Six sub-sessions
Chair: Christine Blake, Drivers of Food Choice Program
Speakers and presentations
Ilse de Jager, Wageningen University
Food and Nutrient Gaps of Smallholder Farming Households in Rural Northern Ghana
Francis Oduor, Bioversity International
Does Caregivers’ Nutritional Knowledge and Attitudes Mediate Seasonal Shifts in Children’s Dietary Patterns – A Case Study of Rural Young Children in Western Kenya
Michelle Holdsworth, University of Sheffield
Dietary Transitions in Ghanaian Cities: Using Innovative Methods to Map the Social and Physical Food Environments that Drive Consumption of Unhealthy Foods and Beverages, to Identify Contextually Appropriate Policies and Interventions
Alan de Brauw
Market Access and Changes in the Structure of Food Consumption: Evidence from Nigeria
Elinlaa Kivaya, Sokoine University of Agriculture and Emory University
Multiple Regression Analysis of Drivers of Food Choice in Tanzania
Carolyn Auma
What Influences (Un)Healthy and Environmentally (Un)Sustainable Dietary Practices Among Uganda Women? A Qualitative Study Using Photovoice
Q&A
Recording
Food and Nutrient Gaps of Smallholder Farming Households in Rural Northern Ghana
Ilse de Jager, Wageningen University
Introduction
Current transformations of food systems driven by climate change, urbanization, income growth, and population growth are often associated with unhealthy diets: they fail to provide sufficient, diverse, nutritious and safe food for all. Food-based dietary guidelines (FBDGs) provide guidance to policy makers, the private sector, and consumers to redesign food systems and to improve diets of vulnerable populations. As appropriate FBDGs are based on the actual dietary patterns and their costs, it is assumed that the recommended foods are available, affordable, and acceptable for the population under study.
Methods
Using quantitative dietary intake data of infants and young children in rural northern Ghana, we developed local FBDGs. We studied whether these FBDGs are supported by the diversity and quantity of a household’s food production among 329 households, identifying nutrient and food gaps at the household level. In addition, we conducted a rough analysis of food gaps at district and national level. Further, as diversifying crop production is often mentioned as a potential solution for increasing the diversity of foods available, we studied individual associations between household’s food production diversity, household’s food and nutrient coverage and child’s dietary diversity and nutrient adequacy.
Findings and Interpretations
We found that 40 percent of our subjects were stunted and their nutrient intakes were far below the required amounts: the probability of adequacy for most nutrient intakes was below 50 percent. At household level, the developed FBDGs were on average not able to sufficiently cover the household requirements for fat (60.4 percent of recommended nutrient intake (RNI)), calcium (34.3 percent), iron (60.3 percent), vitamin A (39.1 percent), vitamin B12 (2.3 percent) and vitamin C (54.6 percent). The food production of over half the households supplied insufficient calcium (75.7 percent), vitamin A (100 percent), vitamin B12 (100 percent) and vitamin C (77.5 percent) to cover their needs. Of all households, about 60 percent of the households did not produce enough to cover their optimised required amounts of grains and legumes and none of vegetables. At district and national level, grain requirements were covered at both levels (267 percent and 148 percent, respectively); legume requirements only at district level (268 percent) but not at national level (52 percent); and vegetable requirements not at both levels (2 percent and 49 percent, respectively). We found that household’s food production diversity was positively related with household’s food and nutrient coverage. However, household’s production diversity, food and nutrient coverage were not related with the child’s dietary diversity and nutrient adequacy.
Conclusions
This study shows that although local FBDGs are based on actual dietary patterns and costs, the availability of food can be a limiting factor in the ability of populations to adopt the FBDGs. Therefore, the promotion of food-based dietary guidelines through nutrition education or behavior change communications activities alone is not enough to lead to improvements in diets. Additional strategies are required such as agricultural- and market-based strategies, especially for vegetables, in combination with nutrition-specific interventions including food fortification, and home fortification options. These may offer opportunities to further facilitate adoption of recommendations and provide additional support to improve diets of vulnerable populations.
