Menu
Menu
Menu
Menu
Background: Anemia and underweight among women are major public health challenges. Access to health services can improve dietary behaviors and women’s nutritional status. We examined whether exposure to health services is associated with women’s dietary practices in Tanzania. Methods: Data come from a cross-sectional baseline survey among 5000 female primary caregivers who were randomly selected via two-stage sampling, prior to implementing a maternal and child nutrition program. We ran frequencies on women’s exposure to existing health facility-based counselling, community health worker visits, and attendance at women’s support groups. We examined associations between exposure to these interventions and maternal diets and adjusted for sociodemographic covariates using ordinary least squares regression and ordered logistic regression. Results: A third of the sample (34.1%) had received any antenatal care (ANC) during their most recent pregnancy or had been advised by anyone about nutrition (37.0%). 68.0% had never had a community health worker (CHW) speak to them about their children’s health and 9.4% had participated in a women’s group. Only 8.0% of mothers ate more than usual during pregnancy and 7.1% ate more types of foods. After adjusting for mother’s age, education and household assets, women who received nutrition advice were 1.3 times (95% CI: 1.1, 1.7) more likely than mothers who did not to eat more during pregnancy. Receiving antenatal care (ANC) and advice on nutrition before, during, and after pregnancy and delivery were highly associated with the mother eating more types of foods. Hearing from a CHW about children’s health but not support group attendance was often associated with various dietary practices. Almost all measures of access to health services were significantly associated with mothers’ frequency of eating in the previous 24 h. Receiving advice on nutrition during pregnancy and after giving birth and CHW contact were associated with mothers’ dietary diversity in the previous 24 h. Conclusions: Several program exposure variables—especially being counselled about nutrition—were associated with improved dietary practices. Improving service delivery at scale may contribute to improved dietary behaviors in larger populations, given the associations we describe, along with findings from the existing literature.
This study used data from a cross-sectional baseline survey using two-stage cluster sampling to randomly select respondents. The survey was conducted in January and February of 2016, prior to implementing a large, integrated nutrition project (Addressing Stunting in Tanzania Early or ASTUTE). In 2015, IMA (Interchurch Medical Assistance) World Health and its consortium partners launched a five-year projected funded by the United Kingdom Agency for International Development (UKAid). The project targeted women and children in five Northwestern regions of Tanzania, collectively representing a population of 10.2 million and more than 750,000 stunted children. These five regions were selected because they had a high prevalence of stunting and anemia and poor maternal and child feeding practices relative to the rest of the country. The baseline informed program design. This study’s intent is to 1) document government and other implementing partner program coverage prior to the start of the ASTUTE project, and 2) assess associations between exposure to existing government programs and maternal health practices. After baseline, the ASTUTE project was implemented in five lake zone regions of Tanzania (Geita, Kagera, Kigoma, Mwanza, and Shinyanga) and its primary objective was to reduce stunting among children under 5 years of age with improvements in maternal diets, antenatal care seeking, hand hygiene, sanitation, and other behaviors as secondary objectives. Future research will examine associations between exposure to social and behavior change activities that were implemented as part of the ASTUTE project, antenatal care seeking, and maternal diet. Study participants include 5000 female primary caregivers of children aged 0 to 23 months. We recruited respondents from five geographic regions, including Geita, Kagera, Kigoma, Mwanza, and Shinyanga. We used two-stage probability proportional to size sampling, first at the district level and then at the village level in rural areas and neighborhoods in urban areas, employing data from Tanzania’s most recent (2012) census as the sampling frame. Once randomly selected villages or neighborhoods were identified, we selected 20 households from within each village/neighborhood using a spin-the-bottle approach to choose an axis that interviewers could follow to identify the first household for interview. In rural areas, interviewers were required to identify houses at least 200 m apart. In urban areas, we selected every fifth house for interview (in buildings with more than one eligible household, only one household was interviewed). We