This research was conducted in Misungwi District located in Mwanza Region of Tanzania’s Lake Zone (Figure 1). Misungwi District is rural, located 45 km from Mwanza city and at last census (2012) had a population of 351,607 (33). Administratively, the district is sub-divided into 4 divisions, 20 wards, and 78 villages. In 2019, 91% of households in Misungwi District were ethnically Sukuma (13). The Sukuma are a patrilineal society in which women are expected to take care of their husbands and children (34). Those individuals included in this study were low-income, living in villages scattered throughout flatland terrain, and subsisting via the cultivation of maize, millet, rice, sweet potatoes and vegetables, cattle grazing on communal lands, and fishing. Most households surveyed in 2019 reported using firewood (83%) or charcoal (14%) for cooking fuel (83%) (13). Sixty-eight percent of households owned livestock and 62% owned agricultural land (13). Thirteen percent of households were connected to electricity, 80% had mobile phones, 57% owned a bicycle, and about 10% owned a mechanized form of transport (13). Piped water, and advanced sanitation facilities are not common. Each of the four villages considered in this study had a primary school and attendance in primary school in Tanzania is compulsory. However, in Tanzania, there exists a lack of a quality, formal education, especially in rural, poor regions where the long distances to schools and insufficiently qualified and motivated teachers, a lack of teaching materials, textbooks and basic technology, and required financial “contributions” (35) are disincentives for some students (36,37). Moreover, Tanzanian girls are more likely to drop out of school than are boys due to their caretaking responsibilities (38). In 2019, Misungwi District had 48 formal health facilities providing delivery services. The district, along with others in the Lake Zone, has amongst the worst maternal, newborn, and child health indicators in the country (11), and is prioritized by government for maternal newborn health programming.
Insert Figure 1 here.
The study used Criterion-i (39), purposive sampling (40) to identify two rural divisions in Misungwi District and to select study participants who were knowledgeable or experienced with the phenomena under study (41): barriers to receiving Tanzania’s recommended antenatal, delivery and postnatal care services (42). Villages were selected for inclusion by first ranking the four Misungwi District divisions surveyed in 2016 (13) on the basis of accessing antenatal, delivery and postnatal care services, then selecting the two divisions, Mbarika and Inolelwa, with the lowest overall rates of accessing care. Within Mbarika and Inolelwa, cluster randomization (43) was used to identify four wards for study and within each ward, one village was randomly selected, for an overall total of four villages.
A total of 81 illiterate women of reproductive age who were either pregnant or had children and seven influential people, described below, were selected for inclusion in this study. All participants signed informed consent forms, described below. No incentive was provided to the participants, other than refreshments, unless the participant incurred transportation costs to attend the interview(s), in which case transportation costs were refunded.
The illiterate women of reproductive age were recruited by first explaining the purpose and methods of the study to the village leaders and the village-based, volunteer, community health workers (CHWs), then asking them to identify households most likely to have illiterate women of reproductive age. The village leaders and CHWs knew all residents of households in their catchments and had a sense of their literacy status. A second meeting was then held with all members of the village who wished to attend to explain the purpose and methods of the study.
Subsequently, the households of potential illiterate women of reproductive age were visited by field researchers who explained the project. For those pregnant women or mothers of reproductive age who had not completed school beyond the primary level and agreed to continue, literacy was assessed by a standardized protocol (44). Only women who could not read at all were classified as illiterate and eligible to proceed in this study. This process was continued to attain a minimum sample size of 20 illiterate women of reproductive age in each village.
Those perceived to have some influence on women’s decisions concerning antenatal, delivery and postnatal care services were also invited to participate, in order to triangulate experiences of the illiterate women of reproductive age (45). These included opportunistically recruited CHWs and other healthcare providers in each of the four villages. CHWs were community members who were selected by their communities, trained using a national curriculum, and expected to voluntarily provide health promotion education and support emergency referral care (e.g. if a CHW identifies an at-risk mother needing health care, the CHW would ‘refer’ the mother to a health facility) to households in their community, especially to pregnant and newly delivered women. Healthcare providers selected for interviews included nurses and clinical officers providing antenatal, delivery and postnatal services at health facilities. Potential participants were excluded if they had not been active in their roles in the community for at least the prior six months. A total of two influential individuals were sought in each village. The field researchers met this goal in three of the four villages but was able to recruit only one influential individual in one of the villages.
Data were collected July-September, inclusive, 2018, in focus group discussions (FGD), in-depth interview (IDI) or key-informant interviews (KIIs). Semi-structured facilitator guides were used to maintain consistency across FGDs, IDIs, and KIIs (46). To ensure guiding questions resonated with participants, the facilitator guides were piloted twice in two other, similar villages in the Mbarika and Inolelwa divisions. Questions and probes were refined after the pilots to better reflect the context of the region (47).
The morning after women were selected and confirmed, FGDs were held with illiterate women of reproductive age to gain an understanding of factors influencing the antenatal, delivery and postnatal care-seeking practices of this group (48,49). FGDs were held in a community space chosen by the illiterate women of reproductive age. In FGDs, field researchers took a peripheral role to facilitate a group discussion between participants.
Later that day, IDIs were held with individual, illiterate women and KIIs were held with the CHWs and healthcare providers. In IDIs and KIIs, interviewers engaged in a probing conversation with the interviewee (50,51). Individual IDIs were held with illiterate women of reproductive age to explore topics mentioned in the FGDs in more depth. The illiterate women who participated in the IDIs were selected at random from those who participated in the FGDs. KIIs were held with CHWs and healthcare providers as they generally have a good sense of the needs and practices of pregnant women and those with newborns. IDIs and KIIs were conducted in a private location.
FGDs, KIIs, and IDIs were conducted in the local vernacular, Sukuma. Field researchers, comprised of a moderator, note-taker, and an observer, all fluent in Sukuma, facilitated the interviews. FGDs generally lasted 1-2 hours; IDIs and KIIs were 45-60 minutes long and were audio-recorded. Overall, 8 FGDs, two in each village, composed of 8-10 illiterate women of reproductive age each, were conducted with follow-up IDIs completed with 13 (16%) of these women. Seven ‘influencer’ KIIs were conducted: 3 with CHWs and 4 with healthcare providers.
Recorded Sukuma interviews in were transcribed and translated directly and verbatim into Swahili as Swahili was the primary language of the Tanzanian researchers. Transcriptions and translations were checked for accuracy by four of the Tanzanian researchers, fluent in Sukuma and Swahili, who did not conduct the original interviews or transcription/translation. Two additional Sukuma speakers conducted Sukuma source transcripts quality checks. Resulting Swahili transcripts were then translated to English by Tanzanian researchers fluent in English and Swahili.
To provide a systematic account of the observed phenomena and transform interviews into a set of cohesive and meaningful categories, data were coded in four steps using NVivo (v. 12) (52) and, in step five, the importance of themes was determined. In step one, four randomly selected transcripts, including one IDI, one KII and two FGDs, were used to develop a coding template. Here, each of these transcripts was coded individually and the final codes subsequently agreed upon. In step two, four additional transcripts were selected at random and new codes were added if they did not fit with the initial codes. Step two resulted in the final codebook for the study. In step three, 18 additional transcripts were coded for a total of 26 (8 FGDs, 11KIIs and 7 IDIs) of the 28 transcripts were coded after which it was determined that saturation was reached; that is, new themes or sub-themes were unlikely to emerge from analysis of additional transcripts (53). In step four, thematic analysis was used to collapse the codes into basic themes and subthemes (54). In step five, frequency reports were created and cross-tabulations were used to determine the relative importance of themes in the data and their linkages to one another.