Does Caregivers’ Nutritional Knowledge and Attitudes Mediate Seasonal Shifts in Children’s Dietary Patterns – A Case Study of Rural Young Children in Western Kenya
Francis Oduor, Bioversity International
Introduction
Globally 800 million people are still hungry, and another two billion people are suffering from hidden hunger. Rural households dependent on rain-fed agriculture are worst hit. They experience variations in food and nutrient availability occasioned by seasonality of production patterns. This results in periods of nutrient abundance in plenty seasons followed by periods of nutrient inadequacies and malnutrition. This pattern contributes to a cycle of deteriorating health and nutrition status and deprives children of their ability to realize full developmental potential. This study investigates the role of caregiver’s nutritional knowledge and attitudes in mediating effects of seasonality on children’s diets.
Methods
Repeated cross-sectional surveys were conducted on 151 randomly selected households with children aged 12-23 months in the plenty and lean seasons to collect dietary data of the children and assess the caregivers’ nutritional knowledge and attitudes related to child feeding. Dietary data was collected using two non-consecutive quantitative 24-hour recalls and caregiver’s nutritional knowledge and attitudes assessed using an interviewer-administered questionnaire. Quantities of food ingredients consumed by children were converted to individual nutrient intakes using food composition tables in the Lucille for food intake software and the micronutrients of the diets assessed using the nutrient adequacy ratio (NAR) and mean adequacy ratio (MAR) of 11 micronutrients (niacin, thiamine, riboflavin, vitamin B6, vitamin B12, vitamin C, vitamin A, folate, iron, zinc, and calcium). The responses to knowledge and attitudes were assigned scores ranging from 0 to 12 for knowledge and -19 to 19 for attitudes in order to obtain scale variables. A correct response to knowledge question earned 1 score while for the attitudes a positive attitude was awarded a score of 1, negative attitude score of -1 and an ‘ok/so – so’ or no response was awarded a score of 0.
Findings and Interpretations
Sixty-five percent of the caregivers had attained a primary level education or less. There was a positive modest correlation between caregivers’ nutritional knowledge and their attitudes (r=0.3, p<0.000, α=0.01). The MAR of the children’s micronutrient intakes was 0.84 (±0.11) during the plenty season decreasing to 0.80 (±0.11) during the lean season (t150 =3.53, p<0.001). The NAR was more than 0.7 for all the nutrients during the plenty season except for zinc, iron and calcium. The minerals zinc, iron and calcium were the most problematic micronutrients with both seasons recording high prevalence of dietary inadequacies of up to 84 percent. Multivariate regression to model the effect of caregivers’ nutritional knowledge and attitude scores while controlling for the effect of socio-demographics and MAR at first season (plenty) found that caregivers’ nutritional knowledge (ß = -.007, SE=.003, p=.027, 95% CI= (-.013,-.001) ŋ2 =.033) but not attitudes (ß = -.003, SE=0.002, p=.209, 95% CI= (-.007, .002); ŋ2 =.011) had significant effects on the changes in MAR between seasons. Improving the caregivers’ knowledge score by one unit would result in a reduction of the deviation in the quality of the diet by 0.7 percent.
Conclusions
Our study confirms findings of earlier studies that indicate seasonal variations in dietary intakes of children among rural households; however, we further demonstrated that caregivers’ nutritional knowledge is an important factor that mediates seasonal changes in diets. Therefore, increasing nutrition knowledge among these households will help stabilize dietary intakes across seasons. Moreover, we recommend that programs that aim at improving dietary intakes among children should consider changing attitudes in addition to nutrition education.
Dietary Transitions in Ghanaian Cities: Using Innovative Methods to Map the Social and Physical Food Environments that Drive Consumption of Unhealthy Foods and Beverages, to Identify Contextually Appropriate Policies and Interventions
Michelle Holdsworth, University of Sheffield
Introduction
Dietary habits are transitioning in urban Africa and nutrition-related non-communicable diseases (NR-NCDs) are becoming an important public health problem, particularly amongst women. Energy-dense nutrient-poor (EDNP) food and beverage consumption is associated with increased energy consumption and lower micronutrient intake. The overconsumption of EDNP products is therefore implicated in the onset of NR-NCDs and contributes to micronutrient deficiencies. This DFC Dietary transitions in Ghana project 1) investigates the factors in the social and physical food environments that drive consumption of EDNP food and beverages in women and adolescent girls, and 2) harnesses this understanding to develop context-relevant interventions to improve diets.
Methods
Setting: Two Ghanaian cities of different demographic transition stages: Ho: population>270,000; Accra: population>1.8 million.
Study populations: n=192 women/adolescent girls throughout the reproductive life course (13-49 years). Community informants (n=24) and national stakeholders will also be interviewed 2017-2019.
