Six, 10-minute abstract-driven presentations.  

Speakers and Presentations:

• Chair: A. Laxmaiah, National Institute of Nutrition

• Abu Hayat Md. Saiful Islam, Bangladesh Agricultural University (BAU)
  Patterns, determinants and food and nutrition security implications of home gardens in Bangladesh: Evidence from nationally representative household panel data
  Slides/ Recording

• Amy Ickowitz, Center for International Forestry Research (CIFOR)
  From growing food to growing cash: The impact of agrarian transitions on diets and nutrition in rural Indonesia
  Slides/ Recording

• Hollyn Cetrone, Northwestern University, USA
  A participatory agroecological intervention decreases depression amongst female smallholder farmers in Singida, Tanzania
  Slides/ Recording

• Marianne Santoso, Cornell University, USA
  Greater production diversity, more men’s involvement in household chores, and lower women’s depression mediate improvements on child’s dietary diversity in a participatory agroecological intervention in Singida,Tanzania
  Slides/ Recording

• Heather Ohly, University of Central Lancashire, UK
  Exploring the cultural acceptability and sustainability of biofortification in Pakistan
  Slides/ Recording

• Mia Blakstad, Harvard University, USA
  A cluster-randomized home gardening program improves dietary diversity and food security among rural Tanzanian women
  Slides/ Recording

• Q&A
  Recording

Abstracts:

Patterns, determinants and food and nutrition security implications of home gardens in Bangladesh: Evidence from nationally representative household panel data

Abu Hayat Md. Saiful Islam, Bangladesh Agricultural University (BAU)

Introduction:  Different research and programs are advocating for production diversification as it can improve food and nutrition security. Consequently, there is much attention towards home gardens as a strategy to enhance household food security and nutrition. But evidence is lacking, and most of the previous studies were project-based evaluations using observational rather than rigorous econometric methods. Under this backdrop, this study aims to identify the determinants and impact of home garden ownership and home garden production diversity on household and women’s dietary diversity in Bangladesh.

Methods:  Using the International Food Policy Research Institute (IFPRI) Bangladesh Integrated Household Survey (BIHS) two-round nationally-representative panel data, we used the double hurdle model, which assumes that households must pass two hurdles: i) decide whether to do home gardening or not; ii) extent of production diversification (intensity of production diversity) which is conditional on the first decision. To estimate the relationships between food and nutrition (FN) and home garden, we define food and nutrition security indicators for household and women (FNit) as a function of household home garden ownership and diversity in home garden (HHGit), individual variables (Iit), household demographic characteristics (Hit), community variables (Cit).

FNit = b0 + b1HHGit + b2 Iit + b3 Hit + b4Cit +ci+ εit

The equation is estimated by applying the Poisson Fixed Effects Model. We suspect that the estimated home garden ownership and home garden production diversity effect would suffer from systematic selection bias. To overcome the selection bias problem, we have used a fixed-effects (FE) estimator including year dummies which leads to a two-way FE model. FE models have recently been used to control for selection bias in different contexts.

Findings:  For measuring key independent variables, i.e. home garden production diversity, we use the number of crop species produced on the homestead. On the other hand, for measuring key outcome variables, i.e. dietary diversity, we have used the food variety score (no. of food items consumed) and the dietary diversity score (no. of food groups consumed). The double hurdle model results shows that homestead area, household head age, education, farm size, and household head’s wife earning status are significantly and positively and household off farm income is significantly and negatively associated with the probability of having a home garden, as well as the production diversity in the home garden. We find that access to market, which was measured by distance, mainly influences the probability of having a home garden, not the diversity in the home garden. The Poisson Fixed Effect regression result shows that home garden ownership and production diversity in the home garden is positively associated with dietary diversity of household and women. Further the role of other factors that may influence dietary diversity, such as market access measured by distance from market and households selling and buying status, off farm income, agricultural technology adoption and other socio-economic variables, are also analyzed.

Conclusions:  Home-based food production and home gardens in particular have the potential to address malnutrition in developing countries like Bangladesh. We use two waves of BIHS panel data and relevant panel econometrics. The result shows that having home garden and higher home garden production diversity is positively associated with household’s and women’s nutritional needs. Thus future policy and programs should focus on promoting home gardens and improving home garden production diversification to a certain extent, along with facilitating education, increasing income, and women’s empowerment.

