Encouragingly, in our study of 1166 health workers from three different cadres across South Africa and Zambia, the majority of health workers reported low or moderate emotional exhaustion and depersonalization and moderate to high personal accomplishment, indicating that few health workers in the treatment arms of our trial met Maslach’s criteria for burnout towards the beginning of the trial.
When testing for risk factors for emotional exhaustion, we found higher levels of emotional exhaustion among more educated health workers, among CHiPs compared to health facility staff, among those with more years providing services, among those who tested negative for HIV at their last HIV test, and among those who witnessed stigmatizing behavior among their colleagues.
Our study had several strengths. We recruited a large and systematic, as opposed to opportunistic, sample of health workers from three cadres within 14 communities across two countries. We used electronic capture devices that allowed participants to fill in responses to questions without the need to disclose their answers to an interviewer. Limitations include that although the MBI has been used in the local setting before, it was originally developed for higher income settings and the evidence of its psychometric properties in South Africa and Zambia is still emerging. As in many studies of health workers, response rates in our study were lower than in surveys in general populations, at 66% overall, but comparable to other surveys measuring HIV stigma among current and future health care workers with response rates in published literature as low as about 40% [30, 31]. Finally, we cannot exclude the possibility of residual confounding for factors that we were unable to measure in our study.
The levels of emotional exhaustion reported by this population of health workers are lower than both the reference values provided in the MBI manual and previous studies using the MBI among similar populations. The highest mean emotional exhaustion among any cadre in any country was 15.4, compared to the mean of 22.19 reported in the manual for the medical occupation subgroup, which is based on an initial sample of 1,104 American physicians and nurses [29]. All means are also lower than those found in a study conducted in 2003 among health workers in South Africa, which found a mean emotional exhaustion score of 24.2 [20]. Similarly, two unrelated studies in 2009 and 2013 conducted among health workers in Malawi found that more than one third of health workers in the former scored high on the emotional exhaustion subscale while 55% of health workers in the latter scored moderate or high [32, 33]. A separate study conducted in South Africa also found that 35% of registered nurses experienced high emotional exhaustion [34]. For comparison, no more than 18% scored high on the emotional exhaustion subscale among any cadre in our study, and no more than 36% scored moderate or high.
There are several reasons why the reported levels of emotional exhaustion may have been lower in this study. One possible explanation is the timing of the study, which was conducted at the beginning of a large and well-known trial. On one hand, the start of the trial may have reduced levels of emotional exhaustion by giving providers new hope and tools to serve their clients living with HIV, including the ability to offer full ART access to all patients regardless of CD4 count in Arm A communities. On the other hand, the launch of the trial may have made health workers reticent to express feelings of burnout for fear of reprisal or, in the case of the CHiPs, loss of their new position. Another possible explanation for lower than anticipated reported levels of emotional exhaustion could be problems with the translation or adaptation of the survey items. Every effort was made to reduce this possibility by completing a thorough back translation process and consultation with indigenous language speakers in both countries. In Zambia, survey translations were available to respondents in a paper format in addition to the English version on the device. In South Africa, Xhosa and Afrikaans versions of the survey were available on the device in addition to the English version. However, the majority of respondents chose to complete the survey in English. A preference to complete the survey in English may have led to differential interpretations of concepts and meaning by second language speakers. However, it is also possible that the use of self-administered surveys on the devices reduced levels of reporting bias. Finally, it is possible that health workers with higher levels of burnout were less likely to complete the survey at all, meaning that selection bias resulted in lower detected emotional exhaustion.
Our results indicate that emotional exhaustion significantly differed according to level of education, health worker type, years of experience providing health services, health worker serostatus and reports of witnessed stigmatizing behaviors.
Results indicating that emotional exhaustion was significantly higher for those who completed higher education were not consistent with Hu et al. [23] but were consistent with Maslach’s findings [21]. However, the lack of significant difference by marital status runs counter to findings reported by Maslach [21]. While married people have previously been shown to experience lower levels of burnout, this effect may have been obscured in our study due to the large cohort of single CHiPs who were newly recruited into the study and had not previously provided HIV services.
CHWs reported lower levels of emotional exhaustion than other health workers. This may be partially due to differences in job attributes [18]. Specifically, CHWs were not trained as part of the PopART intervention and were therefore not beholden to meeting specific care targets related to the study implementation. Instead, they worked in less structured roles that allowed for some control over their work schedule, particularly in Zambia, which could have shielded CHWs from some of the targets or pressures to serve clients that other cadres faced. Lower emotional exhaustion among CHWs could also be reflective of their long-term presence in these communities, rooted in ideas around community responsibility, duty, and care, which may provide some protection from exhaustion.
Our results suggest a positive association between years providing HIV services and emotional exhaustion among health workers in Zambia and South Africa. A previous systematic review found inconsistent results for the association between years providing HIV services and emotional exhaustion, with different studies showing both positive and negative associations [18]. Our results support the theory that the emotional care burden builds up among healthcare workers providing HIV services over time.
In this analysis, health workers who self-reported living with HIV had significantly lower emotional exhaustion. Study staff familiar with health workers living with HIV in these settings shared that they had observed them taking particular care to ensure that their workload was manageable, which may have a protective effect against developing emotional exhaustion [35]. This suggests that future studies wishing to explore the relationship between HIV status and burnout among health workers should statistically adjust for their workload. Additionally, as discussed previously in relation to the CHWs, the motivation for employment in this line of work may also be protective for health care workers living with HIV, in that they have self-selected because of a deep sense of purpose and self-satisfaction derived from improving the health and wellbeing of others living with HIV, which may serve as a buffer against emotional exhaustion. Finally, because HIV status was self-reported, it is possible that health workers experiencing higher levels of internalized stigma and emotional exhaustion may have chosen not to disclose their HIV status.
The association between emotional exhaustion and witnessing a colleague stigmatize clients living with HIV indicates a relationship between observed stigma towards clients living with HIV in the health facility and the psychosocial wellbeing of health workers providing care, even though the health workers were not the target of the stigmatizing behaviors. This is corroborated by a study of South African nurses that examined various predictors of burnout using the MBI and observed a significant association between stigma and depersonalization [36].
This finding highlights the pernicious effect of stigma on the mental health and wellbeing of health workers [37], regardless of their HIV status, and demonstrates that stigma-reduction efforts in health care settings are crucial not only for the health and wellbeing of potential clients but for the staff as well [38].