Tools: Socio-ecological approach using innovative qualitative and quantitative methods. So far, we have:
- Collected dietary information on 192 participants using a qualitative 24-hour recall that accounts for participants’ social practices. Following the classification of foods as EDNP or not using a nutrient profiling tool, we will explore what drives EDNP food and beverage consumption by analysing how ‘tempo,’ ‘periodicity,’ and ‘synchronisation’ are associated with EDNP intake.
- Completed Photovoice interviews with 64 participants. These will lead to the development of community photography exhibitions in Accra/Ho (launch in June 2018).
- Mapped food environments using Geographic Information System (GIS) to assess the availability of foods/beverages and the food advertising environment in both cities.
The next phase integrates these findings on drivers into developing priorities for context-specific interventions by using innovative methods to consult with local community informants (Community Readiness Model); and national stakeholders (the Food-EPI tool across the seven domains of food environments in the INFORMAS framework for benchmarking policy in relation to healthy food environments).
Findings and Interpretations
Preliminary findings indicate that drivers of food choice act across all socio-ecological levels. Food safety, encompassing microbiological safety, food handling, toxicity from pollutants, and the sanitary environment around food outlets emerged as key drivers of food choice. The cost of food and financial access were also identified by participants, especially in Ho. GIS mapping has indicated that a diverse range of both healthy and unhealthy foods and beverages are available in both cities. Sugar sweetened beverages are widely advertised in Accra and Ho (over half in Accra and one-third in Ho of food advertisements). All drivers have been incorporated into the development of an evidence-informed conceptual framework to illustrate the social and physical environmental drivers of EDNP food and beverage consumption that have emerged, which will be further developed and validated. The framework will identify the key drivers that could be targeted so that culturally appropriate interventions can be developed. The adaptation of a nutrient profiling tool is underway to classify the 144 foods consumed as EDNP (or not) using nine nutrients to encourage and three to avoid. Forty potentially relevant policies to prevent NR-NCDs were identified in Ghana across the seven domains of food environments in the INFORMAS framework.
Conclusions
The use of these novel qualitative and quantitative methods in this context has allowed us to shed light on how food environments are shaping food choices. The triangulation of findings and building of an evidence-informed framework will allow us to identify drivers that are important and specific to African cities. Using methods that give communities a voice (PhotoVoice and the Community Readiness Model) allow unique context relevant insights. The Food-Epi assessment of the food policy environment has indicated that there is already a lot of relevant policy activity in Ghana. An important challenge is to ensure that such policy does no harm on other forms of malnutrition, therefore the double-duty nature of such actions will be assessed during the prioritization process.
Market Access and Changes in the Structure of Food Consumption: Evidence from Nigeria
Alan de Brauw
Introduction
As food systems change, diets and diet quality also change in response. This process varies substantially over space and time, depending on underlying preferences, food availability, and market integration. In Nigeria, the food system currently has clear deficiencies, with high levels of both stunting and wasting (43.6% and 10.6%, respectively) and rising obesity (as high as 22% in Lagos). In this paper, we examine how household-level food consumption changes, both with increases in consumption levels and changes in market availability, in order to understand how diets are evolving in Nigeria.
Methods
We use all six rounds of the Nigerian LSMS-ISA survey in our analysis, with surveys conducted both post-planting and post-harvest in 2010-2011, 2012-2013, and 2015-2016. The panel study included 5000 households in 494 communities across Nigeria including both urban (156) and rural (338) areas. These data allow us to isolate changes in household level food consumption both seasonally and over time. Using the household consumption module, we calculate food expenditure aggregates for broad categories including grains, fruits, vegetables, meats, pulses, dairy, and fats, incorporating both food consumed within and away from home. We can then link these expenditures to both total household consumption, as a welfare measure, and changes or perceptions in changes of market access.
We conduct both descriptive and econometric analyses. In the descriptive analysis, we examine trends in the data to understand how food consumption is evolving over time, by region and location (rural/urban) as well as by region (North/South). In the econometric analysis, we use specifications suggested by Deaton and Paxson (1998), looking at changes in welfare on food expenditures while using household level and community-survey round fixed effects and exploring heterogeneity by region, wealth, and rural areas.
Findings and Interpretations
Preliminarily, we find a number of intriguing trends in the data. Our initial analysis has largely been focused around trying to describe both trends in consumption generally and to understand how consumption both evolves over time and by region.
