Session 7A: Tools and methods for solving agriculture, nutrition, and health challenges
byANH Academy
Academy Week Research Conference
| Agriculture, Nutrition, Public Health
Date and Time
From: 28 June 2019, 11:10
To: 28 June 2019, 13:15
BST British Summer Time GMT+1:00
Location
Country: India
Open Full Event

 

Seven, 10-minute abstract-driven presentations.  

Speakers and Presentations:

 

  • Chair: William Masters, Friedman School of Nutrition Science and Policy, Tufts University
  • Ibukun Owoputi, Cornell University
    He said, she said: Using pile sort methods to explore differences in decision-making and resource allocation for food, agriculture, and other costs among couples in Tanzania
    Slides

  • Frances Knight Center for International Forestry Research (CIFOR)
    Agrifood: A new modelling tool to inform decision making in agriculture-nutrition programming by rapidly examining the potential nutritional, agricultural, environmental and social implications of promoting alternative sets of food-based
    Slides / Recording

  • Gregory Cooper, SOAS University of London
    Using spatial group model building approaches to identify food system challenges, policy levers and sustainable evolutionary pathways in Bihar, India
    Slides/ Recording

  • Amit Kumar Barui, Cornell University
    Smartphone based point-of-use determination of aflatoxin in peanuts to ensure safety
    Slides

  • Lukas Pawera, Alliance of Bioversity International and CIAT
    Developing new quantitative indices for assessing importance, underutilization and potential of edible species for dietary diversity
    Slides

  • Marianne Santoso, Cornell University
    The Household Water Insecurity Experiences (HWISE) Scale: Development, validation, and implementation of a household water insecurity measure for low- and middle- income countries
    Slides/ Recording

  • Jan Priebe, National Resource Institute, University of Greenwich
    Evaluating interactive voice response (IVR) surveys for measuring dietary intake and time use
    Slides/ Recording

 

Abstracts:

He said, she said: Using pile sort methods to explore differences in decision-making and resource allocation for food, agriculture, and other costs among couples in Tanzania

Ibukun Owoputi, Cornell University

Introduction:  Interventions aimed at improving maternal and child nutrition should reflect an understanding of the family context. Household dynamics can have a large influence on individuals’ ability to benefit from interventions designed to improve health. Spouses may disagree about who makes decisions or controls assets and income within a household. Understanding these discrepancies can help uncover why many programs targeting women do not achieve their full impact. In two regions of Tanzania, we used pile sorts—a qualitative research method—to understand differences in spending, allocation of resources, and decision-making among couples in rural farming households.

Methods:  Couples who were pregnant or had a child <24 months old were purposefully selected based on demographic characteristics (e.g. age, education, and polygamous marriage). Men and women completed the pile sorts separately. Participants received artificial money (30,000 TSH, equivalent to 13 USD) and were asked to allocate this money in three rounds. Participants sorted the money into different piles for expenses related to food, education, household expenses, agriculture, livestock, healthcare, leisure, loans, alcohol, business, savings, and “other.” In round one, participants were asked how they and their partners would normally allocate this money together if they were to receive it as an additional amount. In round two, participants were asked to allocate this money by themselves as if their partner was not there. In round three, participants were asked how they thought their partner would allocate the money. In each round, we probed for participants’ allocation within each category (e.g., types of foods, or special foods for babies). Participants were also asked how these types of decisions are normally made in the household, which round was similar to how they actually allocate resources, and if any other household members normally participate in these decisions.

Findings:  Individuals in the same household had their own preferences/priorities that differed from their spouse. Women often had control over specific resources, such as growing certain crops. Initially, most men and women often reported that joint decision-making was typical for household expenses (round one). However, on further investigation women frequently reported that their husbands had the final say over total allocation of income. Domestic violence was a pervasive issue, as women often reported that they would go along with their husbands’ decisions to avoid arguing or being beaten and some women did not feel that they actually had any say in household spending. Women frequently reported that they did not know what their husbands spent household money on, but suspected that it was alcohol or other leisure activities. Polygamous couples often disagreed on spending on basic household items. Almost all men and women reported that, jointly or alone, they would allocate money for business, savings, and/or for emergency use if children or other family members fell ill. In addition, both men and women reported they would hide extra money given to them from their spouse so that they could decide alone how to spend the money.

