In the nursing profession, in general, it has been noted that workload is a major workplace stressor (Glazer & Gyurak 2008). This is especially true in South Africa, where HIV and AIDS places a greater burden on carers and, thus, an increased workload among hospital nurses (Coetzee et al. 2013; De Wet & Du Plooy 2012).

With over 5 million persons living with HIV, the sheer number seeking treatment creates overcrowded conditions in clinics (Coetzee, Kagee & Vermeulen 2011), resulting in an inordinate workload for nurses. Specific additional duties to caring for patients receiving antiretroviral therapy (ART), such as providing adherence counselling, assisting patients in obtaining social grants and advising on ways to prevent medication side effects, add substantially to nurses’ existing workload (De Wet & Du Plooy 2012).

It has been reported that in some contexts, nurses come into conflict with other players in the clinic or hospital, such as their managers, other nurses, doctors, patients and patients’ relatives (Guidroz, Wang & Perez 2012). Numerous studies around the world have shown robust associations between interpersonal conflict and burnout in various contexts (García-Izquierdo & Ríos-Rísquez 2012; Guidroz et al. 2012; Kitaokai & Masuda 2012).

A lack of adequate resources including lack of supplies, lack of equipment, broken equipment and unavailable resources has been shown to be among the top workplace stressors in several countries (Glazer & Gyurak 2008). In South Africa, poor workplace infrastructure and the unavailability of medical supplies were key reasons health workers either left the public health sector or emigrated from South Africa (Pendleton, Crush & Lefko-Everett 2007). Organisational constraints, workload and interpersonal conflict at work have been robustly associated with emotional exhaustion, a key aspect of burnout (Engelbrecht et al. 2008).

Engaging with the emotional issues around death or dying is a strong predictor of burnout (Garrosa et al. 2008). Nearly 90% of South African nurses caring for persons living with HIV empathised strongly with their patients and found it difficult to see patients experience suffering and ultimately death (Davhana-Maselesele & Igumbor 2008).

HIV-related stigma has been defined as ‘a “process of devaluation” of people either living with or associated with HIV and AIDS’ (UNAIDS 2007:9). Secondary HIV and AIDS stigma experienced as a result of being associated with people living with HIV (PLWH) has been identified as an important occupational stress factor among nurses (Haber, Roby & High-George 2011). HIV caregivers, including nurses, have been ostracised by community members and even by friends and family (Van Dyk 2007). For example, among nurses in five African countries, including South Africa, perceived HIV and AIDS stigma was shown to predict job dissatisfaction (Chirwa et al. 2009).

Demographic factors such as age and job status have been found to influence the intensity of burnout among nurses (Garrosa et al. 2008). For example, among a sample of Spanish nurses, age and job status accounted for 16%, 14% and 12% of the variance in emotional exhaustion, depersonalisation and personal accomplishment, respectively (Garrosa et al. 2008). Younger nurses in the study experienced greater levels of burnout than older nurses.

In South Africa, high levels of emotional exhaustion were found among nurses who worked both in public and private hospitals (Coetzee et al. 2013), with 45.8% scoring in the elevated range (greater than 27) on the Maslach Burnout Inventory (MBI). Further, 53.8% of the participants in Coetzee’s study who worked in public hospitals scored in the high range for emotional exhaustion compared with 40.6% who worked in private hospitals. Similarly, in a study of 818 South African nurses, 34.6% and 20.4% of participants scored in the elevated range on emotional exhaustion and depersonalisation, respectively (Van der Colff & Rothmann 2012). High levels of burnout were also found in a study of over 500 nurses from the Free State province in South Africa, with reported mean scores of 31.34 for emotional exhaustion and 17.80 for depersonalisation (Engelbrecht et al. 2008). Additionally, 68.7% and 85.1% of participants reported elevated levels of emotional exhaustion and depersonalisation, respectively. Among 935 critical care nurses in private and public South African hospitals, the mean score of 27.04 on the MBI fell in the elevated range, with 59.4% scoring in the moderate range for emotional exhaustion, and 16.4% scoring in the high range (Klopper et al. 2012). Further, 68.5% and 3% of the sample reported moderate and high levels of depersonalisation, respectively (Bruce & Sangweni 2012). The research indicates that nurses experience a high rate of burnout. Furthermore, a number of factors have been associated with burnout in previous research: that is, age, job status, workload, interpersonal conflict at work, organisational constraints, death and dying-related stress, and HIV and AIDS stigma by association. The aim of this study was to test the ability of these factors to predict burnout.

In the context of the challenging terrain of HIV treatment, we investigated the predictors of three dimensions of burnout, namely emotional exhaustion, depersonalisation and personal accomplishment among a sample of South African nurses caring for patients living with HIV. We tested the predictors (i.e. age, job status, workload, interpersonal conflict at work, organisational constraints, death and dying-related stress and HIV and AIDS stigma by association) on each of these dimensions of burnout. The objectives of the study were as follows:

to determine the extent to which age, job status, workload, interpersonal conflict at work, organisational constraints, death and dying-related stress and HIV and AIDS stigma by association were predictors of emotional exhaustion

Age, gender, race, marital status, living situation, education level, job status, time in the nursing profession, work details, income level, place of birth and languages spoken were obtained using a demographic questionnaire.

