High burnout among the South African orthopaedic community: a cross-sectional study

Background Burnout is epidemic among physicians, with the orthopaedic speciality displaying one of the highest rates of burnout in international studies. The burnout rate of the South African orthopaedic community is unknown. This study aimed to determine the prevalence and causes of burnout, as well as the coping mechanisms and associations with burnout, in South African orthopaedic surgeons and trainees. We conducted a cross-sectional, secure, online survey of members of the South African Orthopaedic Association. The survey assessed demographic characteristics, workload, professional fulfilment and burnout (utilising the Stanford Professional Fulfilment Index), associated workplace distress conditions, causes of and coping strategies for burnout. A response was not compulsory for any question. Statistical analysis was performed to assess for independent associations with burnout. hundred of 155) of respondents were male. Registrars accounted for 17% (27 of 155), while 83% (128 of 155) were qualified specialists. Respondents were in orthopaedic practice for a median of 17 years (IQR 9–28). Sixty per cent (76 of 127) practise in private, 17% (22 of 127) in public and 23% (29 of 127) in both sectors. The overall burnout rate was 72% (113 of 156). Burnout was associated with being the parent of young children and having fewer hours of sleep on call. Registrars were more likely to have burnout than consultants (OR 5.68, 95% CI 1.3–25.2). Gender, practice setting and subspeciality were not associated with burnout. Self-reported causes of burnout that were found to be associated with actual burnout were: ‘hours at work’, ‘lack of free time’ and ‘work–life imbalance’. No self-reported coping mechanisms were found to be protective in this cohort, but the use of alcohol as a coping mechanism was associated with an increased likelihood of burnout (OR 3.9, 95% CI 1.4–10.7). Respondents felt that the concurrent experience of the COVID pandemic at the time of running the survey reduced their experience of burnout. in the South African community is 72%. Trainees were found to be particularly vulnerable. There appears to be a need to develop, assess and implement effective system-related initiatives aimed at reducing the among and trainees Africa. middle-income (LMICs), much work been done in (RSA) to assess the prevalence of burnout among A national study found higher rates of burnout in medical doctors than comparative international figures. 18 Studies from various provinces reported rates between 52% and 100% in doctors of varying career levels and specialities, with none of these investigating orthopaedic surgeons exclusively.


Introduction
Burnout is described in the International Classification of Disease (ICD) 11th revision as an occupational phenomenon resulting from chronic workplace stress. 1 The syndrome is described as having three components: emotional exhaustion (lack of enthusiasm for work), depersonalisation (cynicism) and lack of professional satisfaction (low sense of personal accomplishment). [2][3][4][5][6][7] More than two decades of research have expounded the epidemic, which has notably affected the health professions disproportionately. 8 Burnout is particularly relevant in the health professions, due to the numerous downstream effects on the health system, on patients, and on the individual themselves. The estimated cost of burnout to the US health system is $4.6 billion annually. 9 This is due to higher staff turnover, staff attrition, early retirement, reduced work effectiveness and professionalism, absenteeism and presenteeism (being at work when one should be off sick). 4,5,[7][8][9][10][11] Physicians experiencing burnout have displayed higher rates of depression and suicidal ideation, broken relationships, increased alcohol and drug use, reduced work satisfaction, altered prescribing habits and test ordering, increased risk of medico-legal lawsuits and reduced cognitive ability. [7][8][9][10][11][12][13] Physical health can also be affected as burnout can result in higher rates of diabetes mellitus, cholesterol, pain, fatigue, cardiovascular disease and early death. 7,11 Within the branches of medicine, orthopaedic surgeons in the USA have one of the highest burnout rates. 4 Various international studies have demonstrated burnout rates of between 16% and 85%. [3][4][5]7,14 This variability is similar to that seen among general surgeons, and can be accounted for by differing career stages, workload, work hours and number of calls. 2,4,7 Orthopaedic trainees are at higher risk, with reports suggesting work hours, electronic medical records (EMR), lack of sleep and lack of supportive mentors as common predisposing factors. 5,15 Two studies identified female orthopaedic surgeons as being at greater risk of experiencing burnout. 15,16 Despite the paucity of data on physician burnout in low-and middle-income countries (LMICs), much work has been done in South Africa (RSA) to assess the prevalence of burnout among doctors. 17 A national study found higher rates of burnout in medical doctors than comparative international figures. 18 Studies from various provinces reported rates between 52% and 100% in doctors of varying career levels and specialities, with none of these investigating orthopaedic surgeons exclusively. 6,[19][20][21][22] We aimed to determine the burnout rate in South African orthopaedic surgeons and trainees. In addition, we wanted to evaluate the perceived causes and coping mechanisms, screen for the presence of associated workplace distress conditions, determine the social impact of burnout and determine the independent risk factors for the development of burnout in this cohort. While not a primary aim, we also evaluated the impact of the COVID pandemic on burnout.

