Job satisfaction and turnover intentions among health care staff providing services for prevention of mother-to-child transmission of HIV in Dar es Salaam, Tanzania

Background: Option B+ for the prevention of mother-to-child transmission (PMTCT) of HIV (i.e., lifelong antiretroviral treatment for all pregnant and breastfeeding mothers living with HIV) was initiated in Tanzania in 2013. While there is evidence that this policy has benefits for the health of the mother and the child, Option B+ may also increase the workload for health care providers in resource-constrained settings, possibly leading to job dissatisfaction and unwanted workforce turnover. Methods: From March to April 2014, a questionnaire asking about job satisfaction and turnover intentions was administered to all nurses at 36 public-sector health facilities offering antenatal and PMTCT services in Dar es Salaam, Tanzania. Multivariable logistic regression models were used to identify factors associated with job dissatisfaction and intention to quit one’s job. Results: Slightly over half (54%, 114/213) of the providers were dissatisfied with their current job, and 35% (74/213) intended to leave their job. Most of the providers were dissatisfied with low salaries and high workload, but satisfied with workplace harmony and being able to follow their moral values. The odds of reporting to be globally dissatisfied with one’s job were high if the provider was dissatisfied with salary (adjusted odds ratio (aOR) 5.6, 95% CI 1.2-26.8), availability of protective gear (aOR 4.0, 95% CI 1.5-10.6), job description (aOR 4.3, 95% CI 1.2-14.7), and working hours (aOR 3.2, 95% CI 1.3-7.6). Perceiving clients to prefer PMTCT Option B+ reduced job dissatisfaction (aOR 0.2, 95% CI 0.1-0.8). The following factors were associated with providers’ intention to leave their current job: job stability dissatisfaction (aOR 3.7, 95% CI 1.3-10.5), not being recognized by one’s superior (aOR 3.6, 95% CI 1.7-7.6), and poor feedback on the overall unit performance (aOR 2.7, 95% CI 1.3-5.8). Conclusion: Job dissatisfaction and turnover intentions are comparatively high among nurses in Dar es Salaam’s public-sector maternal care facilities. Providing reasonable salaries and working hours, clearer job descriptions, appropriate safety measures, job stability, and improved supervision and feedback will be key to retaining satisfied PMTCT providers and thus to sustain successful implementation of Option B+ in Tanzania. A facility-based, quantitative, cross-sectional study was conducted in 36 public health facilities in Dar es Salaam, Tanzania, between March and April 2014. The health facilities were sampled from all public-sector facilities providing PMTCT services in two (Ilala and Kinondoni) of the three districts in Dar es Salaam, where about 70% of the city’s population reside [29]. The two districts represent the smallest (Ilala) and the largest (Kinondoni) of the three districts of Dar es Salaam both in terms of population and the number of health facilities [30]. The reason for choosing public rather than private clinics is that the majority of the population attends public facilities and most of the workforce shortages are reported from public facilities [11]. During the study period, Tanzania was phasing out Option A (antenatal and intra-partum ART prophylaxis and nevirapine for HIV-exposed infants) and adopting Option B+ following the 2013 WHO recommendations [3]. PMTCT services in Tanzania are routinely provided free of charge and in accordance with the national guidelines that are regularly revised to match the WHO guideline updates [3, 13]. Out of 150 health facilities in these two districts (49 in Ilala and 101 in Kinondoni) providing the integrated antenatal care (ANC) and PMTCT services, 92 non-public facilities were excluded. Ten public facilities were further excluded due to lack of permission from authorities to access them. Out of the remaining public facilities, all 18 facilities from Ilala and another 18 matching (on facility level and size in terms of patient visits) facilities from Kinondoni were selected. A self-administered questionnaire was distributed to all providers who were present in the PMTCT clinic on the day of the interview. About 2% of the providers were excluded because they did not give consent to participate. Finally, the questionnaires were distributed to 250 health care providers who predominantly provide PMTCT services. These included nurse officers and nurse midwives with a college diploma (trained for at least 3 years), as well as public health nurses and nurse assistants with a college certificate (trained for at least 2 years). We used a structured questionnaire created and validated in terms of content in Mali and Senegal to ensure its suitability for the sub-Saharan African context [31]. The questionnaire had been developed in four stages, where the first included the testing of four dimensions for measuring job satisfaction based on previous work, the Measure of Job Satisfaction [32] and the Job Descriptive Index [33]. The second stage involved a comprehensive literature review and expert consultations leading to the addition of two more dimensions. The third stage involved testing of the new questionnaire in Senegal and Mali in 2008 and 2010, respectively; after that, the questionnaire was modified by adding three more dimensions. The final stage involved the testing and validation of the final questionnaire using a combined psychometric analysis and expert consultations [31]. The questions were organized into the following categories: socio-demographic characteristics, such as age and gender; professional qualifications; overall job satisfaction; and satisfaction with different aspects of the job. The overall job satisfaction was assessed using one specific question, and the satisfaction with different aspects of the job was assessed using nine domains with a total of 42 questions (minimum 3 and maximum 5 questions