Association Between Nutrition Social Behavior Change Communication and Improved Caregiver Health and Nutrition Knowledge and Practices in Rural Tanzania

Background: Efforts to improve infant and young child feeding practices include the use of nutrition behavior change communication among caregivers of children under 5 years. We assessed the association between monthly participation in community-level nutrition group meetings on caregiver health and nutrition knowledge and practices (KPs). Methods: Data from a community-based cross-sectional survey conducted in the Eastern and Southern Highland Zones of Tanzania were used. Indices were developed for caregivers’ knowledge of nutrition, health and childcare, household (HDD) and young child dietary diversity (CDD), and vitamin A (VA) intakes. The comparison of means and proportions was assessed using Student’s t-test and the Chi-square test, respectively, between the caregivers participating in nutrition group meetings and non-participants. The impact of the number of nutrition meeting attendance on caregiver KPs scores was examined using multiple regression. Results: Of 547 caregivers surveyed, 49.7% attended nutrition group meetings and received information on nutrition social behavior change communication (SBCC). Overall, 28% of participating women had a moderate level of nutrition knowledge, 62% had a high level of VA knowledge, and 57% had a high level of health and childcare knowledge. Participation in nutrition group meetings was significantly associated with the health and childcare knowledge score (HKS), HDD and CDD scores, and household and young child VA intake; the magnitude of the associations was greater for caregivers who attended at least four meetings. Conclusion: The findings emphasize the need for programs that seek to address the issues present in the use of nutrition SBCC at the community level to improve maternal or caregiver KPs and subsequently the nutrition status of infants and young children. This community-based cross-sectional survey was conducted between August and September 2017 in all the seven VISTA Tanzania project intervention districts. The project districts were Gairo and Ulanga in the Morogoro region; Mufindi and Iringa districts in the Iringa region; and Wanging’ombe, Chunya, and Mbozi districts in the Mbeya region. The villages in the project intervention districts were enumerated in August 2017 in preparation for sampling the villages and households to be surveyed. In the three regions, farming is the main economic activity. In terms of agro-ecology, the Morogoro region falls in the eastern agro-ecological zone, while Iringa and Mbeya regions are in the Southern Highlands agro-ecological zone. Both agro-ecological zones receive the highest annual rainfall in Tanzania and are homes to major water bodies that influence the eco-climate, while the numerous rivers are used for many small-scale irrigation schemes. Maize, cassava, rice, potato, and sweetpotato are the main staple crops grown. Cattle raring, small ruminants as well as poultry farming are widely practiced in these regions. Sweetpotato is produced mainly for home consumption and is consumed as boiled, roasted, or deep-fried storage roots. However, sweetpotato leaves in Tanzania are also consumed in local diets and are a common green vegetable in the rural and urban markets. At the beginning of the project in 2015, there were no documented data on the proportion of households consuming OFSP to our knowledge, which would have been very important for our research in the project target district as a benchmark. However, the project baseline survey revealed that only 0.4% of the households had consumed OFSP during the previous 24 h (19). Elsewhere in Tanzania, a study conducted in 2012, in the Lake agro-ecological zone, revealed that about 2% of households consumed OFSP at least one time every week (20). We anticipate, through the implementation of the VISTA Tanzania project in these selected districts in the eastern and southern agro-ecological zones, where the prevalence of vitamin A deficiency (VAD) is high (36%) among children of 6–59 months (3), that it will be more beneficial to achieve a higher (10%) consumption of provitamin-A-rich OFSP during the previous 24 h among the participating households. The study targeted households with children <5 years old (6–59 months). Caretakers of these children were the primary respondents. These primary caretakers also participated in the monthly nutrition group meetings in the communities that were established and run by the CHWs. These primary caretakers were mostly the biological mother of the children or the grandmother. During each monthly group meeting, CHWs provided improved maternal, infant, and young child nutrition counseling (16). The main communication aid for the CHWs was a desk-sized set of counseling cards containing eight lessons with each page on the chart containing illustrated examples of healthy practices on the front, with the accompanying messages on the back (4–5 key messages per topic). The main lessons were as follows: (1) healthy mothers during pregnancy; (2) healthy eating; (3) VA; (4) biofortification; (5) infant feeding; (6) OFSP benefits; (7) growing OFSP; and (8) creating a kitchen garden and planting fruits. At each community group meeting, the CHW was to conduct lesson #5 and present one additional topic. Crucially, at these meetings, cooking demonstrations that utilized OFSP and other locally available nutritious and VA-rich foods were carried out by the CHWs with the full participation of the group members present. The curriculum and training models that were used to train the CHWs were adopted from our previous study in western Kenya (16) and modified and adapted for this current nutrition-sensitive project in Tanzania. Eligible participants included women of 17–45 years of age, residing in the study villages and who are either the biological mother or primary caretaker of a child (6–23 months) or who have a confirmed