Methods
This cross-sectional study was performed by way of a ten-minute online survey. The survey was administered using Checkbox ® Survey Software (© Checkbox Survey, Inc. 2002Inc. -2021. The questionnaire comprised six sections, namely, demographics, workload, the Stanford Professional Fulfilment Index (PFI), screening questions for associated workplace distress conditions, perceived causes of burnout and coping strategies for burnout. In considering the COVID pandemic and its potential effect on burnout, each section concluded with a question to determine if the pandemic had affected the response. There was a single open-ended question. The remaining questions required numerical answers or selection of appropriate single or multiple options.
The Stanford PFI is a validated assessment tool comprising 16 questions scored on a 5-point Likert scale of agreement; four questions aimed at establishing the presence of work exhaustion (WE); and six for interpersonal disengagement (ID). If the combined average score for these two variables was calculated at 1.33 or greater, it constituted burnout. 10 The other six questions assessed professional fulfilment (PF). Respondents were deemed professionally fulfilled if they scored an average of three points or greater for these six questions. 10 Five additional workplace distress conditions may be associated with burnout. 'Moral injury' is defined as the distress caused in a person that arises from the challenge of simultaneously knowing what care patients require but being unable to provide it due to constraints that are beyond their control. 23 'Compassion fatigue' is 'the change in empathetic ability of the caregiver in reaction to the prolonged and overwhelming stress of caregiving'. 24 'Approval addicts' are those whose self-esteem is attached to how other people perceive them. 25 'Imposter phenomenon' is when a person has the psychological belief that they are a fraud, and fear being recognised as an imposter, despite objective evidence of success. 26 The fifth condition, termed 'second victim syndrome', is used to characterise the feeling of responsibility that a healthcare worker experiences when a medical error occurs. 27 For each of these conditions, a single or multiple questions were adapted from existing validated scores, or formulated from defining characteristics of the syndrome, for the purpose of screening for each of the conditions. 25,[27][28][29][30] All the members of the South African Orthopaedic Association (SAOA) with listed email addresses were eligible to participate. Invitations were emailed to members, with an attached link to the secure online survey. To proceed with the survey, respondents had to consent to participation, after reading an information page on the nature of the study. Survey responses were completely anonymous, and respondents were limited to a single response. The survey was run for a total of six weeks over May and June 2021. Two reminders were sent using the same platform at twoweekly intervals. No questions were compulsory to complete; however, the PFI section was completed by all respondents.
The survey responses were exported to Microsoft ® Excel for Mac, Version 16.50 (21061301), where descriptive statistics were summarised and the open-ended question encoded by the general theme of response. Non-parametric continuous variables were summarised as medians with interquartile range (IQR), and categorical variables were expressed as percentages with counts. Statistical analysis was performed with jamovi statistical software (version 1.6.23.0). 31 Univariate binomial logistic regression was employed to assess the associations of the risk factors with the presence of burnout. A multiple regression model was developed using reverse stepwise elimination to determine independently associated variables. Significance was set at p < 0.05.

Demographic details
Of the SAOA members contacted, 960 had active email addresses. One hundred and fifty-six members completed the survey, constituting a response rate of 16%. The median age of respondents was 46.5 years (IQR 37-58). Ninety per cent of respondents (139 of 155) were male, and 10% were female (16 of 155). Qualified orthopaedic surgeons constituted 83% of the sample (128 of 155), and 17% (27 of 155) were registrars. Sixty per cent of the specialists were private practitioners (76 of 127), 17% (22 of 127) practised exclusively in the public sector, and 23% reported working in both public and private (29 of 127). The median number of years in practice was 17 (IQR 9-28).

Relationship factors
Of the entire group, 94% (143 of 153) reported being in a relationship. Ninety-four per cent of the respondents reported that

Working hours
Thirty-five per cent of respondents reported working between 40 and 50 hours per week (54 of 156), 25% worked between 50 and 60 hours (39 of 156), and 31% 40 hours (46 of 156) or less. Eleven per cent work more than 60 hours a week (17 of 156).
Most respondents reported that outpatient clinics, together with ward rounds or surgery, accounted for between 30% and 40% of the workload on average, followed by administration (20%), while teaching and research each contributed less than 10%. Forty per cent of respondents reported between two and four weekday calls

Causes of burnout
Regarding causes of burnout, 'too many bureaucratic tasks'

Workplace distress conditions
Of the screening questions for workplace distress conditions captured in Table IV, the most notable findings were that 84% of respondents (130 of 155) screened positive for moral injury, 71%

