not calculated. However, for this more general approach, model building strategies for a dichotomous outcome frequently employ an approximation of 10 individuals in the less common category per variable to develop a model appropriate to the sample size. This principle was kept in mind in the development of the multiple regression model to avoid overfitting. Variables with small cell sizes that required Fisher's exact testing on univariate assessment were also not included in the modeling procedure to limit concerns related to sparse data.
| Results|| |
Of the 17 program directors contacted 15 agreed to distribute the survey. A total of 114 completed questionnaires were received from a potential maximum of 518 (22% response rate). General characteristics of the respondents are summarized in [Table 1].
Of the 114 respondents meeting the inclusion criteria 39 (34%) scored as being at high risk of clinical burnout. Several factors showed a significant association with high risk of burnout on univariate analysis [[Table 2]; see [Appendix B] for all associations]. When factors with univariate P < 0.05 were analyzed together as covariates using multiple logistic regression the combination of inadequate personal/family time (IFT), a personal history of mental health or substance abuse-related issues (MHSA), and moderately to poorly approachable staff/senior residents (PAS) remained significant [Table 3]. From the model-based estimates residents with none of the three aforementioned factors (n = 42 [37%]) had a 10% probability of being at high risk of clinical burnout. Those with one (n = 46 [ 40%]), two (n = 20 [18%]), or all three of these factors (n = 6 [5%]) however had predicted high risk probabilities of 32%–40%, 69%–75%, and 93% respectively [Table 4]. Effect modification could not be assessed due to the relatively small cell sizes and subsequent model instability when two-way interaction terms between these terms were included in the models.
|Table 2: Statistically significant univariate predictors* of high versus moderate-to-low burnout risk|
Click here to view
|Table 4: Predicted probability of high burnout risk using the empirical form of the multiple logistic regression model|
Click here to view
Differences in the sensitivity and specificity of utilizing these risk factors singularly or in combination are presented in [Figure 3]. When the presence of any of these three predictors was used to identify sample participants in whom high burnout risk may be a concern nearly all who did score highly on the MBI were detected (34 of 39); however, half of those considered to be lower risk were incorrectly identified (38 of 75). Correspondingly, the presence of at least one of the three risk factors in an individual had a sensitivity of 87.2% and a specificity of 49.3% for high burnout risk. In contrast, if all three risk factors were required in an individual before high burnout risk was suspected only 5 of the 39 individuals identified using the MBI ™ would be detected; however, individuals who scored as lower risk were almost never labeled as a false positive (1 of 75). Under this criterion the presence of all three factors had a sensitivity of 12.8% but a specificity of 98.7% for high burnout risk. The corresponding values for the use of two or more positive factors as high risk identification were 21 of 39 and 5 of 75 (sensitivity 53.8% and specificity 93.3%).
|Figure 3: Sensitivity and specificity representation. When the presence of any, two or more, or all three factors were used to detect high burnout risk the sensitivity and specificity values respectively were 87.2% and 49.3%, 53.8% and 93.3%, and 12.8% and 98.7%|
Click here to view
Of the three risk factors identified in the preceding paragraph, IFT was the most frequently occurring predictor. Fifty-eight participants (51%) reported that their amount of personal/family time was “poor” to “extremely poor” with 34 (30%) citing it as their only risk factor. Each of the three factors included in our predictive model had similar individual probabilities for high risk of burnout (32%, 34%, and 40% for IFT, PAS, and MHSA respectively) but PAS and MHSA were considerably less common (PAS was indicated by 29 individuals in total [25%] with 7 [6%] having PAS as their only risk factor; MHSA was indicated by 17 individuals in total [15%] with 5 [4%] having MHSA as their only risk factor).
| Discussion|| |
Our study, a nationwide anonymous survey of Canadian general surgery residents, aimed to capture the frequency of such residents being at high risk of clinical burnout (34%) and to identify factors that would be amenable to intervention for prevention of its inception or deterioration. From the 114 participants who responded it was recognized that IFT, MHSA, and PAS are important, potentially remediable, and independent predictors of high burnout risk.