Conclusions:  Individuals within a household can have decision-making power over different domains of the household, and these differences may affect health and nutrition outcomes. Understanding household power relationships is important for improving intervention design, implementation, and effectiveness. Exploration of how intra-household decision-making affects uptake of recommended health behaviors is key to understanding family response to interventions. Although women and men may report joint decision-making, actual practices differ due to household power structures and domestic violence. Discordance among couples can affect the success of nutrition and agricultural interventions, especially if couples hide money from each other and disagree over spending on food/agriculture.


Agrifood: A new modelling tool to inform decision making in agriculture-nutrition programming by rapidly examining the potential nutritional, agricultural, environmental and social implications of promoting alternative sets of food-based

Frances Knight Center for International Forestry Research (CIFOR)

Introduction:  Improving dietary quality, for women and young children, in LMICs requires interventions across health, food and agricultural systems that are designed with understanding of their intersectionality, rather than in isolation. Such interventions should address the availability of and access to a sustainable supply of nutritious foods as well as the appropriate preparation and consumption of these foods. There is a lack of user-friendly tools that can rapidly generate evidence to guide programmatic decisions that consider implications for and the impact of the local agricultural context to support nutrition.

Methods:  The Agrifood tool is based on Multi-Criteria Decision Analysis (MCDA). Users begin with a number of options, food-based recommendations or local foods. By examining their scores across nutrition, agriculture, gender and environmental indicators with user-derived rankings, users compare and prioritise these options based on their implications for nutrition and agriculture outcomes. Indicators were identified through qualitative interviews with stakeholders from the international agriculture-nutrition community (n=67). The semi-structured interviews focused on objectives used when designing nutrition-sensitive agriculture programmes and sources of data. A literature review was also undertaken to explore possible data sources and existing metrics. A shortlist of indicators was included in the beta version of the tool, which will identify, based on the local situation and user-prioritisation of indicators, optimal sets of food-based recommendations to promote in a specific agriculture-cultural context. After users select and enter input data for 6-12 possible options at the individual-level and rank indicators, the software will estimate the implications for each potential combination of options and simulate partial or whole diets. The beta software and MCDA methods will be piloted through application to a CGIAR agriculture and nutrition programme in Mozambique in February 2019 to further refine the indicators, data inputs and software tool.

Findings:  Qualitative interviews were held with agriculture-nutrition programme staff from NGOs (n=30; 42%), researchers from academia (n=25; 35%), donor organisations (n=10; 14%) and governments (n=7; 10%) in Africa, Asia, Latin America, Europe and North America. Based on these interviews, 32 individual indicators classified into 12 areas were initially identified by participants, including nutritional implications of promoting or producing particular commodities, acceptability of these commodities, access to, availability and costs of inputs, implications for time use and income, gender, production diversity, land and water availability and use, seasonality of production and resilience. Further consultations led to the development of a list of 10 criteria for each option; nutritional value of a recommended food, cost of putting a recommendation into practice, acceptability of recommended practice, cost and availability of inputs to produce a recommended food/combination of foods, labour inputs and gender implications, water and land requirements, estimated income from production of recommended commodity, seasonality of commodity and climate resilience. The findings from the pilot study and the final tool will be presented.

Conclusions:  The Agrifood tool is being designed to provide agricultural users without a nutrition background, nutrition users without an agricultural background, or groups of interdisciplinary colleagues working together a way to prioritise food combinations or sets of food-based recommendations that are most appropriate to promote in a specific agricultural-cultural context. Agrifood will allow the examination and comparison of expected risks and benefits across selected nutritional, agricultural, social, and environmental criteria with the objective of promoting actions to improve dietary quality whilst also being as feasible and acceptable as possible for local agricultural production.

 

Using spatial group model building approaches to identify food system challenges, policy levers and sustainable evolutionary pathways in Bihar, India

Gregory Cooper, SOAS University of London

Introduction:  Despite being India’s sixth largest state in terms of fruit and vegetable production, Bihar’s horticultural system is characterised by fragmented value chains, erratic environmental conditions, and a preference to supply major population centres. In turn, demand in rural areas is supressed by inaccessible markets, price inflation from commission agents and a general lack of nutritional awareness. Since January 2016, the non-governmental organisation, Digital Green, has aggregated the fruit and vegetable production of ~28,000 farmers in Bihar. However, the ability of the ‘LOOP aggregation service’ to achieve equitable livelihood and nutritional outcomes at both ends of the value chain remains uncertain.