From growing food to growing cash: The impact of agrarian transitions on diets and nutrition in rural Indonesia

Amy Ickowitz, Center for International Forestry Research (CIFOR)

Introduction:  Indonesia has been undergoing rapid changes on many fronts in the last two decades: economic development, dramatic land use change, and dietary changes resulting in new health and nutrition burdens. A key driver of these changes is a transition away from traditional agriculture to oil palm production. We present findings from a Drivers of Food Choice (DFC)-funded project to show how these changes in land use, social organization, and market integration are affecting local diets and health in two sites in Indonesia, one in Borneo and one in Papua.

Methods:  West Kalimantan, Borneo, is in the midst of major oil palm expansion, but still has substantial areas of traditional swidden farming. Papua has much less land in oil palm than Borneo, but it is considered to be the new ‘oil palm frontier.’ We randomly selected approximately 250 indigenous Dayak households with children between the ages of 12 and 59 months across 15 traditional swidden villages and 250 Dayak households across 18 oil palm villages in the same districts in West Kalimantan. In Papua, we randomly selected approximately 250 traditional hunter/gatherer households with children between 1 and 12 years of age and 250 households of similar indigenous Papuan ethnic groups where at least one member works in oil palm. In addition to questions about agricultural and health practices, we conducted a 24-hour dietary recall of mothers and their oldest children under 59 months in three seasons throughout 2017 in West Kalimantan and in one season in Papua. We documented the source of each food/ingredient consumed so that we could determine the relative contributions of forests, fallows, farms, and markets to local diets across the two groups. We collected anthropometry data for mothers and children to measure nutritional status and anemia rates.

Findings:  While overall dietary diversity at the food group level was very similar among the Bornean households across the two groups, there was variation in individual foods consumed, especially in the dry season. Most notably, while chicken was the dominant flesh food consumed among women in oil palm households, in the dry season, women in the traditional households ate quite a lot of wild meat. The average quantity of meat consumed, however, was not statistically different across the two groups. The biggest difference in terms of food group consumption was in the consumption of leafy green vegetables: women in swidden households consumed more leafy greens than their commercialized counterparts. Children in oil palm households had higher rates of wasting in two seasons than their traditional counterparts. In Papua, dietary diversity was similar across the two groups, but the composition shows different dietary patterns. Those in traditional households ate more tubers (sago), fish, and fruits than those in oil palm households, while those in oil palm households ate more cereals (rice), vegetables, and highly processed foods. None of the measures of nutritional status were statistically different, however, anemia rates for women in the oil palm communities were significantly higher than those in traditional households.

Conclusions:  A transition from traditional agriculture to modern cash crop systems is advocated by development professionals, conservationists, governments, as well as donors. Indeed, this is one of the key changes that defines economic development. The results presented here as well as from other studies suggest that there may be negative consequences of these transitions in terms of diet and health. Food and nutrition researchers can help to design policies to ensure that diets in transitioning communities do not worsen when new livelihood opportunities are embraced.

A participatory agroecological intervention decreases depression amongst female smallholder farmers in Singida, Tanzania

Hollyn Cetrone, Northwestern University, USA

Introduction:  In 2015, depressive disorders led to over 50 million Disability-Adjusted Life Years (DALYs) lost globally, with more than 80% occurring in low- and middle-income countries. Depressive disorders are also risk factors for a number of adverse maternal and child health outcomes. Nutrition-sensitive agriculture interventions have been posited to decrease depression, however this relationship has never been quantified. Therefore, we aimed to: [1] quantify the impact of a participatory agroecological intervention on depression amongst smallholder farmers, and [2] investigate the role of food security, social support, and gender equity as mediators of that impact.