As outlined in the introduction the existing literature on burnout has demonstrated multiple serious negative consequences of the condition for both physicians and their patients.,,,,,,,,,,, To date no studies have attempted to determine its prevalence among residents in Canadian general surgery training programs. By evaluating general surgery residents specifically, using only a small number of comparatively easily determined and potentially remediable factors, we believe that program directors could not only identify “at-risk” residents but also institute specific measures at an early stage before the development of full-blown burnout. This study, therefore, adds to the existing literature firstly by providing an indication of the prevalence of high clinical burnout risk among this specific population and secondly by highlighting three key factors which can be identified and acted upon by programs in order to improve both the mental and physical well-being of their residents. This identification of the three key risk factors is beneficial from a practicality perspective as although there are several validated tools available for the measurement of burnout their metrics render them impractical for the real-world evaluation of residents in either a large-scale or longitudinal format.
Comparing our results to those of other similar studies is fraught with difficulties. A recent similar study looking at the levels of burnout among 665 American general surgery residents from 246 accredited programs reports a 69% rate of high burnout risk but, significantly, does not include the response rate yielded or the maximum potential number of responses. Another study conducted among French surgical trainees shows a 52% risk of high burnout with a response rate of 66%. Our rate of 34%, therefore, seems low by comparison; however, when one examines the literature in detail, there is considerable heterogeneity with regard to the definition of high burnout risk employed even when the same tool is used for its measurement. In order to avoid contributing to this heterogeneity the authors chose to adhere to the definition of high burnout risk as described in the original MBI ™ published in 1981. The factors included in our predictive tool do appear to be congruent with the majority of the findings of similar studies in which burnout appears to be most closely related to working hours and environment within specific specialties at the trainee or resident level.
One clear potential limitation of this study is the 22% response rate which is certainly suboptimal but by no means unusual for questionnaire surveys distributed to physicians., Consequent to the low response rate is the potential for a Type 1 statistical error in the predictive model, namely that the relative infrequency of the PAS and MHSA risk factors in the data set, coupled with their statistically significant associations with the outcome, could have been proven to be false positives if a greater response rate had been achieved. For this reason the variables of age (35 years or older) and long interval since medical school (>5 years), although statistically significant and potentially important, were not included in the multiple regression modeling. Other factors may not have been recognized due to inadequate power.
Additionally, the low response rate raises the potential issue of inadvertent selection bias. Residents more prone to burnout may have been more likely to choose to participate in a study where burnout is the central feature thus overestimating the prevalence for the population as a whole. Similarly, residents at high risk of burnout may identify with the proposed factors more readily than those at low risk thus increasing their enthusiasm for completion of the survey and overstating any associations with the outcome. In an effort to better evaluate the representativeness of our sample a review of national general surgery resident characteristics reported by CAPER for the corresponding academic year, as well as those reported in a 2012 Canadian study, suggests that our sample is relatively similar to Canadian general surgery residents in regard to age, sex, marital and parental status, and international medical graduate status., It is notable that in our sample, compared to the 2012 survey, teaching frequency was reported to be considerably lower and rate of poor or moderate teaching satisfaction considerably higher although these differences may have been definition dependent. However, in terms of inadequate work-life balance, 42% of the 2012 survey respondents felt this to be true: similar to the 50% in our sample who felt that they worked too many hours. Comparison of the available characteristics is provided in [Appendix C].
The cross-sectional nature of this work is another limitation to our conclusions of cause and effect - burnout itself can precipitate a loss of perspective, particularly regarding familial and professional relationships; therefore, it could be that IFT and PAS are in fact the consequences rather than the causes of the condition. This might explain why seven of the eleven factors showing a statistically significant association with high burnout risk following univariate evaluation [Table 2] were from the attitudes toward residency domain. Whether these factors precipitate burnout, or it is indeed the reciprocal, is a dynamic that cannot easily be determined, particularly using an impersonal data collection method such as an electronically distributed questionnaire.
