Who are CHWs? An ethnographic study of the multiple identities of community health workers in three rural Districts in Tanzania

Background: Numerous studies have examined the role of community health workers (CHWs) in improving the delivery of health services and accelerating progress towards national and international development goals. A limited but growing body of studies have also explored the interactions between CHWs’ personal, communal and professional identities and the implications of these for their profession. CHWs possess multiple, overlapping roles and identities, which makes them effective primary health care providers when properly supported with adequate resources, but it also limits their ability to implement interventions that only target certain members of their community, follow standard business working days and hours. In some situations, it even prevents them from performing certain duties when it comes to sensitive topics such as family planning. Methods: To understand the multiple identities of CHWs, a mixture of qualitative and ethnographic methods was utilized, such as participant observation, open-ended and semi-structured interviews, and focus group discussions with CHWs, their supervisors, and their clients. The observation period began in October 2013 and ended in June 2014. This study was based on implementation research conducted by the Connect Project in Rufiji, Ulanga and Kilombero Districts in Tanzania and aimed to understand the role of CHWs in the provision of maternal and child health services in rural areas. Results: To our knowledge, this was the first study that employed an ethnographic approach to examine the relationship between personal, communal and professional identities, and its implications for CHWs’ work in Tanzania. Our findings suggest that it is difficult to distinguish between personal and professional identities among CHWs in rural areas. Important aspects of CHW services such as personalization, access, and equity of health services were influenced by CHWs’ position as local agents. However, the study also found that their personal identity sometimes inhibited CHWs in speaking about issues related to family planning and sexual health. Being local, CHWs were viewed according to the social norms of the area that consider the gender and age of each worker, which tended to constrain their work in family planning and other areas. Furthermore, the communities welcomed and valued CHWs when they had curative medicines; however, when medical stocks were delayed, the community viewed the CHWs with suspicion and disinterest. Community members who received curative services from CHWs also tended to become more receptive to their preventative health care work. Conclusion: Although CHWs’ multiple roles constrained certain aspects of their work in line with prevalent social norms, overall, the multiple roles they fulfilled had a positive effect by keeping CHWs embedded in their community and earned them trust from community members, which enhanced their ability to provide personalized, equitable and relevant services. However, CHWs needed a support system that included functional supply chains, supervision, and community support to help them retain their role as health care providers and enabled them to provide curative, preventative, and referral services. The data for this article comes from research conducted during two related projects in Kilombero district, Morogoro region. The main project was called Connect, a research intervention study designed to test the impact of using a paid cadre of CHWs that provided integrated maternal, newborn and child health service [23]. CHWs also provided family planning services such as distributing condoms, refilling oral contraceptives, and providing education and referrals (for other family planning methods) in households. The second sub-project was known as Connect Family Planning, which began to operate in 2013. It aimed to contextualize the findings in the first project, which had shown that CHWs had a non-significant effect on contraceptive utilization after two years of their introduction [32]. During the implementation of both studies from 2010 to 2013, CHWs retention was 98% (Kante, Almamy. Personal communication. Aug.21, 2019). The study population comes from rural and peri-urban areas. The residents of Kilombero District are mostly engaged in subsistence farming, cultivating crops like rice, maize, and cassava. Kilombero District is a religiously and ethnically heterogeneous area, populated by both Muslims and Christians [33]. Common ethnic groups include farming tribes such as the Wapogoro, Ndamba, Kaguru, Wangoni, Wahehe as well as recent migrants such as the Sukuma, who are both farmers and pastoralists. At times, conflict would arise between farmers and pastoralists. During the study period, such a conflict led to the death of a police officer, siege of a police station, and having helicopters and other national reinforcements brought to the area [34]. The data informing this article comes from two sources: (i) qualitative data in the form of interviews (IDIs) and focus discussions (FGDs) and (ii) ethnographic data in the form of observations and participation. The qualitative research was part of a larger study, which was registered through the International Standard Randomized Trial register with an award number ISRCTN96819844. The IDIs and FGDs were collected in two phases during March 2012 and during July 2013 from eight villages out of 50 intervention villages in Rufiji, Ulanga, and Kilombero Districts. Qualitative data collected during March 2012 came to be known as Qualitative Appraisal System 1 (QSA 1), while data collected during July 2013 came to be known as Qualitative Appraisal System 2 (QSA 2). Fewer interviews were conducted during QSA II because the aim was to track any changes in the specific themes rather than producing an exhaustive list of themes. The criteria for selecting the villages for QSA 1 and 2 factored in the size of the villages, numbers of WAJAs deployed, and information about health coverage and access. The aim of the data collection was to gain impressions from different stakeholders and perspectives involved in both the provision and receiving of health services. Researchers found saturation in the targeted themes: improvement of maternal and child health, referrals, medical supplies, and increased knowledge of MCH health in order to triangulate the quantitative data. From the village government, participants included Village Executive Officers (VEOs), village chairmen, traditional birth attendants, Village Health Workers, village supervisor, hamlet leaders and WAJAs. From the government and health care providers, the