pregnancy (i.e., by a health worker). Lastly, women must have resided in the study villages during the period of intervention from April 2015 to August 2017. Women who are younger than 17 or older than 45 years of age and who are not in their first or second trimester of pregnancy or the primary caretaker of a child (6–23 months of age) were not eligible. Women who did not reside full-time in the study villages during the period of the intervention were not eligible. There were no known risks to these populations beyond some possible discomfort due to the need to assess certain targeted behaviors of the intervention (OFSP knowledge, production, diet practices, food consumption, etc.) or inevitable survey procedures. This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the Commission for Science and Technology/COSTECH, Tanzania. Written informed consent was obtained from all subjects. The sample size of the end-of-project survey was based on the same principle as the baseline survey of 2015 to enable a comprehensive and objective comparison of the primary KP outcome of household weekly frequency of OFSP consumption among project participants. This outcome was the change in the proportion of households consuming OFSP at least one time a week from 0.4% at baseline to 10% at endline. Similar assumptions were made on expected proportions of household weekly frequency of OFSP consumption and on expected changes according to the data of surveys conducted in the Lake Zone regions of Tanzania (20) and in western Kenya (21, 22). The sample size calculation was done to allow for comparison of the proportions between endline (10%) and baseline (0.4%) data accounting for the complex cluster survey design effect (DEFF of 1.5) (23). Based on an alpha error of 5% and power of 90%, the best estimate of sample size for the primary outcome of household weekly frequency of OFSP consumption was 426 for the seven project intervention districts. This sample size was distributed proportionately among the seven project intervention districts using the probability proportional to the size sampling technique (24). This sample size would allow comparisons for OFSP knowledge, growing practices and consumption, and dietary practices among households between the baseline and endline. The sample size was therefore increased to 512 households to cater for a 20% non-response. The endline survey was conducted in the seven intervention districts of the VISTA Tanzania project. Each district has unique characteristics, including the potential for expanding OFSP production; however, poor nutrition practices and low family income are common features among all the project target districts. The project intervention districts and households were purposively selected because they all fall within USAID Feed the Future's Zones of Influence besides having been used during the baseline survey. The survey used a multi-stage cluster sampling design to select the study respondents (24). The first stage involved selecting sample points (“clusters”) using “probability proportionate-to-size” cluster sampling based on the list of villages from each of the project intervention districts (24). Thus, 50 villages were randomly selected from the total number of villages in the project intervention districts. A list of all the households in the selected village that met the VISTA Tanzania project target intervention criterion was compiled with the help of village agriculture extension officers (VAEOs). The geographical reference of all eligible households was recorded and included in the sample frame for random selection of the eventual respondents. During enumeration, at least 30 families were selected in each village by the village executive officer and the VAEO assisted by the district-based agricultural experts. Emphasis was placed on families with children between 6 and 59 months old. In case households that failed this criterion were listed, they were not geo-referenced for the purpose of saving time since they will not form part of the population that would be sampled for the survey. Informed consent was then obtained from all the eligible participants. Among these households, 11 were randomly selected for individual interviews. Thus, 550 households, each represented by a young child's primary caregiver, were identified for enumeration as the main part of the endline study. Here, a household is defined as a person or a group of persons, related or unrelated, who live together and share a common source of food. Note that the study was oversampled (from 512 to 550) to ensure we have enough power to be able to detect potential differences in other outcomes. In each of the selected households, the primary caretakers of the children were the primary respondents. During the listing stage, in each village, a village leader was interviewed to gather information on the village's access to services such as agricultural extension services, market and health services, and information on other agricultural, health, or any community development intervention that might be happening in their community. A standardized, structured, smartphone-based questionnaire was used in which the responses per respondent were directly recorded by trained enumerators. All the survey tools were prepared by the VISTA staff in collaboration with project implementing partners (IPs), reviewed for accuracy and completeness, translated into Swahili, and pre-tested before administering in the field for data collection. Based on the pre-test results, the questionnaire was accordingly modified and finalized in consultation with the IPs. The questionnaire was divided into modules, with questions in each module intended to capture different information, knowledge, and practices among the target population about sweetpotato in general and OFSP in particular. The modules of the questionnaire were as follows: At enrollment for each respondent, data were collected on basic socio-demographic characteristics, such as age, marital