Methods:  We apply spatial group model building (SGMB) techniques to investigate the barriers facing fruit and vegetable availability in Bihar’s nutritionally vulnerable markets. SGMB combines the concepts of system dynamics with the expert knowledge of stakeholders, aiming to build upon ‘value chain snapshots’ by capturing the feedbacks, delays and decision-making variables that drive system behaviours across time and space. Moreover, SGMB introduces a spatial dimension to traditional group modelling approaches through the use of ‘LayerStack’: a facilitation toolkit made up of a series of acetate sheets overlaying a map of the area of interest. LayerStack provides an offline geographical information system (GIS) that enables stakeholders to draw and visualise the various stocks and flows that drive value chain dynamics across time and space. We conducted four SGMB sessions in each of Bihar’s Bhojpur and Muzaffarpur districts during early 2019, involving stakeholders from agricultural production, LOOP aggregation, trading, and market retailing. Once familiarised with systems thinking and LayerStack in session 1, the stakeholders and facilitation team worked to co-develop and evaluate a systems dynamics model of the LOOP aggregation service and its associated value chain processes.

Findings:  The SGMB sessions unlocked a series of insights for the parameterisation of the system dynamics models and their underlying modules (e.g., farm production, marketing pathways and price formation). First, LayerStack mapping uncovered the spatial gradients underpinning fruit and vegetable production and consumption, including cultivation potential, market availability and population density. Second, spatial dynamics were complimented by temporal reference modes, providing time-series estimates for processes like trader availability, LOOP adoption and daily market prices. Third, the SGMB sessions provided a platform to understand the value chain deficiencies from a stakeholder perspective. For example, alongside insufficient market capacities, fruit and vegetable traders often felt locked-out of rural haats due to a lack of trust-based relationships with village retailers. Therefore, the SGMB sessions facilitated hypothetical rural supply scenarios, considering the feedbacks on local retail prices and the impacts on farmer revenues. Fourth, with the periods between SGMB sessions used to code the information into formal system models, validation was aided by stakeholders comparing the model’s structure to reality and the simulation time-series to empirical observations. We also discuss the challenges faced when organising and facilitating the SGMB sessions – helping to guide the application of SGMB to other food systems in future.

Conclusions:  In contrast to traditional modelling procedures, we found that SGMB provides an approach to incorporate stakeholder perceptions and values into formal model parameterisation. Moreover, where food system challenges are geographical, the use of LayerStack mapping helps stakeholders and modellers to speak a common language through the visualisation of stocks, flows and feedbacks. Ultimately, the SGMB approach was found to provide a more diverse range of information than available through quantitative datasets, household surveys and published literature alone, whilst ensuring that model development is iteratively evaluated by local system knowledge and expertise.

 

Smartphone based point-of-use determination of aflatoxin in peanuts to ensure safety

Amit Kumar Barui, Cornell University

Introduction:  Aflatoxins (AF) are highly toxic and carcinogenic secondary metabolites produced by Aspergillus flavus and A. parasiticus. Consumption of contaminated food cause health problems or even death. Exposure to AFs over a longer period can cause hepatocellular carcinoma. AF exposure can also cause stunting in children. Several methods have been described for the determination of AFs. These techniques require extensive sample preparation, expensive equipment and well-trained personnel. In order to pave the way for the easy, rapid and sensitive detection of aflatoxins in food we are describing here a smartphone-based detection platform for determination of AFs in peanuts to ensure safety.

Methods:  The system consists of a (i) a reusable smartphone accessory, (ii) a disposable custom test strip for aflatoxin, and (iii) a smartphone app. The test is started by placing a drop of food sample/extract on the test strip which contains all the necessary reagents in it. This follows the addition of running buffer to perform the reaction on the strip. After the reaction is complete two red colored lines are observed on the strip. The intensity of both red lines (test, control) changes with the concentration of aflatoxin present in the sample being analyzed. The smartphone/device-based camera captures an image of the test strips and performs post-processing to provide a quantitative output about the aflatoxin concentration. To analyze peanut samples, a small household coffee/spice grinder was used to grind the peanut samples before extraction of AFs. 15 mL of AF extraction solvent (Methanol:Water::80:20, 4% NaCl) was added to 5 g of grinded peanuts followed by vigorous mixing. The mix was kept undisturbed to allow layer separation. Appropriate quantity of supernatant was diluted with water (40:60) and were subject to analysis using the aforementioned system.