respondents included health care workers (doctors, clinicians and nurses) and members of the Community Health Management Team (CHMT). A total of 88 IDIs and 24 FGDs were conducted by native Swahili-speaking interviewers (see Table 1). On average, each interview lasted for forty-five minutes to one hour and the FGD took between sixty and ninety minutes. The FGDs averaged 12 respondents, including women and men, categorized by age, gender and profession [35]. An additional file has been included that shows more details on the questions administered (see Additional file 1). Upon consent, interviews were audio recorded and then translated into English by experienced translators. In both rounds of qualitative data collection, the same interviewers were used for consistency. Prior to data collection, the interviewers were trained by a senior IHI researcher on research ethics and confidentiality as well as how to correctly phrase the interview and focus group guides. Community and district authorities assisted the researchers in identifying respondents from the Connect sites. The researchers had a list of positions in the community as well as categories. The local authorities would introduce the researchers to the appropriate individual occupying the requested position or a representative from the community for the requested categories. A list of the positions and categories of the respondents can be found in Table 1. Descriptions of IDIs, FGDs and Types of Informants The second source of data for this article comes from ethnographic research that involved observation and participation in WAJA’s professional, communal and personal activities. Two researchers were involved in the ethnographic study, a medical anthropologist completing his doctoral degree and a research assistant with a university degree. The ethnographic study occurred over a nine-month period from October 2013 to June 2014. It involved four villages in Kilombero District: Katindiuka, Lumemo, Mlabani and Kisawasawa. Researchers accompanied supervisors distributing supplies to the WAJAs, attended training of village health teams, observed mass meetings of WAJAs and district supervisors, reviewed WAJAs’ monthly reports with the supervisors, visited health centers and dispensaries, interviewed health workers at the centers and dispensaries, and interviewed community members about WAJAs’ services. Our research also included an observation period that entailed visiting six WAJA, four females and two males, three times a week for 12 weeks in their villages. We spent an average of six hours a day observing and participating in WAJA activities alternating between morning and evenings among the four villages. We kept a daily record of our observations in the form of field notes, and we discussed salient findings and topics at the end of each day. The ethnographic data focused on WAJA’s professional work including household visits, which entails case management, educational sessions on maternal and child health and family planning, referrals, patient check-up and consulting with supervisors and fellow WAJAs over the phone. The data also included WAJA’s “non-professional” money-making roles primarily farming but also brick making boda boda (motorcycle) taxi driving and the pressing of palms for oil. The other data was around their communal roles – attending meetings, prayers and community gathering such as funerals, weddings and baptisms. Prior to beginning the ethnographic research, we analyzed QSA 1 and 2 qualitative data to discover broad themes and topics related to the WAJAs’ reception in their own villages. Three team members were involved in reading the IDI and FGDs including one member who was part of the data collection in QSA 1 and 2 and two members who were not part of the data collection team. The initial analysis of the qualitative followed grounded theory procedure, an inductive research method that strives to generate concept within the data, privileging description over abstract categories and engaging in constant comparison between data sets [36]. We pursued open coding to determine the frequency, similarities, relationships and contexts that shape WAJAs reception in the study area. Based on the general findings, we adopted categories such as lack of medicines, delays of salaries, kinship relations and income generation activities all of which emerged from the reading of IDIs and FGDs. This was achieved through writing memos, sharing notes and conducting discussion among the three researchers. We also recorded salient quotes and information from interviewees based on the general analysis of the IDIs and FGDs data. We used these first impressions, themes and topics to guide us but not to determine the scope of our ethnographic research. We repeated the same process for our ethnographic data. We pursued open coding to analyze our field notes by constantly reflecting on themes generated earlier from the qualitative data, noting frequency and similarity of themes, levels of detail, and relationships among the IDIs, FGDs and ethnographic fieldnotes. We wrote memos, shared notes and met to discuss to draw out the connections, similarities and different aspects of our data. At the axial coding stage, we engaged in an overall interpretation of data based on the patterns that emerged. This procedure verified earlier observations such as delay of medical supplies but also revealed a richer picture of the implications of such findings to WAJA’s professional role. Axial codes were combined to provide an explanatory framework of how WAJAs’ roles and identities blur and interact with implications to their work. At the final, integrative stage, we generated a working theory to explain how previous and new roles interacted positively and negatively on WAJAs’ work. The strength of this study was the use of ethnographic approach to understand community’s reception of the introduction of a specific type of community health workers. Therefore, the researchers spent nine months observing the WAJA and were able to triangulate and contextualize the findings in QSA I and II. To minimize observer’s effect, researchers accompanied WAJAs on their pre-planned work schedules instead of creating activities for the researchers. The limitation of the current study is that the data used was collected five years ago. Several of the issues that undermined CHWs professional and personal identity such as delays of salaries or medical supplies may have changed or resolved. Another weakness is the study’s sample size particularly for observational data. The researcher observed only six WAJAs from four villages in Kilombero district, which means the issues and factors observed during the study were place and time-specific.