Effect of burnout on respondents
Burnout had a large or severe effect on 24% (37 of 155) of respondents, moderate effect on 37% (58 of 155), and a small or minimal effect on 39% (60 of 155) of respondents, respectively. Burnout impacted relationships in 51% (79 of 154) of cases. Thirtysix per cent (56 of 155) of respondents said they would take a

(156) 72%
Categorical variables expressed with counts and percentages of total *Professional fulfilment was scored on a scale of 0-4, where 0 is not true and 4 is entirely true # Work exhaustion and interpersonal disengagement scored on a scale of 0-4, where 0 is not at all and 4 is very much

Imposter phenomenon No Yes
Have you ever been surprised by your successful completion of a task or received promotion and felt it was most likely as a consequence of luck or 'being in the right place at the right time'? 99 (152) 65% 53 (152) 35% Do you find it hard to accept compliments or praise due to fear you will be 'found out' as a fraud or not as intelligent as you perceive people think you are? 109 (154)   Prefer not to answer 6/9 (67%) 3/9 (33%)  Table V provides a summary of the variables associated with burnout. Of the demographic and call-related characteristics, 'age of youngest child' (p ≤ 0.001), 'age of oldest child' (p = 0.002) and on-call hours of sleep (p = 0.003), were inversely associated with burnout. That is to say; burnout was associated with respondents who had younger children and fewer hours of sleep on call. Burnout was associated with being a registrar (p = 0.022), with no association to the current year of training ( Figure 3). There was no association between practice sector, gender, relationship status or workload. Of the self-reported causes of burnout,    respondents that selected 'hours at work', 'lack of free time' and 'work-life imbalance' as causes of burnout, were more likely to have measured burnout (Figure 4). No coping mechanisms were found to be associated with reduced burnout, but if respondents selected 'alcohol' as a coping mechanism, they had an increased risk of burnout ( Figure 5). Screening positive for moral injury or one measure of compassion fatigue, placed respondents at increased risk of burnout. Burnout was associated with the graded 'impact of burnout' measure (the greater the reported impact, the greater the risk of burnout), as well as to the reported impact on relationships. The multiple regression model that best explained the presence of burnout included the following factors: 'lack of free time', compassion fatigue, alcohol as a coping mechanism, 'impact of burnout on life', 'impact of burnout on relationships', age of children and on-call hours of sleep. Compassion fatigue, and the impact of burnout on relationships were found to be independently associated with burnout.

Impact of COVID on burnout
We found several associations between COVID and measured burnout (Table VI). Respondents who reported that COVID had affected their sleep, had an increased risk of burnout, with the