Findings:  In this assay, decrease in test line intensity is corelated with the increase in AFB1 concentration in the sample. Test to control line (T/C) ratio of the test strips were computed employing the algorithm developed in the lab. A calibration curve for the determination of AFB1 in spiked (156-5000 pg/mL) peanut extract was prepared. It shows a polynomial relationship between AFB1 concentrations and T/C ratio with a R2 of 0.91.

Conclusions:  The smartphone-based point-of-use system can determine aflatoxin in peanuts at levels as prescribed by various international agencies to ensure food safety. The data collected on the mobile platform can be uploaded to a database for effective monitoring and prevention of food safety violations.

 

Developing new quantitative indices for assessing importance, underutilization and potential of edible species for dietary diversity

Lukas Pawera, Alliance of Bioversity International and CIAT

Introduction:  Recently, there has been a considerable effort in developing metrics for assessing human diets yet quantifying the importance and potential of individual edible species within diets have often been overlooked. Therefore, to better understand the importance of the potential of biodiversity for improving human diets, we propose simple quantitative indices for assessing species’ contribution to dietary diversity.

Methods:  The proposed indices are inspired by quantitative ethnobotany while the data collection is aligned with the standard procedure for measuring dietary diversity. The indicators take into account edible plant parts consumed, thus highlighting the dietary potential of species which in some instances are a source of multiple food categories. The indices can be used to assess a species actual contribution to dietary diversity, theoretical maximal contribution and a level of underutilization. The indices have been tested on a sample of 100 ethnic Minang and 100 ethnic Mandailing women of reproductive age respectively, from cocoa farming households in West Sumatra, Indonesia.

Findings:  All food plant species consumed in the last 24 hours were identified and their contribution to dietary diversity estimated and compared between both ethnic groups. The most underutilized as well as the most potential species for dietary diversity were also identified by the indices. The species which are a source of foods from more than one food category (e.g. Carica papaya: dark green leafy vegetables, other vegetables, vitamin A rich fruits and vegetables) were, despite reaching the greatest potential for dietary diversity, found to be highly underutilized.

Conclusions:  The wider adoption of the indices is feasible, as the data are collected through qualitative 24-hour food recalls. However, a detailed list of all foods consumed needs to be recalled and data analysis must be performed at the species level (e.g., consumption of Carica papaya in different food categories would be aggregated and counted for the species). The proposed metrics are useful for identifying promising underutilized species by quantifying their level of “underutilization” and their potential to improve dietary diversity. The indices can be also used for project and impact monitoring, particularly for measuring changes in consumption of target species.

 

The Household Water Insecurity Experiences (HWISE) Scale: Development, validation, and implementation of a household water insecurity measure for low- and middle- income countries

Marianne Santosa, Cornell University

Introduction:  Human health, nutrition, and agriculture are predicated on water. To date, progress towards equitable and sufficient water has been primarily measured by population-level data on water availability. To understand the many ways that problems with water impact health and well-being, however, higher resolution measures that include data on availability, accessibility, sufficiency, and reliability of water are needed. Therefore, we developed the Household Water Insecurity Experiences (HWISE) Scale to measure household water insecurity in an equivalent way across disparate cultural and ecological settings.

Methods:  Cross-sectional surveys were implemented in 8,127 households across 28 sites in 23 low- and middle-income countries: 10 in sub-Saharan Africa, 9 in Latin America, and 9 in Asia. Data collected included 32 items on experiences of water insecurity in the prior month; socio-demographics; water acquisition, use, and storage; household food insecurity; and perceived stress. We retained water insecurity items that were salient and applicable across all sites. We used classical test and item response theories to assess dimensionality, reliability, equivalence, and validity.