The advantages and disadvantages of web-based questionnaires and their usefulness in epidemiological studies have been well documented in the literature. Although they possess significant advantages such as earlier responses, quicker completion, and reduced costs, web-based studies have been shown to have consistently lower overall response rates compared to the more traditional modes of data collection. However, potential selection bias notwithstanding, the authors believe that the number of residents shown to be at high risk of burnout in our study is sufficiently great to warrant further attention.
Additional thought should be given to one of the fundamental aims of the study, namely the assumed potential to remediate the identifiable factors. The most appropriate model identified IFT, a personal history of MHSA, and PAS as the three factors which in combination gave the most reliable prediction for high risk of burnout. While one can conceive of how each of these could be addressed, any remedial measures implemented by a program could have implications relevant to surgical training. The concept of IFT is clearly subjective and how best to address this without significantly impacting either clinical training or patient care would be challenging. It is, however, by far the most common of the three factors and is perhaps the one that would be easiest to recognize and resolve. In contrast, if a resident perceives that staff are unapproachable, there are only a limited number of avenues by which a program director can resolve this. It is not unheard of for staff to have the responsibility of teaching residents taken away from them, but this is not always feasible, particularly if the staff member in question is a senior figure in the faculty. Similarly, it should be borne in mind that the source of the personality clash may not be the staff member and that the problem may originate with the residents themselves. Finally, although supportive measures can easily be provided for residents with MHSA, the perceived career implications and ethical issues surrounding the disclosure of this information may be a barrier to its detection and subsequent resolution. This requires a larger cultural shift in the perception of MHSA at all levels within the department and the larger institution in which it is based.
The identified risk factors are good starting points for identifying and preventing burnout among general surgery residents, but further refinement would be useful. While screening for all three of these factors, with concern raised at the detection of any of the factors, appears to identify virtually all residents at high risk, it also identifies many who are not. In contrast, requiring positive responses on a minimum of two factors before intervening would miss many residents in potential distress; however, given the high specificity of this approach, a positive response to more than one of the three risk factors should raise substantial concern for high burnout risk. Even so, the findings should be approached cautiously given the low response rate to the survey.
| Conclusion|| |
Our results suggest that one-third of general surgery residents in Canada are at a high risk of clinical burnout. They also suggest that there are specific, identifiable risk factors which show an association with this outcome. Recognizing these factors there are measures that residency programs could adopt, both to limit their impact as a means of preventing burnout in vulnerable subgroups and in treating those already experiencing symptoms. By doing so programs would be benefitting resident well-being and education while minimizing the risk to patient safety.
Financial support and sponsorship
No financial support or sponsorship was received in conjunction with this study.
Conflicts of interest
There are no conflicts of interest.
Ethical conduct of research
This study was approved by the University of Saskatchewan Research Ethics Board
| References|| |
Ruotsalainen JH, Verbeek JH, Mariné A, Serra C. Preventing occupational stress in healthcare workers. Cochrane Database Syst Rev 2014;12:CD002892.
Maslach C, Schaufeli WB, Leiter MP. Job burnout. Annu Rev Psychol 2001;52:397-422.
Kumar S. Burnout and doctors: Prevalence, prevention and intervention. Healthcare (Basel) 2016;4:37.
Oskrochi Y, Maruthappu M, Henriksson M, Davies AH, Shalhoub J. Beyond the body: A systematic review of the nonphysical effects of a surgical career. Surgery 2016;159:650-64.
Cook AF, Arora VM, Rasinski KA, Curlin FA, Yoon JD. The prevalence of medical student mistreatment and its association with burnout. Acad Med 2014;89:749-54.
Ishak W, Nikravesh R, Lederer S, Perry R, Ogunyemi D, Bernstein C, et al.
Burnout in medical students: A systematic review. Clin Teach 2013;10:242-5.
Dyrbye LN, West CP, Satele D, Boone S, Tan L, Sloan J, et al.
Burnout among U.S. medical students, residents, and early career physicians relative to the general U.S. population. Acad Med 2014;89:443-51.
Ishak WW, Lederer S, Mandili C, Nikravesh R, Seligman L, Vasa M, et al.
Burnout during residency training: A literature review. J Grad Med Educ 2009;1:236-42.
