Impact of reliable light and electricity on job satisfaction among maternity health workers in Uganda: A cluster randomized trial

Background: Maintaining a motivated health workforce is critical to health system effectiveness and quality of care. Scant evidence exists on whether interventions aimed to strengthen health infrastructure in low-resource settings affect health workers. This study evaluated the impact of an intervention providing solar light and electricity to rural maternity facilities in Uganda on health workers’ job satisfaction. Methods: We used a mixed-methods design embedded in a cluster randomized trial to evaluate whether and how the We Care Solar Suitcase intervention, a solar electric system providing lighting and power, affected health workers in rural Ugandan maternity facilities with unreliable light. Facilities were randomly assigned to receive the intervention or not without blinding in a cluster-randomized controlled trial. Outcomes were assessed through two rounds of surveys with health workers. We used regression analyses to examine the intervention’s impact on job satisfaction. We used an inductive approach to analyze qualitative data to understand the study context and interpret quantitative findings. Results: We interviewed 85 health workers across 30 facilities, the majority of whom were midwives or nurses. Qualitative reports indicated that unreliable light made it difficult to provide care, worsened facility conditions, and harmed health workers and patients. Before the intervention, only 4% of health workers were satisfied with their access to light and electricity. After the installation, satisfaction with light increased by 76 percentage points [95% confidence interval (CI): 61–92 percentage points], although satisfaction with electricity did not change. Experience of negative impacts of lack of overhead light also significantly decreased and the intervention modestly increased job satisfaction. Qualitative evidence illustrated how the intervention may have strengthened health workers’ sense of job security and confidence in providing high-quality care while pointing towards implementation challenges and other barriers health workers faced. Conclusions: Reliable access to light and electricity directly affects health workers’ ability to provide maternal and neonatal care and modestly improves job satisfaction. Policy makers should invest in health infrastructure as part of multifaceted policy strategies to strengthen human resources for health and to improve maternal and newborn health services. Trial registration socialscienceregistry.org: AEARCTR-0003078. Registered June 12, 2018, https://www.socialscienceregistry.org/trials/3078 Additionally registered on: ClinicalTrials.gov: NCT03589625, Registered July 18, 2018, https://clinicaltrials.gov/ct2/show/NCT03589625)