per domain). The satisfaction questions (Additional file 1) were Likert-type, based on a 5-point scale coded on a scale from 1 (strongly disagree) to 5 (strongly agree). The questions were formulated to ask if the providers agreed that they were satisfied with various aspects of their job, such as remuneration, equipment and work context, workload, duties, harmony in the workplace, on-the-job training, management, moral satisfaction and job stability. We consulted and obtained permissions to use the scale from the corresponding author Fournier Pierre [31]. In collaboration with language experts, the scale was translated from French to English and finally to Kiswahili (local language). Questions on health care providers’ intentions to leave their current job and their perceptions on various PMTCT options were developed by the research team, e.g. “How frequently do you think about leaving your current job?”, and the responses were marked on a 7-item Likert scale ranging from 1 (never) to 7 (very often). We also asked the providers about which PMTCT Option (A, B or B+) they themselves thought was the best for the mother and her baby, versus which option they thought would be preferred by the patients versus which option they thought would be most feasible for implementation in the Tanzanian public health system. To ensure quality and reliability, the data collection tool was piloted at different health facility levels (hospital, health centre and dispensary level), before initiating the study, and the inputs were used to modify and improve the questionnaire. Twenty Tanzanian research assistants with a medical education background were recruited and trained to understand and use the research protocol and data collection tools. After obtaining a written consent, the self-administered questionnaire was filled in by the providers during work breaks or after a shift was completed. Providers were informed that participation was voluntary and that they could withdraw from the study at any time. Completing the questionnaire took approximately 45 min, and participants were reimbursed with a small amount equivalent to a 1-h salary, 10 000 Tanzanian shillings ($5), after completing the survey. To ensure confidentiality, providers were informed to not write their names on the questionnaire to avoid any identifying information. We obtained permission to conduct the survey from both the district and health facility authorities in charge. The Research and Ethics Committee of Muhimbili University of Health and Allied Sciences (MUHAS) gave ethical clearance for the study. Categorical variables were summarized as proportions, and Fisher’s exact test was used to test for association between various factors and job satisfaction. We calculated the Cronbach alpha for each domain, and the coefficients showed that except for the job stability domain where a coefficient 0.69 suggested fair reliability, the domain scores included in the questionnaire (0.77 to 0.92) were very reliable. For the purpose of this analysis, we converted responses from a 5- or a 7-item Likert scale to binary outcomes. The primary dependent variable for this analysis was “overall satisfaction with the current job”, with five different response alternatives on a 5-item Likert scale. Health care providers who reported to be “neutral”, “dissatisfied” and “very dissatisfied” were categorized as “dissatisfied”, while those who reported to be “very satisfied” or “satisfied” were categorized as “satisfied”. The secondary dependent variable was “intention to leave the current job”, with seven possible Likert-scale responses: “very often”, “fairly often”, “often”, “occasionally”, “rarely” and “almost never” were considered as having an intention to leave, while those who reported “never” were considered to have no intention to leave their current job. The explanatory variable covered job satisfaction in nine domains, converting 5-item Likert-scale replies into binary outcomes dividing the responses into “dissatisfied” (including “neutral”/“dissatisfied” and “very dissatisfied”) versus “satisfied” (“very satisfied” and “satisfied”). Other independent variables included provider baseline characteristics such as age, gender, education level, professional cadre and the time in current position. We also assessed the association between the health care provider’s perception towards various PMTCT options and their job satisfaction or intention to leave their current job, which we thought was of special interest during the policy transition period from Options A and B to Option B+ in Tanzania. The statements in each domain were used to create separate logistic regression models for satisfaction with the current job and for the intention to leave the current job. We selected one variable with the lowest p value, provided that the p value was less than 0.2, from each model. For both main outcomes (satisfaction with current job or intention to leave the current job), the independent variables that were significantly associated with the dependent variable were used to create the final multiple logistic regression models using a backward elimination method. Lastly, the relationship between respondents’ satisfaction with their current job and their intention to leave this job was assessed in a separate model. All models were adjusted for cluster effect at the facility level taking into consideration that several health care providers were interviewed at the same facility. We also examined possible multiplicative interactions between predictors in the final multiple logistic regression models by adding their product terms. A formal p value for departure from interaction was obtained using a likelihood ratio test comparing the maximized log-likelihoods of the model with and without the product terms. The results are presented as odds ratios (ORs) and their 95% confidence intervals. The data analysis was done using Stata, release 14 (StataCorp LP; College Station, TX).