Nutrition Assessments

Prior to conducting any surveys and assessments, a one-day training course was held in Malawi on November 16, 2016 for key project partners who were going to supervise data collection. As part of the training, the purpose of the survey, sample size calculation, survey objects and household questionnaire were reviewed and discussed. Responsibilities for enumerator training, field supervision, selection of surveyors and timeline were also finalised. Following which enumerator training in both Malawi (3-7 December, 2016) and Zambia (8-10 December, 2016), were conducted.

Using the Individual Dietary Diversity Score (IDDS) questionnaire, baseline data was subsequently collected, cleaned and basic descriptive analysis was extracted from it and most crucially, the food items were categorised into food groups for the dietary intake.

The results of the baseline IDDS surveys showed that on average, 3.42 food groups were consumed by women of reproductive age (WRA) during the day before the interview.

Furthermore starchy stable foods were consumed by 99% of the women while 87% consumed dark leafy greens and only 28% consumed meat, poultry and/or fish. In addition, less than 1% of women ie., 2 women consumed dairy products while only 2% had eggs.

In summary only 18% of women achieved the threshold for consuming 5 or more out of 10 IDDS food groups the day before the interview. Therefore, 79% of the women were more likely to have inadequate micronutrient intakes as they consumed fewer than 5 food groups.

To understand the food an nutrition security situation in both Malawi and Zambia project locations, an agrobiodiversity assessment was essential to determine what foods were locally available. To complement the agrobiodiversity assessment, a seasonal food availability assessment was also performed which helped determine the use of local foods in local production systems, markets and diets to support objectives to expand production and consumptions of nutritious food. Subsequently it was possible to map the availability of foods within essential food groups from January to December. This proved to be extremely helpful in not only planning the interventions in the value chains of interest but also in developing useful nutrition education messaging which would help households to diversify their diets.

In each project site, two focus group discussions (FDGs) were conducted for a total of four FGDs with approximately 10 targeted participants per group for a total sample size of 40 FDG participants (both men and women).

To elicit information from different subsections of the community, a cross-section of individuals in agricultural production representing different age groups were included in the FDGs. These two tables show the findings from the FDGs conducted.

A methodology to assess no/low/medium/high availability of all reported foods across each calendar month was utilised. As evidenced from the tables above, it was found that the typical Malawian and Zambian diets consisted mainly of large portions of energy-dense foods with a low diversity in foods consumed especially when it came to protective and body-building foods (two of the three food group guidelines in Zambia).

To better understand daily operations of the selected value chains, value chain assessments were conducted with the main of objective to evaluate value chain dynamics from harvest to consumption while identifying bottlenecks and constraints.

Current agro value chain assessments are based on production and economic parameters and do not consider factors that are relevant in revealing the nutritional, environment or safety aspects of the chain. Subsequently, the project found that the typical agro value assessment inadequate in providing a clearer picture of the current state of the value chain. Therefore, the value assessment was redefined to include other aspects that were more relevant to building the capacity of stakeholders in the project. A broader value chain assessment model was adopted which included activity and process analysis, measurement of processing parameters, material resource use (food and water), cost of activities and drivers, energy consumption and waste generation and consumer purchasing decision evaluation. The following four new aspects were integrated into the current value chain assessment:

• Knowledge assessment
• Nutritional value/assessment of products and losses along the chain
• Techno-environmental assessment of current practices
• Product and process safety along the chain

Redefined value chain assessment model

Optimising agri-food value chains is an essential part when addressing food security issues since they are closely linked to achieving human satisfaction. The project developed a methodological approach using consumer-focused indicators to assess a consumer- based value chain and its correlation to food security.

The framework analysed the following broad dimensions (made up of indicators) for different stages along the consumer-based value chain: social, environmental, economic, operational, quality, perception and attitude, agility, governance and management dimensions.

Because agriculture has been put at the forefront when dealing with food and nutrition issues, the project sought to assess the performance of the bean value chain in meeting consumer preferences within the context of food security. A multidimensional performance index was developed and applied when the performance of the common bean value chain was assessed as well as its structure and dynamics. The findings showed that bean production was characterised by:

• The use of primitive tools and recycled seeds
• Manual and time-consuming activities
• Inefficient storage
• Threshing
• Loss management techniques

For these reasons there were low bean yields, low quality of bean products and losses incurred. The common bean value chain poses strong buyer power, minimal supplier power, a considerable threat to entry, intense rivalry among actors and a minimal presence of substitutes. Knowledge transfer was largely informal and unidirectional as well as internal and mainly centred on farming practices. Results revealed that agility along the chain was very low because value chain actors did not have the necessary assets to respond adequately and quickly to the dynamic environment within which they operate. The quality assessment revealed threshing, sorting and storage conditions were suboptimal and led to a lack of homogenous and clean beans, broken beans, darkening beans, increased cooking time, reduced shine and damaged beans.