field-tested the survey instrument among mothers and fathers then revised and finalized it prior to administration by IPSOS (Institut de Publique Sondage d’Opinion Secteur) Tanzania, which is part of a global data collection firm. We scripted the questionnaire onto a mobile data collection platform and uploaded it to Android mobile devices used for data collection. We obtained informed consent from all study participants—written if the respondent was literate and by thumb print if not. The National Institute for Medical Research in Tanzania and relevant local government authorities authorized the research (NIMR/HQ/R.8a/Vol.IX/2344). Three research teams, trained by IPSOS Tanzania, administered the questionnaire. Interviewers conducted one-hour face-to-face interviews in Kiswahili. Interviews took place at the caretaker’s place of residence. We made three attempts to contact mothers in their residence, after which replacement households were selected. There were 150 refusals total in the five regions (2.9% of all individuals contacted). Upon completion of data collection, IPSOS Tanzania compiled survey results for cleaning and analysis. Outcomes included whether the mother, at any time during her most recent pregnancy resulting in the birth of the youngest living child, ate more than usual (a single, subjective measure) and consumed more types of foods than usual (also, a single, subjective measure). Additional outcomes included the number of times the mother ate in the previous 24 h and her dietary diversity based on seven food groups (grains, legumes and nuts, dairy, flesh foods, eggs, vitamin A rich foods, and other fruits), also measured for the 24 h prior to interview. Each of these behaviors was self-reported. Our measure of dietary diversity differs somewhat from the current 10-item Minimum Dietary Diversity for Women (MDD-W) index used by the Food and Agriculture Organization [20]. In particular, the MDD-W considers pulses (beans, peas and lentils) to be separate from nuts and seeds. The MDD-W also includes a category for other vegetables, in addition to dark, green leafy vegetables. However, when our baseline was carried out (2016), the Food and Agriculture Organization had not yet published new guidance about measuring maternal dietary diversity. Thus, our dietary diversity score is consistent with global standards as of early 2016. Wherever possible, we used the same questions as those used in the 2015 Tanzania Demographic and Health Survey (TDHS) [7]. However, the TDHS does not ask about the amount and types of food consumed during pregnancy, exposure to counselling on maternal nutrition, contact with community health workers, nor participation in support groups. Each question not included in the TDHS was pre-tested then modified based on results from pre-testing. Demographic data included information on the mother (ethnicity, religion, years of schooling, literacy, and age plus whether she personally owned a mobile phone), household (housing construction and assets ownership, whether the household owned a radio or TV, and the number of other children in the family), and community (travel time to the nearest market and health facility). The asset indicator was created by summing the number of assets respondents indicated they owned out of 13 possible assets, including bicycles, cars, carts, radio, and television, among others. The household construction index was created based on the construction materials used for the floor, roof, and walls of the dwelling ranging from three (if the walls, floor, and roof were made of rudimentary materials) to nine (if the walls, floor, and roof were made of finished materials). Access to services pertains to the availability of safe drinking water sources (e.g., protected wells, a public standpipe) and safe sanitation (e.g., a flush toilet). Pit latrines were not considered to be improved sanitation, per the Joint Monitoring Program of the WHO. The household wealth index was adapted from a previously validated index [21]. The index is comprised of the two sub-indices described above. An average of the two indices was used to calculate an overall wealth score, with possible values ranging between 0 and 1. Higher wealth scores indicate higher socioeconomic status. Stata 14.2 (College Station, Texas, USA) was used for all statistical analyses. We calculated chi-squares and t-tests to gauge unadjusted associations between exposure to health programs and services and measures of maternal diet. Ordered logistic regression modeling was used to determine whether associations described in bivariate analysis persisted after adjusting for maternal age and education as well as household assets. These covariates were chosen based on conceptual and statistical considerations (including the need to avoid collinearity and overfitting models). Ordinary least squares modeling was conducted for outcomes that were continuous (mother’s dietary diversity and frequency of eating in the previous 24 h).
N/A