Discussion
We found a burnout rate of 72% in South African orthopaedic surgeons and trainees, which falls on the higher end of the spectrum compared to international studies (between 16% and 85%). 5 Numerous studies have identified several common associations with burnout and recognised groups within the health professions at higher risk of developing burnout. 2,8,32 The majority of associations reported with burnout are health-system related or work environment issues such as: lack of autonomy, too many bureaucratic tasks, administrative responsibilities, high workload, remuneration, working hours and calls (correlated to sleep deprivation). [2][3][4]8,10,11,14,15,[33][34][35][36] The other causes pertain to work-life imbalance including: having dependants and inability to balance work and family responsibilities. 2,5,7,11,32,36 Specific groups shown to be at greater risk for developing burnout comprise: physicians of younger age, female gender and trainees. 2,5,14 In our cohort, we found similar risk factors, such as having children of younger age and fewer on-call hours of sleep. Although most respondents reported that burnout was caused by 'lack of compensation' and 'bureaucratic tasks', the factors that placed them at increased risk were rather, 'hours at work', 'lack of free time' and 'work-life imbalance'. The significantly higher rate of burnout in registrars compared to specialists is consistent with findings from the USA and Australia. [37][38][39][40] This is a concern for training coordinators, as among many implications, burnout has notably been shown to contribute to poorer examination results. 40 We found no association between gender and burnout in our study; however, the study was underpowered to detect a difference.
Physicians attending to cancer patients have been shown to have an increased risk of burnout, and in the orthopaedic context specifically, the highest rates of burnout were in arthroplasty surgeons in China (85%). 10,41 Neither arthroplasty nor orthopaedic oncology were found to be associated with burnout in our study. Identifying as a paediatric subspecialist tended towards being protective against burnout but was underpowered to detect significance. This was an unanticipated finding. The Paediatric Orthopaedic Association of North America (POSNA) reports burnout of 38% among their members, while 28% (two of seven) of orthopaediatric specialists in our cohort had burnout. 16 Once burnout has been established, international reports show that it contributes to: higher rates of depression and suicidal ideation, broken relationships, increased alcohol and drug use, and reduced work satisfaction. 7,8,[10][11][12][13] Suicidal ideation rate (10%) in our cohort was similar to that of a USA Medscape study (13%) in which orthopaedic surgeons had the second highest rate across the specialities represented. 14 Also from the Medscape report was the finding that 47% of workplaces do not have programmes to manage burnout, compared to 63% in our cohort. 14 In addition, the Medscape respondents reported that even if workplace programmes to manage burnout did exist, 42% of them would be unlikely to attend it, similar to 38% reporting this in our cohort. The Medscape respondents reported 'fear of stigma in the workplace', or felt the condition was not 'severe' enough to warrant attendance at such a programme as reasons for unlikely attendance. Irrespective, participation in wellness programmes has only minimally improved burnout scores; likewise, resilience training studies reflect conflicting results. 7,12,33 We feel that this is because the main causal agent is at the system or organisation level, rather than as a result of an individual's behaviours. One of our respondents put it eloquently: 'The likelihood that I would participate in any burnout program would depend on whether I perceive the program to be useful. Viewing burnout as something that can be addressed at the individual level is pointless. It does not help to tell someone to exercise or meditate or something when they are being traumatised by an external system that doesn't allow time for those things anyway.' This emphasises our finding that no self-reported coping mechanisms were found to be protective against burnout. In addition, it echoes the current research direction, which has shifted focus toward changes that need to occur at a system or organisational level, such as administrative support, reduced working hours, supportive mentorship and advocacy for member support in medical societies, which hold promise. 4,10,11,14 With the advent of COVID-19, the South African health system has seen an increase in burnout due to increased demand for already strained resources. 42 Fear of contracting the virus, fear of infecting family members, and reduced staffing have also been contributory. 43 Younger doctors, female doctors and trainees remain at higher risk of burnout during the pandemic. [44][45][46] Interestingly, there are conflicting reports of burnout rates in healthcare workers that are in direct contact with COVID-positive patients. Some studies report reduced burnout rates or no difference in burnout rates. 47,48 Those who have lower burnout rates than their colleagues may be benefiting from 'a return of a sense of autonomy', or 'the ability to display altruism', or 'the ability to relate to their patients', argue Hartzband and Groopman. 35 We explored the possible confounding nature of the COVID pandemic on our findings. Consistent with Medscape's National (US) Physician burnout reports from successive years (2019-2021), where orthopaedic surgeon burnout decreased after the onset of COVID (from 38% in 2019 to 34% and 33% in 2020 and 2021 respectively), respondents in our study indicated that their selections on the measured burnout index were typically reduced since the advent of the pandemic. It bears mentioning that these responses are subject to recall bias. However, should this result be an accurate reflection of burnout during this time, we surmise that the majority representation of orthopaedic specialists, private sector practitioners, had a reduced workload due to elective surgery restrictions, and that could have contributed to this finding.
In addition to this bias, there were other limitations. A priori sample size estimates for the regression model we utilised would have necessitated a 27% response rate; however, our response rate was 16%. Although this was comparable to previous studies in orthopaedic cohorts (the majority of which quote rates around 20%, but range from 6-94%), it meant that our study was underpowered to detect significance for several variables. 5 Burnout studies regularly experience poor response rates. It is likely this reflects a lack of time to complete extensive questionnaires with minimal incentive, which we believe to be true in our context. 5 In our questionnaire, we used the Stanford PFI Measure. While it is a validated burnout measurement scale, it limited comparison to only those studies that used this measure or those that reported an overall burnout score. We deemed this acceptable, however, as overall burnout was our outcome of interest. Furthermore, the information sheet and consent form detailed that the survey aimed to evaluate the presence of burnout in orthopaedic surgeons. This could have introduced response bias and exhibited a priming effect on results. However, some of the questions explicitly asked about 'burnout', and as such, the investigators felt this effect would have occurred despite blinding respondents.
Respondents were blinded to the fact that we screened for other associated workplace distress conditions, for which we found high rates in all five conditions. This uncovered a knowledge dearth that warrants further investigation. Other gaps we did not explore in this study include whether burnout rates differ between provinces within South Africa, or across rural and urban boundaries, findings which have been detected in previous burnout investigations both in South Africa and abroad. 6,19,22,49,50 Conclusion Burnout is prevalent in the South African orthopaedic community (72%). Registrars are particularly vulnerable, and further investigation is necessary to identify risk factors particular to this group. Modifiable system-based risk factors, and thus potential targets for initiatives to reduce the burnout rate, included: 'hours at work' and 'lack of free time'. High rates of positive screening for associated workplace distress conditions warrants further investigation.

Additional information
Should you, after reading this article, recognise the symptoms of burnout in yourself and desire support, psychological assistance is available to you by contacting the Healthcare Workers Care Network helpline at 0800 21 21 21. Alternatively, the South African Orthopaedic Association has initiated a Mentorship Programme, available to all members of the SAOA. To get in touch, please email info@saoa.org.za.