status, education, occupation, household size, and composition. Data on agricultural resources and household assets were also collected to provide a context for understanding the overall results of this research. There were two teams of fieldworkers during the data collection phase of the survey. Each team comprised 11 enumerators, a team leader among enumerators, and a CIP staff as a supervisor. The team leader had the responsibility for visiting teams in the field, ensuring that households are selected accurately, and adequate survey tools and other logistics are available. The supervisor was also responsible for deciding how to overcome unexpected problems. Each problem encountered and each decision made were recorded and included in the supervision report. At the end of each workday, the team leaders conducted a wrap-up session with the team to discuss any problems encountered during the day and reviewed all questions and tracking forms to ensure accuracy and completeness. After a review of each completed computer-assisted personal interview (CAPI)-entered data, a backup was created before closing the day's work through Bluetooth technology. The interview of each caretaker of the eligible and selected household (HH) took approximately 50–70 min, and questions were asked in the Kiswahili language. Each interview was conducted at the home of the participant after she/he was reminded of the informed consent that was procured during the household listing exercise. At the end of each day, the team leader with the assistance of the supervisor reviewed the completed CAPI questionnaires and discussed issues and concerns about the day's interviews. The issues were addressed using field notes, and if necessary, interviewers would return to pertinent HHs to correct the errors. Six outcome variables that characterize maternal KPs were constructed. These variables were as follows: (1) nutrition knowledge score (NKS), (2) HKS, (3) household dietary diversity score, (4) young child dietary diversity score, (5) caregiver VA intake, and (6) young child VA intake. NKS and HKS were derived from key variables using equally weighted summative item scores (see Annex 1 in Supplementary Materials for a list of survey items used). Weights were not used to generate knowledge scores as the selected items were relatively homogenous and equally important and, therefore, would not benefit from the added complexity of weighting and would risk incorrect weight assignment to items (27). The NKS ranged from 0 to 14 points, and the HKS ranged from 0 to 13 points. The dietary diversity score (DDS) and VA intake score, for both caregiver/household and young children, were the primary health practices of interest. The caregiver's and young child's DDS were constructed from a 24-h food recall, adding the number of different food groups out of 12, which were consumed by the household within the last day (28). Specifically, for the household DDS, 13 food groups were included in the index calculation for households: (1) cereals, (2) roots and tubers, (3) vegetables, (4) fruits, (5) meat and poultry, (6) eggs, (7) fish and seafood, (8) pulses, legumes, and nuts, (9) milk and milk products, (10) oils or fats, (11) sugar or honey, (12) bio-fortified foods, and (13) miscellaneous foods (beverages and related foods). The OFSP was categorized as a biofortified food with both high energy and VA content. Each food group was scored as 0 if not consumed during the past 24 h and 1 if consumed during the same period. The dietary diversity index was obtained by summing the scores for the 13 food groups. Therefore, the possible range of the dietary diversity index was from 0 to 13. The household DDSs were grouped into tertiles: with a score of 0–3 categorized as “low;” a score of 4 as “medium;” and a score of 5 and above as “high.” For computation of a young child's (aged 6–59 months) DDS, the food groups used were as follows: (1) grains, roots, and tubers, (2) VA-rich plant foods, (3) other fruits or vegetables, (4) flesh foods (meat, fish, poultry, and seafoods), (5) eggs, (6) pulses, legumes, and nuts, (7) milk and milk products, (8) any oil or fat-fried/cooked food, and (9) any bio-fortified staples. Each food group was scored as 0 if not consumed during the past 24 h and 1 if consumed in the same period. The dietary diversity index was obtained by summing up the scores for the nine food groups. The possible range of the dietary diversity index was from 0 to 9. The child dietary diversity scores were then grouped into tertiles, with a score of 0–2 categorized as “low;” a score of 3 as “medium;” and a score of 4 and above as “high.” The frequency of VA consumption score was calculated using the Helen Keller International (HKI) food frequency index model to assess the household risk level of VAD (25). This model counts the frequency of how certain foods are eaten over time although it suffers from a failure to capture actual amounts of each food consumed. However, this model was validated against biochemical indicators and can be used to adequately predict whether VAD is a public health problem in the population. A household was considered to be at risk of VAD if the mean frequency of consuming VA from animal sources was 4 days per week or less or the mean frequency of total consumption of animal and plant sources of VA was 6 days per week or less. The frequency of the VA consumption score was calculated by first summing the number of days during the previous week the child or the caregiver consumed VA-rich food from ananimal source. Then, the number of days the child or caregiver consumed VA-rich food from a plant source was summed and divided by 6. The following formula was used in calculating the index: Weighted total consumption days (Cw) = Total number of days animal sources of Vitamin A consumed (TVA) + Total number of days plant sources of Vitamin A consumed (TAP) divided by 6. The weighted VA consumption score (C) is equal to the total number of days the child or mother consumed VA-rich