Many of the basic facts in the data are as expected. As consumption increased, non-food consumption increases faster than food consumption. Rural areas have higher consumption levels of grains despite lower consumption overall; and dairy is more often consumed in the North while fruits and meat are more often consumed in the South. of all other categories, whereas the North has higher consumption of grain, pulses, and dairy and the South broadly shows higher consumption of fruits and meat.
Informed by these trends, we next use the econometric approach described above to study the relationship between overall household expenditures and budget allocations to the different food expenditure categories. Somewhat surprisingly, we find as consumption increases, relative to the trend, both grain and meat consumption rise. Simultaneously, we find a decline in the share of expenditures on vegetables and oils. These responses are stronger in the North and in rural areas. Analysis incorporating changes in market access is ongoing.
Conclusions
In sum, our preliminary findings are that as households have increases in consumption, they eat more grains and meat, and reduce their consumption of vegetables; this finding appears robust to various econometric specifications. Although we are not measuring diet quality directly, correlations between household level consumption and diet quality imply that if incomes and consumption were to rise substantially, diet quality could get worse — at least on average — rather than improve. Further analysis is exploring the difference in these findings both by location – e.g., urban versus rural — and by variables that measure how market access has changed over the course of the panel, particularly in rural areas where such variation may be more important for diet quality.
Multiple Regression Analysis of Drivers of Food Choice in Tanzania
Elinlaa Kivaya, Sokoine University of Agriculture and Emory University
Introduction
Several studies have addressed the impact of seasonality on food choice in sub-Saharan Africa in terms of childhood nutrition, human capital development, and household caloric consumption. This paper seeks to examine the relationship between seasonality, affordability, and frequency of consumption employing regression analysis of Cost of the Diet data collected in two livelihood zones in Tanzania. The paper serves two purposes: 1) expand the use of the Cost of the Diet methodology for secondary analysis and 2) provide a nuanced understanding of food consumption habits in Tanzania.
Methods
The Cost of the Diet (CoD) methodology was used to collect market data and food habits from two livelihood zones identified by the Household Economy Approach (HEA): Mvomero and Tanga. Over the course of 12 months, data were collected from 11 markets, 41 traders, and five villages in Mvomero for 252 food items. In the Handeni data were collected across seven markets, 27 traders, and four villages for 252 food items. Food habits and frequency of consumption data were collected using interviews and validated through focus group discussions according to the standard CoD methodology.
Lastly, regression analysis was employed to examine the relationship between self-reported frequency of consumption and the following variables: price (per 100 grams), seasonality (market availability by season), price volatility (standard deviation of price across three seasons), and affordability (percentage of daily household income required to purchase a standard portion of each food item). According to Kaminski et al., (2016) seasonal variation can result in a 10 percent adjustment in caloric intake in poor urban households. The CoD methodology enables an assessment of collected market data disaggregated by livelihood zone, providing a nuanced understanding of the drivers of food consumption of poor rural households.
Findings and Interpretations
According to the regression analysis, annual average price and seasonality were the most statistically significant drivers of food choice for the households surveyed. Price volatility and affordability were the least statistically significant. Other factors not included in the model e.g., convenience, additional inputs needed for cooking (water, cooking fuel, etc.), taboos, taste, and personal preference, are not accounted for in the model, but also constitute significant drivers of food choice.
Affordability of each food item was calculated based on cost per standard portion (as determined by the CoD food items database) as a percentage of daily household income per family member for very poor households. This approach was replicated from a Miller et al. (2016) study that interpreted affordability as the price per recommended serving of food items. The study also found that food items that cost ≤20 percent of household daily income per family member were considered affordable. Wild foods may result in a reduction in the annual cost of the diet for poor and very poor households, particularly in areas with large concentrations of agro-pastoralists. For this reason, daily household income per family member was used to determine affordability rather than percentage of household food expenditure.
Conclusions
One of the challenges identified of assessing seasonality in markets in SSA is the collection of food price data primarily in urban locations. While urban markets may provide a closer approximation of national average prices, they may not be representative of markets accessible to poor and very poor households. By using the CoD methodology, markets were identified and surveyed where consumers are primarily comprised of the lowest two wealth groups. In addition, the food item price provided is the quantity or volume of food most frequently purchased by low-income consumers. The aim of which is to provide better information for the formulation of policies and interventions designed to target poor and very poor households.
