We designed a two-part study to describe occupational burnout and utilization of HIV services among providers in the primary care centres of the Lusaka, Zambia, public health sector, where services for HIV care and treatment have rapidly expanded since 2004 [3
]. We recruited physicians, clinical officers (the equivalent of physician assistants in the United States and Europe), nurses, midwives and pharmacy staff employed at government clinics. Thirteen sites were chosen, all providing long-term HIV care and treatment. At each facility, other primary care services are provided, including general outpatient care, antenatal services, child health services and tuberculosis treatment. Characteristics of each facility and catchment size are shown in Table .
District health facilities in Lusaka, Zambia, that served as sites for this study, March 2007 to July 2007
This study included qualitative and quantitative study components. Our qualitative methods consisted of six focus group discussions and four key-informant interviews. One group session was held for each of the following: midwives, physicians, clinical officers and pharmacy staff. Two groups of nurses were convened, as they comprise the vast majority of health care workers. Participants were stratified according to clinic of employment and professional cadre, and then randomly selected. Overall, 13 invitations were sent out for each group (i.e. one representative per clinic).
Discussions were held in a private conference room at a local nongovernmental organization and each lasted approximately 90 minutes. Trained study staff served as facilitators; written notes and tape recordings were transcribed.
Key informants were identified by physicians based at the study clinics and were recruited to participate in one-on-one interviews. The four interviews included two providers living with HIV, one HIV-negative and one with unknown HIV status. Each lasted approximately 60 minutes.
Data were analyzed by manually compiling common themes and matrices and discussing these data between investigators, study staff and representatives from the Lusaka District Health Management Team.
In parallel, we surveyed active health care providers across the 13 clinics to quantify their perceptions of occupational burnout and HIV service utilization. All staff reporting to work over a three-week window were asked by their supervising nurse managers to complete a 36-question survey. Each questionnaire had a statement of consent attached; completion of this consent was necessary for inclusion in the analysis. Drop boxes were provided so that participants could return their questionnaires anonymously.
In the survey, prevalence of occupational burnout was based on a single question that has been validated against a full occupational burnout scale [22
]. Respondents were asked to quantify their level of burnout from a five-item scale: (1) "I have no symptoms of burnout"; (2) "Sometimes I am under stress, but I don't feel burned out"; (3) "I am definitely burned out and have occasional symptoms of burnout"; (4) "The symptoms of burnout I'm experiencing won't go away"; and (5) "I feel completely burned out and I am at the point where I need to make some changes or seek some sort of help". If respondents selected (3), (4), or (5) from the scale, they were categorized as having occupational burnout. We also asked numerous supporting questions to better understand types of burnout in this population.
To determine utilization of HIV services, we relied primarily on use of HIV testing services over the past 12 months. Information regarding demographic characteristics, employment history and HIV knowledge and perceptions was also collected.
In our statistical analysis, we calculated unadjusted and adjusted relative risks (RR) with 95% confidence intervals (95% CI) to identify predictors of occupational burnout and HIV service utilization [23
]. We adjusted for individual and clinic-level variation in hierarchical logistic models by means of the SAS GLIMMIX Procedure [24
]. Covariates included demographics and other independent variables associated with the outcome at p < 0.10 in unadjusted models. All analyses were performed with SAS version 9.1.3 for Windows (Cary, North Carolina, United States of America).
All participants provided written informed consent to participate. This study was approved by the University of Zambia Research Ethics Committee and the University of Alabama at Birmingham Institutional Review Board.