The study was conducted in maternity facilities in Central, Eastern, and Western regions of Uganda [25]. Uganda has a population of 44.3 million in 2019, with a total fertility rate of 4.8 live births per woman [26]. In 2016, about 74.2% of deliveries were assisted by a skilled birth attendant and 73.4% of deliveries occurred in a health facility [27]. Uganda’s health system is divided into public and private sectors. The public sector consists of national and regional hospitals and a tiered district health system composed of health centers at four levels. Most lower-level health centers are not connected to the central electricity grid and power shortages are frequent [28, 29]. A 2015 survey suggested that kerosene lamps supplied 42% of lighting needs in Health Center IIs [29]. In our study, 40% of health facilities did not have electricity or relied on lanterns as the primary source of light at the time of the baseline survey [30]. The intervention, a “Solar Suitcase” manufactured by the non-government organization We Care Solar, is a complete solar electric system that contains high efficiency movable LED lights for medical use, rechargeable headlamps, USB ports for charging cell phones and small medical devices, and a portable fetal heart rate Doppler [30]. One Solar Suitcase was installed in each facility, with 2–4 overhead LED lights for each delivery room, depending on its size. A full description of the Solar Suitcase and its implementation, including training of health workers, maintenance, and costs have been previously published [25, 30]. We used an embedded mixed-methods design to evaluate whether and how the intervention affected health workers’ job satisfaction and their experiences of working at night [31, 32]. Specifically, we embedded qualitative data collection within an experimental design before and after the intervention implementation to understand the study context and interpret quantitative results. The experimental study was a stepped-wedge cluster-randomized controlled trial. The trial was conducted between June 2018 and April 2019. Level II, III, and IV health centers that lacked access to a reliable, bright light source in the maternity ward were eligible for inclusion in the study. Facilities were randomized into one of two groups of 15 facilities to either receive the intervention in the first or second sequence. The focus of this analysis used data collected from the baseline (when no facility received the intervention) and the first follow up survey (6 weeks after the first 15 facilities had received the intervention) (Fig. 1). Within facilities, we interviewed all consenting health workers who were involved in labor and delivery. The interview consisted of both quantitative survey questions and qualitative open-ended questions. The purpose of the open-ended questions was to elicit narratives of the study context and intervention impact to facilitate interpretation of quantitative findings. Enumerators followed an interview guide to ask these open-ended questions and recorded health workers’ responses as field notes. Details of the trial design, including sampling methods, randomization, and sample size, have been previously published [25, 30]. Study flow chart A total of 15 female and 5 male enumerators conducted the health worker surveys. These enumerators had at least a certificate in Comprehensive Nursing or Midwifery and completed a 2-week training on research protocol, data collection methods, and human subjects research. With permission from the district and facility leaders, enumerators interviewed health workers at a private space using a questionnaire with both closed and open-ended questions. The research team comprised two researchers with PhDs in health policy and mixed-methods training (SR and WC), a researcher with a PhD in economics (JC), a researcher with a PhD in medicine (PW), and a researcher with a post-graduate diploma in monitoring and evaluation (BM). Three team members were female (JC, SR, and WC) and two were male (BM & PW). Before the trial began, we calculated the minimum detectable effect size for health workers’ satisfaction with light and electricity to be 0.62, assuming a total of 3 health workers per facility across 30 facilities, a mean health worker satisfaction score of 2 (out of a 1–5 range), and an intra-cluster correlation (ICC) of 0.3. In practice, our observed sample size was about 2 health workers per facility, the baseline satisfaction score was 3.1, and the ICC was 0.2. Using these parameters, our ex post minimum detectable effect size for health worker satisfaction was 0.36. The pre-registered primary outcomes included health workers’ overall job satisfaction as well as their satisfaction with light and electricity. Job satisfaction was measured via an index and calculated as the mean of health workers’ responses to four statements about their motivation to work, how satisfied they are with their job, the morale level at their department, and their plans to stay at the same job, with higher scores indicating greater job satisfaction. Satisfaction with light and electricity is a binary variable equal to one if a health worker “agreed” or “strongly agreed” to both statements: “I am satisfied with the availability and brightness of light in this facility” and “I am satisfied with the availability of electricity in this facility.” Detailed definitions of quantitative outcomes are provided in Appendix Table ​Table44. Outcome measurements and definitions For pre-registered secondary outcomes, we constructed an index to measure health worker’s experiences of the impact of lack of overhead light during nighttime deliveries in the past month. The index was calculated as the mean of 14 items that measured how often health workers conducted deliveries at night without overhead light; had to hold a torch (i.e., flashlight) in hand to see a patient; experienced lack of light that affected normal care provided; delayed care; feared to move around the facility; and were affected in ability to suture, find/use equipment, conduct examinations of the mother, provide emergency care, provide newborn care, monitor fetal heartrate, administer medication, clean up after delivery, and manage infection control. Responses were scored on a 1–5 Likert scale from “Never” to “Every nighttime delivery”, so that higher scores on the index indicated more frequent occurrence of negative impacts of lack of overhead light. In addition to the pre-registered outcomes, we examined each of the indicators that constituted the 4-item job satisfaction index individually. We further examined outcomes that measured concepts closely related to health workers’ job satisfaction, such as job security, self-confidence, adequate support in terms of supplies and equipment, and workload. Responses to these measures were assessed on a 1–5 Likert scale to indicate levels of agreement, with higher scores indicating greater satisfaction. As part of the open-ended questions, health workers were asked to comment on the impacts of lack of overhead light on their job, difficulties with using kerosene lanterns or candles, memorable situations where lack of lighting affected patient care, comparison between lack of lighting and other challenges at facilities, and their feelings about working without reliable light. After receiving the intervention, health workers were asked to comment on whether the intervention helped them to care for patients and what kind of challenges they encountered in using the intervention. First, we evaluated the effects of the intervention on health worker outcomes by analyzing quantitative data collected from health worker surveys. To supplement the quantitative findings, we analyzed qualitative data collected from the open-ended questions. Quantitative analysis We used linear probability models that included facility fixed effects to estimate the impact of the intervention on primary and secondary outcomes at the level of health workers. Standard errors in all models were clustered at the facility level. To assess the robustness of the models, we used alternative model specifications including non-linear models (logistic for binary outcomes and Poisson for count outcomes), facility random effects, inclusion of health worker control variables in regression models, and adjustment of standard errors using the wild cluster bootstrap method given the small number of facilities [33]. In addition, we assessed whether the results were driven by compositional changes in health workers after installation of the intervention by examining heath worker retention rates as a robustness check. Data were analyzed with Stata version 15.1 [34]. Qualitative analyses We used an inductive analytical approach to analyze the qualitative data [32]. One member of the research team began the qualitative analysis by reading the field notes in full until reaching a high level of familiarity with the content of the text. She organized enumerators’ field notes based on the open-ended questions, created initial codes by open coding the field notes, and coded the text to generate themes. A second member of the research team read the coded text and provided feedback. The two research team members iteratively analyzed the codes, coded the field notes, and finalized categories and themes. Throughout the process, we considered how our training, identity, and world view influenced our interpretation of the qualitative findings and consulted other research team members to draw on interdisciplinary insights. We did not present the qualitative results to health worker participants for comments, but we shared the findings with our in-country team who agreed that the findings reflected the reality in the clinical setting. Qualitative data were analyzed with Taguette version 0.10.1, a web-based text management and analysis software [35].

N/A

• chang, wei
• cohen, jessica
• mwesigwa, brian
• waiswa, peter
• rokicki, slawa