Methods: The Singida Nutrition and Agroecology Project (SNAP-Tz) is a randomized effectiveness trial of a participatory agroecological intervention aiming to improve sustainable agriculture through legume intensification, nutrition, and gender equity lessons in Singida, Tanzania. Food-insecure smallholder farming households with children less than 1 year old were recruited (n=596) in January 2016. Data are from annual surveys conducted in February 2016 and 2018, which included questions on depression, household food security, domestic violence experience (DVE), husband help, and social support. Depression was measured using the Center for Epidemiological Studies Depression Scale (range: 0-60). Probable depression is indicated by the East African-specific cut-off of >179. [1] To quantify SNAP-Tz’s impact on depression, we estimated intention-to-treat using a difference-in-difference (DID) estimate on women’s depression scores. [2] For pathway identification, mediation analysis is done with change in household food security, social support, and gender equity (operationalized as DVE and husband help). All analyses used STATA15, and standard error was corrected for clustering at village level.

Findings:  At baseline, SNAP-Tz female participants depression score were 16.9±10.8, which means that nearly half were at risk of probable depression (43.46%). After two years of interventions, SNAP-Tz decreased that score (β=-2.56, p<0.05), which means that SNAP-Tz women were at an additional 11% decrease for being at risk of probable depression compared to women in the delayed intervention group (p<0.05). Change in food security and domestic violence experience mediates -0.38 (p<0.001) and -0.92 (p<0.05), respectively, of SNAP-Tz’s impact on change in women’s depression score while change in social support and husband help scores did not. Mental health is linked to favorable nutrition and food security and high levels of gender inequity, and high workloads for rural women are linked to adverse maternal health and nutrition, so these findings are consistent with current literature surrounding the possible link between nutrition-sensitive agriculture interventions and depression. Results indicate that change in depression scores is only partially mediated by food security and DVE, so there are likely factors not included in the analysis which affect this pathway. SNAP-Tz’s participatory approach is knowledge intensive and [encourages women in problem-solving stuff] which may result in additional decrease in depression.

Conclusions:  To our knowledge, [1] this is one of the first reported impacts on decreasing risk of depression for a nutrition-sensitive agriculture intervention, and [2] the interventions seemed to operate through food security and gender equity. SNAP-Tz appears to have an important secondary impact on mental health, and therefore future agricultural and nutrition projects should include mental health evaluations as part of their study to see if these outcomes can be generalized. Thereupon it could be concluded that nutrition-sensitive agriculture interventions have the ability to reduce the loss of quality life years for women in farming communities.

Greater production diversity, more men’s involvement in household chores, and lower women’s depression mediate improvements on child’s dietary diversity in a participatory agroecological intervention in Singida,Tanzania

Marianne Santoso, Cornell University, USA

Introduction:  Nutrition-sensitive agriculture interventions are thought to improve children’s diet by improving agricultural production, household income, and women’s empowerment. However, there are few empirical data to test the relative contribution of these pathways to explain the relationship between nutrition-sensitive agriculture interventions and child’s diet. The Singida Nutrition and Agroecology Project (SNAP-TZ; NCT02761876) is a nutrition-sensitive agriculture intervention trial that measured and improved child’s dietary diversity, production diversity, and women’s empowerment in Singida, Tanzania. We therefore investigated the mediating role of improvements in production diversity and women’s empowerment on SNAP-Tz’s impact on children’s dietary diversity.

Methods:  SNAP-Tz is a randomized effectiveness trial of a participatory agroecological intervention where one male and one female ‘mentor farmer’ were elected from the ten intervention villages (n=20) and lead their peers, rural farmers with children <1 years old at baseline (n=587), in learning and experimentation on agroecology with a focus on legume intensification, nutrition, and gender equity. Participants were surveyed about their children’s diet, production diversity (crop nutritional functional richness, range: 0-7) and women’s overall empowerment [Abbreviated Women’s Empowerment in Agriculture Index (AWEAI), range:0-1] in 2016 (baseline) and 2018. We also measured three specific aspects of women’s empowerment: women’s ability to allocate income [average of 0.5 for joint and 1 for final decision-making on WEAI income allocation questions, range: 0-1], men’s involvement in household chores [husband help on 7 activities, range: 0-7], and women’s depression [Center for Epidemiologic Studies Depression Scale, (CES-D)4, range: 0-60]. We estimated the Average Causal Mediation Effect (ACME5) of changes in production diversity and women’s empowerment measures on changes in child’s dietary diversity score (range:0-7) using Stata14, accounting for geographic clustering, baseline measures of both mediating variables and child’s dietary diversity, and social desirability bias.