Qiao H, Schaufeli W. The convergent validity of four burnout measures in a Chinese sample: A confirmatory factor-analytic approach. Appl Psychol Int Rev 2011;60:87-111.
Maslach C, Jackson SE. The measurement of experienced burnout. J Organ Behav 1981;2:99-113.
Hosmer D, Lemeshow S, Sturdivant R. Applied Logistic Regression. Special Topics. 3rd
ed., Ch. 10. Hoboken, NJ: John Wiley and Sons Inc.; 2013.
Dewa CS, Loong D, Bonato S, Thanh NX, Jacobs P. How does burnout affect physician productivity? A systematic literature review. BMC Health Serv Res 2014;14:325.
Dewa CS, Jacobs P, Thanh NX, Loong D. An estimate of the cost of burnout on early retirement and reduction in clinical hours of practicing physicians in Canada. BMC Health Serv Res 2014;14:254.
Maslach C, Leiter MP. New insights into burnout and health care: Strategies for improving civility and alleviating burnout. Med Teach 2017;39:160-3.
Raptis DA, Schlegel A, Tschuor C, Clavien PA. Job satisfaction among young board-certified surgeons at academic centers in Europe and North America. Ann Surg 2012;256:796-803.
Hall LH, Johnson J, Watt I, Tsipa A, O'Connor DB. Healthcare staff wellbeing, burnout, and patient safety: A systematic review. PLoS One 2016;11:e0159015.
Salyers MP, Bonfils KA, Luther L, Firmin RL, White DA, Adams EL, et al.
The relationship between professional burnout and quality and safety in healthcare: A meta-analysis. J Gen Intern Med 2017;32:475-82.
Harms BA, Heise CP, Gould JC, Starling JR. A 25-year single institution analysis of health, practice, and fate of general surgeons. Ann Surg 2005;242:520-6.
Oreskovich MR, Kaups KL, Balch CM, Hanks JB, Satele D, Sloan J, et al.
Prevalence of alcohol use disorders among American surgeons. Arch Surg 2012;147:168-74.
Shanafelt TD, Balch CM, Bechamps G, Russell T, Dyrbye L, Satele D, et al.
Burnout and medical errors among American surgeons. Ann Surg 2010;251:995-1000.
McCray LW, Cronholm PF, Bogner HR, Gallo JJ, Neill RA. Resident physician burnout: Is there hope? Fam Med 2008;40:626-32.
Bhutani J, Bhutani S, Balhara YP, Kalra S. Compassion fatigue and burnout amongst clinicians: A medical exploratory study. Indian J Psychol Med 2012;34:332-7.
] [Full text]
Elmore LC, Jeffe DB, Jin L, Awad MM, Turnbull IR. National survey of burnout among US general surgery residents. J Am Coll Surg 2016;223:440-51.
Chati R, Huet E, Grimberg L, Schwarz L, Tuech JJ, Bridoux V, et al.
Factors associated with burnout among French digestive surgeons in training: Results of a national survey on 328 residents and fellows. Am J Surg 2017;213:754-62.
Pulcrano M, Evans SR, Sosin M. Quality of life and burnout rates across surgical specialties: A systematic review. JAMA Surg 2016;151:970-8.
Grava-Gubins I, Scott S. Effects of various methodologic strategies: Survey response rates among Canadian physicians and physicians-in-training. Can Fam Physician 2008;54:1424-30.
Cunningham CT, Quan H, Hemmelgarn B, Noseworthy T, Beck CA, Dixon E, et al.
Exploring physician specialist response rates to web-based surveys. BMC Med Res Methodol 2015;15:32.
Annual Census of Post-M.D. Trainees 2015-16. Ottawa, Ontario: Canadian Post-M.D. Education Registry, Association of Faculties of Medicine of Canada, 2015-2016.
Tomlinson C, Labossière J, Rommens K, Birch DW. The Canadian general surgery resident: Defining current challenges for surgical leadership. Can J Surg 2012;55:S184-90.
van Gelder MM, Bretveld RW, Roeleveld N. Web-based questionnaires: The future in epidemiology? Am J Epidemiol 2010;172:1292-8.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3], [Table 4]