Consumers’ preferences and needs were not found to be adequately met because performance assessment revealed low scores in food security indicators. The common beans value chain scored below average on all indicators and these low performance scores can be attributed to inefficient performance and management of activities along the value chain, low stake holder involvement, lack of financial and technical capacity, low trust and lack of value creation opportunities.

The bean value chain obtained lower scores for agility, management and a higher score for the economic dimension. Further analysis revealed correlations between dimension and food security indicators while cluster analysis showed similarities among value chain actors based on performance scores with majority of actors found within the cluster characterised by higher scores in affordability and accessibility. Through the project, product and process quality improvement mechanisms were applied to strengthen the capacity of value chain actors to produce optimum amounts of quality beans that would meet consumer preferences.

Common Bean Value Chain Map

After the focus group discussions held in the first year of the project it was decided that trained enumerators would be used to conduct dietary monitoring in both Zambia and Malawi. Food lists were generated using focus group data from each target area were then adapted to fit the local context in Zambia and Malawi using local names for each food. These food lists were then separated into the ten food groups which are used to calculate the Minimum Dietary Diversity for Women of Reproductive Age (MDD-W) and IDDS. The list-based MDD-W method was used to provide a template for enumerators to probe respondents for foods which they consumed in the previous 24 hours.

After the focus group discussion held in the first year of the project it was decided that trained enumerators would be used to conduct dietary monitoring in both Zambia and Malawi. Food lists were generated using focus group data from each target area were then adapted to fit the local context in Zambia and Malawi using local names for each food. These food lists were then separated into the ten food groups which are used to calculate the Minimum Dietary Diversity for Women of Reproductive Age (MDD-W) and IDDS. The list-based MDD-W method was used to provide a template for enumerators to probe respondents for foods which they consumed in the previous 24 hours.

While the first part of the dietary monitoring focused on gathering data on foods that respondents consumed in the previous 24 hours, the second part of the dietary monitoring tool comprised of three questions each targeting participants’ consumption of the beans, dark leafy vegetables and fish food groups. The final part of the dietary monitoring survey focused on Knowledge, Attitudes and Practices (KAP) around food preparation and processing. These questions aimed to measure the level of learning as well as its application in practice in relation to behaviours for improving the nutritional content of family foods. For each participant, 11 rounds of data were collected throughout various times in the year from September 2017 to May 2019 in both Malawi and Zambia.

Results in Malawi showed that 100% of the participants responded that they ate a Staple in the previous 24-hour period. Thus, the most consumed food groups were Staples, Dark Green Leafy Vegetables (DGLV) and Other Vegetables across all rounds. The DGLV variety varied across each round of dietary monitoring in Malawi with rape, mustard, bean leaves and pumpkin leaves most commonly reported. In the Other Vegetables category, tomatoes were reported by the majority of respondents in each round while onions were reported highly in most rounds.

Similarly, in Zambia in all rounds, all participants reported that they consumed Staples in the previous 24 hours. In addition, DGLV were consumed by all responded in most in Zambia however, in January, May and July of 2018, a few respondents did not report consuming DGLV. Other vegetables were consumed by 30-71% of respondents across the 11 rounds of dietary monitoring of which tomato was most common food in the Other Vegetable food group consumed in the dry months. Mushrooms however, were the most common Other Vegetable consumed in the rainy season (November-January) and onions were reported to be consumed in the hot season (September-October).

Following the difficulties and costs incurred while collecting nutrition monitoring data, the project developed a mobile tool to help collect nutrition data. This concept of dietary intake assessment and nutritional status evaluation was based on artificial intelligence (AI) tools such as computer vision and machine learning which used a combination of techniques like preprocessing, feature extraction, classification and nutrient profiling. This allowed for the estimation of the nutrient factors from users’ meals.

In its application, the user captures an image of the food before consumption using a smart mobile device. The developed classification model then identifies the class of the food in the image as well as its nutrient composition. The nutritional quality of the food is then predicted using SAIN-LIM nutrient profiling algorithm and data obtained from dietary composition databases. The results showed that assessing and predicting the nutritional factor of a user’s mean from its image would in addition help monitor household nutrition, could improve and facilitate proper control of dietary intake and provide overall maintenance of balanced health conditions.

With the support of local partners, households were mapped out and central locations were chosen in five different areas per country to hold nutrition education large adult learning events (200-300 people). For these events, key message cards were developed based on desk review of nutrition education programs and materials in Zambia and Malawi. To begin with, 16 key message cards were designed and tested in Chitipa District (Malawi) for understanding as baseline survey showed that much of its population was illiterate.

Ultimately, 2500 key message cards were printed, laminated and bound for distribution to target beneficiaries and stakeholders. During large adult learning events, key message cards were distributed to households and extension officers aided in tracking which household received cards. Feedback received from participants and observers were very positive and indicated that this adult learning method was very successful and well appreciated by the community members.