food items from animal sources plus the adjusted consumption from the plant sources. The following animal and plant sources were included in the estimation of the index. The cut-off point for adequate frequency of VA intake was 6 for the weighted consumption score. Participation in community-level nutrition group meetings was the primary independent variable of interest. We hypothesized a priori that primary caregiver attendance and participating in nutrition group club meetings was a key source of nutrition and health knowledge acquisition. Further, based on our review of the literature and our previous study in western Kenya (29), we identified other maternal, household, and community-level factors that can potentially confound the association between nutrition group meeting participation and nutrition and health KPs. We used a conceptual framework (Figure 1) adopted from our study in western Kenya (29) to guide the selection of variables for our adjusted regression analysis. At the household level, we considered the status, age, educational level, and employment of the household head, as well as household size as potential confounders of caregiver's participation in nutrition group meetings. Maternal or caregiver socio-demographic characteristics, such as age, marital status, educational status, involvement in agricultural activity and selling agriculture products, engagement in salaried employment, cultivation, and consumption of sweetpotato (OFSP), were identified as potential confounding factors in the association between nutrition social behavior change through club meeting participation and outcome KP variables. Conceptual framework of the association between participation in nutrition group meetings and knowledge and practices of caregivers of young children. Adopted from Perumal et al. (29). The household wealth index was used as a proxy for the socioeconomic status of the household and was created by summing the values of different predominantly discrete data household variables, such as the type of housing and roofing, the presence and type of toilet, the source of water during the dry season, and the source of cooking fuel, as well as the possession of durable household assets such as radio, television, telephone/mobile, solar panels, gas cooker, bicycle, water pump, motorcycle, car /truck, tractor, and generator. A wealth index based on ordinal variables for these data was created to allow comparison across sites (30). The wealth index scores were then grouped into tertiles, with a score of 0–11 categorized as “poor;” a score of 12–14 as “medium;” and score equal to or >15 as “high.” The maximum score for the wealth index was 29. At the community level, the presence of a nutrition club, village health and nutrition committees, and trained CHWs were considered as factors that potentially affected participation in nutrition group meetings and caregiver KPs. CSPro-supported CAPI data entry system was used to collect and collate data. In CAPI, the enumerators used smartphones to enter responses on site during the interview. The CAPI application enabled interviews to be conducted face-to-face and determined the question order and performed editing of responses as well as skip patterns. CAPI, therefore, offered a flexible approach to collecting and editing the data, resulted in better data quality, and improved the efficiency of interviewing and final data processing. The endline survey was conducted under a common goal for each village and household sampled in the districts with the intention of pooling the data for analysis. Thus, every effort was made to ensure consistency in the survey execution at every household. All the data were subsequently combined for all the sampled villages and households through a centralized database management system. After data collection and collation, reports were generated using Stata version 14.1 (StataCorp., College Station, TX) for basic logic, range, and missing data checks. The data were then cleaned and locked for analysis. Descriptive statistics, including frequencies and proportions for categorical variables and mean with standard errors for continuous variables, were generated for study participants. In bivariate unadjusted analyses, we used Chi-square and Fisher’s exact tests (for proportions) and Student’s t-test (for continuous variables) to compare baseline characteristics of the study participants between the two groups (those attending at least one nutrition group meeting vs. non-attendance). We hypothesized a priori that caregivers who participated in nutrition group meetings will demonstrate higher nutrition and health and childcare knowledge and, consequently, better household and young child dietary diversity and VA intake. Differences in nutrition and health and childcare knowledge and dietary diversity and VA intake were compared between the two groups. With the use of cluster-adjusted regression analysis, we examined the differences between participants and non-participants in the nutrition group meetings. This accounted for the cluster sampling and hierarchical nature of the data. We used multiple linear regression analyses to assess the effect of nutrition group meeting participation (none, <4, and ≥4 visits) on nutrition and health and childcare knowledge, better household and young child dietary diversity, and VA intake, adjusting for caregiver and household-level potential confounders, such as maternal age, marital status, education, employment, household size, sweetpotato cultivation, wealth index, source of consumed OFSP root, and status of head of household. The model was developed by backward stepwise elimination, removing the covariate with the largest p-value at each step until the remaining variables were significant at the 0.05 level in the final model. All statistical analyses were assessed by using SAS 9.4 (SAS Institute Inc., Cary, NC). A p-value of 0.05 was deemed statistically significant for all analyses. Data are shown as mean ± standard errors (SEs).