Findings:  SNAP-Tz significantly improved child’s dietary diversity (β=0.53, p=0.004), production diversity (β=0.61, p<0.001), and AWEAI (β=0.18, p<0.001). SNAP-Tz also improved women’s income-allocation decision-making power (β=0.08, p<0.001), men’s involvement in household chores (β=0.53, p<0.001), and women’s mental health (β=-2.43, p<0.001). Improvements in production diversity mediated about 11% of SNAP-Tz’s impact on child’s dietary diversity (ACME: 0.057, 95%CI: 0.012-0.109), while increased AWEAI did not (ACME: 0.022, 95%CI: -0.086-0.133). Looking further into women’s empowerment mediation: increased women’s power to allocate household income did not mediate SNAP-Tz’s impact on child’s dietary diversity (ACME: 0.013, 95%CI: -0.023-0.051), while increased men’s involvement in household chores (ACME: 0.036, 95%CI: 0.002-0.078) and lower women’s depression (ACME: 0.030, 95%CI: 0.002-0.067) mediated 7% and 6% of impact on child’s dietary diversity, respectively. The surprising non-significance of AWEAI and income allocation as mediator might be because SNAP-Tz’s impact on those outcomes was too small. The small impact means women in SNAP-Tz’s households still might not have less say in allocating income. Additionally, it made our sample size underpowered for this mediation analysis. It is important, however, to consider the alternative explanation that women might also make decisions that are not optimal for child nutrition cannot be ignored.

Conclusions:  SNAP-Tz improved child’s dietary diversity through increasing agricultural production diversity, men’s involvement in household chores, and women’s mental health. Even though the mediation effect of increasing men’s involvement in household chores is half of that for production diversity, it underscores the importance of engaging men in household tasks and child care, which is often under-emphasized in many nutrition-sensitive agriculture projects. The importance of improving women’s mental health also highlights another potentially important mediator. This further understanding of the mediating role of production diversity and various aspects of women’s empowerment would allow us to optimize the design of nutrition-sensitive agriculture projects.

Exploring the cultural acceptability and sustainability of biofortification in Pakistan

Heather Ohly, University of Central Lancashire, UK

Introduction:  The BiZiFED project (Biofortified Zinc Flour to Eliminate Deficiency) aims to investigate the potential impact of biofortification as a strategy to alleviate zinc deficiency in Pakistan. The impact of biofortification is contingent on uptake, which in turn is dependent on the socio-cultural acceptability of producing and consuming biofortified flour. This mixed methods study conducted in 2018 explored the context, traditions, knowledge, and attitudes of consumers and farmers, to improve our understanding of the factors that are likely to affect the uptake of biofortified wheat.

Methods:  To obtain a consumer perspective, semi-structured interviews were conducted with household members recently involved in a randomised controlled trial (RCT). The RCT examined the effect of consuming biofortified flour on the zinc status of women living in a rural community in North-West Pakistan. A subsample of 10 (out of 50) households from the RCT was randomly selected. Five male heads of household and five female trial participants were interviewed. The interview schedule was developed collaboratively with the local research team. Questions related to participants’ experiences of using biofortified flour, comparisons with their usual flour, awareness of potential health benefits, willingness to purchase zinc-biofortified flour if available in the future, and wider community perceptions. Interview data were translated into English, transcribed verbatim and analysed thematically. In addition, a questionnaire survey was conducted to explore farmers’ views and perspectives on biofortification. In total, 66 farmers who had grown the biofortified wheat variety (Zincol) were recruited from Sindh Province. The questionnaire was developed collaboratively with the local research team. Questions related to farmers’ awareness of biofortification and their experiences of growing biofortified wheat, including fertilizer use, support/advice received, and any differences in yield and price. Descriptive analysis was completed.