These findings are particularly relevant for understanding consumer behavior of poor and very poor households and the consumption of specific food items. Annual price per 100 grams was found to be slightly more statistically significant as a driver of food consumption habits than seasonality. These results suggest that both annual price and market availability need to be considered when designing policies to increase the consumption of specific, nutritious foods more so than price volatility and affordability per standard portion size.
What Influences (Un)Healthy and Environmentally (Un)Sustainable Dietary Practices Among Uganda Women? A Qualitative Study Using Photovoice
Carolyn Auma
Introduction
Rapid urbanization in low- and middle-income countries is associated with changes in dietary practices, including reduced consumption of traditional, plant-based diets in favour of ‘westernized’ diets rich in animal protein, salt, sugar, saturated fat, and processed food. ‘Westernized’ diets have been associated with high greenhouse gas emissions, i.e. less environmentally sustainable and associated with an increased risk for nutrition-related non-communicable diseases. In sub-Saharan Africa women are particularly vulnerable to these dietary changes. To achieve healthy and environmentally sustainable dietary practices amidst urbanization, an understanding of what influences Ugandan women’s dietary practices is imperative.
Methods
A quota sampling method was used to recruit women (n=73) from Nakawa division, Kampala city (urban), (n=40) and Nakawuka and Bulwanyi parishes, Wakiso district (rural), (n=33) to participate in a mixed-methods study. The study participants, varying in socioeconomic status and occupation, first completed a qualitative 24-hour recall interview and then one woman was randomly chosen, from within each quota, to participate in a subsequent Photovoice exercise. Of the participants sampled for Photovoice (n=18), those who agreed to take photographs (n=14) were given digital cameras to take pictures depicting their dietary practices and food environments within a one-week period, then photographs were discussed in face-to-face interviews. Participants who opted not to take photographs (n=4) agreed to discuss their dietary practices and perceived food environments, during in-depth face-to-face interviews. Data collection took place between September 2017 and January 2018 and interviews were conducted in English, Luganda or Luo. Interviews were audio recorded, transcribed verbatim and analysed thematically. Codes used in the thematic analysis were generated inductively and deductively, i.e., based on the ecological model [8] and social practice framework. This paper will focus on the preliminary results from the Photovoice exercise.
Findings and Interpretations
Factors that influenced (un)healthy and environmentally (un)sustainable dietary practices were identified at the individual level, and in participants' social, physical and macro-level food environments.
Individual: Participants in both settings identified food waste, preference and perceptions, health, physical and sensory food attributes, food safety and hygiene, and ‘adulteration’ as influencing their dietary practices. Food cost and hunger/satisfaction were important among rural participants. Cost, ‘cooking skills’ and ‘type of food’ were hindrances towards more environmentally friendly and efficient cooking fuels, e.g., electricity versus charcoal/firewood.
Social networks: Family was a key influence on participants’ dietary practices. Participants described food choices in terms of care (e.g. plant-based foods as more healthy than animal-based foods); ‘how we grew up’ (parents, grandparents, and family); and family members' food preferences. Good customer relationships with food retailers were important.
Physical environment: Rural and urban participants expressed how important ‘home’ gardens were to produce their own food. Proximity to food retailers, type of food retailers and type of products available were important.
Macro-level environment: The media was highlighted as a mediator of information and consequently attitudes/perceptions towards dietary practices, by some higher-income participants in Kampala; whilst community interventions promoting healthy eating ‘skills’ were salient influences among rural participants.
Conclusions
The preliminary findings from this study illustrate that the dietary practices of participants in both rural and urban Uganda were influenced by factors that fell within four domains. These included the individual-level (personal need, motivation, desire), within social networks (e.g. family and friends), the physical environment (e.g. ‘home’ gardens and neighborhood/community food retailers) and the macro-level environment (e.g. social and cultural norms, media and community-based interventions). Participants’ narratives reveal how (un)sustainable and (un)healthy dietary practices (food sourcing, cooking and eating) might be perpetuated when the requisite elements in each domain come together in the practicalities of daily life and in meeting the demands of work, schooling and parenting. These findings demonstrate how the food environment in which one is situated might influence the execution of dietary practices in daily life and highlight the importance of incorporating factors across these four domains in policy documents or other recommendations aimed at promoting healthy and environmentally sustainable dietary practices in Uganda and other similar contexts.