Findings:  Twelve items about experiences of water insecurity, scored as “never”, “rarely”, “sometimes”, and “often or always”, were retained. Items showed unidimensionality in factor analyses and were reliable (Cronbach’s alpha 0.84 to 0.93). The average non-invariance rate was 0.03% (threshold < 25%), indicating equivalence of measurement and meaning of the scale across sites. Predictive, convergent, and discriminant validity were established based on associations in expected directions with data on water acquisition and use, food insecurity, and perceived stress. Using a cutoff of 12, prevalence of water insecurity ranged from 7.0% among households sampled in Morogoro, Tanzania to 89.0% among households sampled in Cartagena, Colombia.

Conclusions:  The HWISE Scale equivalently measures universal experiences of household water insecurity across low- and middle-income countries. It can be readily used to quantify the role of water insecurity in adverse health, psychosocial, and economic outcomes. It can also be used to monitor and evaluate water insecurity across time, identify vulnerable and at-risk populations for maximally effective resource allocation, and measure the effectiveness of water-related policies and interventions. A globally accepted method for measuring household water insecurity can contribute to an evidence base for clinical and public health recommendations regarding access to equitable and sufficient water.

 

Evaluating interactive voice response (IVR) surveys for measuring dietary intake and time use

Jan Priebe, Natural Resources Institute, University of Greenwich

Introduction:  Innovations in mobile communications, specifically interactive voice response (IVR) surveys, present new opportunities to measure dietary intake and time use with the potential to mitigate limitations of existing methodologies. IVR surveys have potential for this application as they minimize technical and literacy requirements on behalf of the user, making them an accessible channel in low-and-middle-income countries (LMICs) with potential to reach marginalized rural communities. Despite this potential, few studies have evaluated their use in this context. This paper aims to assess the feasibility and relative validity of IVR for estimating women’s time use and maternal/child dietary diversity.

Methods:  Field data collection was conducted from January to February 2018 in Eastern Region, Uganda. Four methods were used to assess maternal time use and dietary intake for women and their 1-year-old child: direct observations over a 15-hour period, 24-hour recalls, IVR and life-logging wearable cameras. Mothers received an IVR phone call every 4-5 hours (3 per day in total) asking pre-recorded questions regarding dietary intake and time use during the preceding period. Answers were given using the numeric keys on the phone. An extensive household survey was completed prior to the research, and questionnaires were administered after data collection to assess respondents’ experience of the ICT-based research methods. The reliability and validity of the IVR survey tool will be evaluated through comparison of its results with direct observations (the gold standard). Descriptive statistics will be used to assess feasibility of using IVR. Bland Altman (BA) will be used to assess agreement (IVR and observed) for estimating dietary diversity and time use patterns. Concordance correlation coefficients by variance components (CCC) will be used to assess agreement of individual food groups and activities, allowing identification of underlying causes of discrepancies (e.g. respondent characteristics) by including covariates.

Findings:  97% of the participants rated their experience with the IVR tool as positive and 99% would be willing to engage in a future study.  However, only 36%  of respondents completed all 3 calls. This paper will use descriptive statistics to further investigate the underlying causes of this missing data, specifically by comparing frequency of causes related to technical (e.g., unavailable network), practical (e.g., respondent busy), and capacity (e.g., inability to navigate service) issues. This analysis will be used to recommend improvements and estimate completion rates in an improved context. In January – February 2019  the validity of the IVR tool will be evaluated through comparison with the direct observations using techniques described in the ‘Methods’ section above. Using BA to compare overall scores will indicate the relative validity of using IVR to estimate common indicators for dietary diversity and time use. Using CCC to assess agreement of individual food groups and activities will highlight weaknesses of the IVR tool (e.g., unclear questions or instructions) where specific categories show lower agreement. By including covariates in the analysis, CCC will also identify respondent characteristics (e.g., age, income, familiarity with technology) that predict lower agreement of IVR with direct observations.

Conclusions:  This paper assesses whether IVR-based data collection tools are viable and valid approaches to measuring dietary diversity and time use among rural populations in low-and-middle-income countries. Preliminary results indicate that while IVR is a well-accepted approach in rural Eastern Uganda, response rates during the study were low. Through further analysis in January – February 2019, causes of this low response will be identified and measures for improvement outlined. The validity of using IVR to estimate indicators of dietary diversity and time use will also be assessed. Where limitations are found, likely causes will be identified and suggestions for improvement given.

 

Speakers:
Login or Join ANH Academy to be part of the conversation