Findings:  In preliminary analysis of the interview data, participants appeared to prefer the flour provided during the RCT to their usual flour purchased from the local market. They preferred the texture of the new flour (good for kneading) and the taste of the bread (sweeter) and found it lighter on the stomach. They reported a range of perceived health benefits, which they attributed to the new flour, including improved digestion and reduced heartburn, aches and pains. They said they would like to purchase biofortified flour – if the price was affordable – and would recommend it to other community members. The survey revealed that farmers were well informed about biofortification. The higher zinc content of the wheat and potential health benefits were important factors in their decision to grow Zincol. However, none of the farmers reported using zinc fertilizers on their crop despite this being recommended practice – reasons for this are unknown. All participants said they received the same price for Zincol as for other varieties of wheat, and 86% said they obtained a higher yield from Zincol. All participants said they planned to continue growing Zincol.

Conclusions:  This is the first study to explore the cultural acceptability and sustainability of biofortification in Pakistan. It offers novel insights into the attitudes and perspectives of local stakeholders and suggests that Zincol could be popular with consumers and farmers, provided it remains affordable. In-depth qualitative findings will be presented at ANH Academy Week 2019. They will include positive and negative perceptions of community members, illustrated with quotations. Study limitations include social desirability bias and small sample sizes. Further research is needed to understand how socio-cultural factors may affect the sustainable uptake of biofortified wheat in wider population groups in Pakistan.

A cluster-randomized home gardening program improves dietary diversity and food security among rural Tanzanian women

Mia Blakstad, Harvard University, USA

Introduction:  Globally, nearly two billion people suffer from micronutrient deficiencies. In Tanzania, 34% of children under five are stunted and 45% of women of reproductive age are anemic. In rural Tanzania, the average diet is monotonous, with the majority of calories consumed coming from carbohydrate staples. Vegetables and other nutrient-dense foods such as nuts and animal products are often more expensive than staple grains and can be difficult to access in areas with insufficient access to markets or trade. Nutrition-sensitive programs that combine agriculture, nutrition and behavioral change can improve both access to and demand for vegetables and other nutrient-dense foods.

Methods:  This analysis considers data from an ongoing cluster-randomized trial of homestead food production at baseline and at 12 months of follow-up (ClinicalTrials.gov NCT03311698). The primary outcome of this intervention was women’s dietary diversity, measured using a locally-adapted food frequency questionnaire, and defined as the number of food groups consumed out of 10. The study was implemented in Rufiji district of Pwani Region in Eastern Tanzania, among 1,006 women of reproductive age with at least one child of 6-36 months and with access to a plot of land or containers where vegetables could be grown. Participating households received 1) agricultural training and inputs to promote homestead food production and dietary diversity, 2) nutrition counselling and 3) other public health messages. Agricultural inputs included seeds for local crop varieties, fertilizer, and watering cans. The intervention engaged the existing community workforce of Agriculture Extension Workers (AEWs), Livestock Extension Workers (LEWs) and Community Health Workers (CHWs) to deliver the training and messages to participants during household visits and at field school sessions. Predictive models were built using restricted maximum likelihood models and robust logistic regression models with random effects for cluster and fixed effects for relevant covariates.

Findings:  Controlling for baseline dietary diversity, wealth quintile, education level, and livestock ownership, women in intervention households consumed more food groups per day at 12 months of follow-up than women in control households. Among women in intervention households, the adjusted odds of consuming at least five food groups per day was twice that of women in control households. Additionally, the adjusted odds of moderate or severe food insecurity among intervention households was lower than that of control households. The positive impacts of the HFP program on food security, on the odds of consuming five or more food groups, and on the average number of food groups consumed are promising. These results were achieved over a relatively short one-year period, and stronger positive effects in the long term are plausible as the participants gain knowledge, confidence, and skills.

Conclusions:  This cluster-randomized home garden trial explored the feasibility of incorporating existing cadres of agricultural extension workers and community health workers to improve the diets of women in rural Tanzania. Our analysis confirmed that integrated nutrition and agricultural interventions increased consumption of nutrient-rich foods. To our knowledge, this study was the first randomized controlled homestead food production trial conducted in East Africa to find significant impacts on dietary diversity. Our results suggest that home gardening programs could yield sustainable health gains if implemented at scale.