The 22-item MBI was used to assess three components of burnout, namely emotional exhaustion, depersonalisation and personal accomplishment (Maslach, Jackson & Leiter 1996). Respondents were asked to indicate the frequency of their experiences, specifically their feelings and attitudes, on a 7-point Likert scale with response options ranging from 0 (never) to 6 (every day). High scores on the emotional exhaustion and depersonalisation dimensions correspond to higher levels of burnout and high scores on the personal accomplishment dimension refer to lower levels of burnout. Maslach et al. (1996) have reported high internal consistency for the three subscales with the alpha reliabilities ranging from 0.71 to 0.90. The MBI has also been used in other South African studies to examine burnout among nurses, with high alpha coefficients for the various subscales (Coetzee et al. 2013; Engelbrecht et al. 2008). For this study, the internal consistency of the MBI as measured by Cronbach’s alpha was 0.85. Cronbach’s alphas for emotional exhaustion, depersonalisation and personal accomplishment subscales were 0.85, 0.60 and 0.67 respectively.

The Quantitative Workload Inventory (QWI) was used to measure the perceived amount of work (Spector & Jex 1998). The QWI has five items which ask about the amount and pace of work. High perceived workload is represented by high scores. High alpha coefficients have been consistently reported for this scale. For this study, the internal consistency of the QWI as measured by Cronbach’s alpha was 0.80.

The four-item Interpersonal Conflict at Work Scale (ICAWS) measured participants’ perceived level of interpersonal conflict at work, which ranges from insignificant personal differences to physical assaults and which can be either overtly or covertly expressed (Spector & Jex 1998). The ICAWS has demonstrated high internal consistency in several studies (Kath et al. 2013). Cronbach’s alpha for the ICAWS in this study was 0.78.

The 11-item Organisational Constraints Scale (OCS) measured the extent to which participants experienced organisational constraints at work. Examples include situations that interfere with job performance such as ‘Inadequate training’, ‘Poor equipment or supplies’, and ‘Inadequate help from others’ (Spector & Jex 1998). The instrument asks respondents to indicate how often the item makes job performance difficult or impossible (Spector & Jex 1998). The range of response options is from 1 (less than once per month or never) to 5 (several times per day). High scores correspond to more organisational constraints. High alpha coefficients have been reported in various studies (Kath et al. 2013). The Cronbach’s alpha for this study was 0.89.

Death and dying-related stress were measured by the Death and Dying subscale of the Nursing Stress Scale (NSS). The NSS has demonstrated high internal consistency in various studies (Rosnawati et al. 2010). For this study, the internal consistency of the NSS as measured by Cronbach’s alpha was 0.86.

The 19-item HIV and AIDS Stigma Instrument – Nurse (HASI-N) was used to measure nurses’ perpetration and experiences related to HIV-related stigma (Uys et al. 2009). We used the nine items of the Nurses Being Stigmatised factor to measure HIV and AIDS stigma by association as experienced by the nurses. The scale has shown high internal consistency (Kohi et al. 2010). For this study, the internal consistency of the HASI-N as measured by Cronbach’s alpha was 0.91.

The regression model explained nearly 40% of the variance in emotional exhaustion (R² = 0.39, F(7, 102) = 9.28, p = 0.00), a medium effect size in the social sciences. Workload accounted for 24% of the variance (ΔF(1, 108) = 34.45, p = 0.001). Job status and interpersonal conflict at work were also significant predictors (p = 0.009) of emotional exhaustion, explaining 4% and 7% of the variance respectively. Organisational constraints was the final significant predictor, accounting for 3% of the variance in emotional exhaustion. Three of the four predictor variables contributed significantly to the variance in emotional exhaustion. These were interpersonal conflict at work (β = 0.20, p = 0.024), job status (β = 0.23, p = 0.009) and workload (β = 0.30, p = 0.001).

The combination of predictor variables explained one-third of the variance in depersonalisation (R² = 0.33, F(7, 102) = 7.22, p = 0.001). The combination of interpersonal conflict and workload explained 14% and 7% in the variance respectively. However, only interpersonal conflict at work (β = 0.27, p = 0.004) was a significant predictor of depersonalisation. Additionally, organisational constraints (β = 0.21, p = 0.034), HIV/AIDS stigma by association (β = 0.26, p = 0.006) and job status (β = 0.21, p = 0.021) significantly predicted the variance in depersonalisation.

The regression model significantly explained nearly one-fifth of the variance in personal accomplishment (R² = 0.18, F(7, 102) = 3.12, p = 0.001). Job status and organisational constraints accounted for 9% and 4% of the variance. HIV and AIDS stigma by association (β = −0.21, p = 0.046) and job status (β = −0.32, p = 0.001) significantly predicted personal accomplishment.