Improving dietary diversity through on-farm production and nutrition education in rural households of Tanzania

Background: The human body requires nutrients from varieties of food groups for a healthy and productive life. These nutrients primarily are from food crops and livestock products, thus make farm production as an important pathway for improving not only dietary diversity but also good nutrition status of a population. However, availability of foods is one thing but the choice and decision on what to eat, how to prepare and allocate the foods to the family members is governed by several factors including Knowledge, Attitude and Practice. The pathways and mechanisms in which dietary diversity is linked with farm production and nutrition knowledge have not been adequately studied. Given the fact that malnutrition is still a huge challenge especially in rural areas of Tanzania, where farm production activities are mainly taking place, it will be important to explore the pathways and associated factors. This study aimed to establish factors associated with household dietary diversity in the context of farm production and nutrition knowledge. The study intended to measure the associations and predictors of household dietary diversity using different indicators of farm production in addition to nutrition knowledge, socio-demographics, economic and agricultural characteristics. Methods: The study was conducted in three phases, the baseline (July-August 2016), intervention (September, 2017 to April, 2018) and the end-line phase July-August 2018. The baseline phase involved cross-sectional data collection from 663 women/caregivers in rural households of Dodoma and Morogoro regions. Nutrition education intervention was given for a period of 8 months. It included 10 days group training with six-month individual training follow-ups in the household. Spouses/ adult men were also invited to participate in the intervention. Training contents included: functions of food, food groups, malnutrition, food preparations, food consumption and homestead food production. Women/caregivers were given pre-test at the baseline and a post-test during the end-line survey to assess the effect of the nutrition education intervention. The end-line-phase involved 577 women/caregivers in rural households of Dodoma and Morogoro regions. Data for household dietary diversity, farm production, nutrition knowledge and other socio-demographic variables were collected using questionnaires. Data were analysed using SPSS version 20. For descriptive statistics, mean and standard deviation were used for parametric data, while median and interquartile ranges were used for non-parametric data. Pearson correlation and Chi-square test were used to determine the correlation and relationship among variables. Mean separation was done by using one-way ANOVA with Tukey’s post hoc test, level of significance was set as P = < 0.05. Binary logistic regression was applied to identify predictors of the association between dichotomous variables. For non-parametric inferential statistics, Wilcoxon signed ranks test was used for comparison of variables before and after the intervention, Mann–Whitney U test for comparison of two groups and Kruskal Wallis test for comparison of more than two groups. The McNemar test was also applied to establish differences in frequencies between baseline and end-line. Results: At baseline, the median household dietary diversity score (HDDS) was low (4.2) with less than half (43.7%) of the households reported consuming at least 5 food groups in all the regions. The Median HDDS differed significantly between regions; Dodoma had lower HDDS (interquartile range) of 3.9 (2.9, 5.2) compared to Morogoro region of 4.5 (3.5,5.8). Furthermore, the baseline results revealed a high level of nutrition illiteracy in the two regions. Only 14% of the population had received nutrition education/information prior to the baseline survey. The mean score for nutrition knowledge and practice was 6.9 (±2.6) out of 20 and only 17% of the study population scored above the mean. At the baseline all households (100%) reported growing at least a single crop and only 52.5% of the households owned at least one type of livestock species. Farm production diversity and growing of specific food groups such as dark green vegetables, other vitamin A rich fruits and vegetables, nuts and seeds had shown to increase HDDS. Despite a mismatch between the proportion of households that reported keeping livestock (52.5%) and those who reported consuming of animal source foods (26%), still livestock keeping had shown to increase HDDS. After adjusting for socio-demographic and other variables in the regression model, the higher farm production diversity, daily food expenditure and nutrition knowledge consistently predicted an increase in household dietary diversity score in both regions. The effect of farm production diversity on HDDS was slightly higher in Dodoma households (ß-coefficient =0.29, t = 5.65, p<0.001) than in Morogoro households (ß-coefficient =0.25, t = 4.62, p<0.001), however, the effects of daily food expenditure on HDDS was more pronounced in Morogoro households (ßcoefficient =0.29, t = 5.87, p<0.001) compared to Dodoma households (ß-coefficient =0.19, t = 3.88, p<0.001). Results of the end-line phase after implementation of nutrition education intervention, indicated that 96% of mothers/caregivers reported having included at least 5 food groups in their household meal compared to the baseline phase (54%). A significant increase in the frequencies of consumption of vegetables, fruits and legumes were noted from the baseline to the end-line survey (87% vs 98%, 63% vs 69% and 76% vs 87%), p<0.001, respectively. Furthermore, the median scores for both nutrition knowledge and nutrition practice increased significantly at the end-line survey. In particular, the findings revealed that proportions of households knowing the importance of growing fruits and vegetable, importance of including different food groups in a meal, dietary enhancers of iron and recognition of malnutrition signs, increased at the end-line phase. The higher scores were recorded among those who had received nutrition education, who had frequently been exposed to nutrition education, whose spouses/men participated in the intervention and those with a formal level of education than their counterparts. Conclusion: The findings of this study entail that any intervention for promoting household dietary diversity should consider farm production diversity, status of nutrition knowledge, household food expenditure and market accessibility. The study has shown that household ability to diversify diets is based on the capacity to produce more varieties of food groups on their farmland. These include production of livestock coupled with production of dark green vegetables, other vitamin A rich fruits and vegetables, nuts and seeds. The importance of exposure to nutrition education came out vividly in this study. Mothers/caregivers who had frequently been exposed to nutrition education had higher household dietary diversity than their counterparts. In addition, household dietary diversity was higher among the households in which their spouses/men participated in nutrition education training than their counterparts. This suggests the necessity of involving men in nutrition education intervention to facilitate the retention of nutrition knowledge and the adoption of desirable dietary practices in households. Furthermore, other dimensions such as food expenditure and market accessibility appeared to enhance household dietary diversity especially when production diversity in the home-stead farm is limited. The study has also shown that, access to the nearest market and an increase in the household food expenditure diminishes the positive effects of farm production diversity over the household dietary diversity. Therefore, observations from this study call for further research to explore the influence of market accessibility and food expenditure on the household dietary diversity especially in rural areas where market infrastructure is poor and per capital income is low.

Scale-N project