Food environments and markets: A
byANH Academy
Open Full Event

 Intro animation:

Food environments and markets studies at ANH2020


Session recording:

ANH2020: Food environments and markets A


Speakers and presentations:

  • Session chair: Jess Fanzo, John Hopkins University
    @jessfanzo @JohnsHopkins

  • Anjali Ganpule-Rao, Centre for Chronic Disease Control, New Delhi
    Association of the food environment with overweight/obesity in Sonipat, India
    Presentation | Slides

  • Chrissie Thakwalakwa, Chancellor College, University of Malawi
    Drivers of food choice in overweight and normal weight mother-child dyads in Malawi
    Presentation | Slides

  • Giacomo Zanello, University of Reading
    @g_zanello @UniofReading
    Rural transformation and the double burden of malnutrition among rural youth in low and middle-income countries
    Presentation | Slides

  • Ramya Ambikapathi, Purdue University
    @iamrumrum @PurdueHHS
    How does the food environment influence household food purchase patterns and nutritional status? Empirical evidence from food vendor mapping in peri-urban Dar es Salaam, Tanzania.
    Presentation | Slides


Association of the food environment with overweight/obesity in Sonipat, India

Anjali Ganpule-Rao1

Nikhil Srinivasapura Venkateshmurthy2

Piyu Sharma1

Ganesh Kumar S2

Prashant Jarhyan2

Rajesh Khatkar2

Avinav Prasad Maddury2

Sailesh Mohan1,2

Dorairaj Prabhakaran1,2

1Centre for Chronic Disease Control, New Delhi
2Public Health Foundation of India, New Delhi


Food environment research has gained importance in recent years in response to the increase in the prevalence of overweight/obesity and non-communicable diseases. Most of these studies are from high-income countries while those from low and middle-income countries are limited. Available literature shows that the food environment encompasses different aspects including distribution of food outlets, which affect food acquisition, food habits and also nutritional outcomes (FAO 2016). We studied the associations between the density of healthy and unhealthy food outlets and overweight/obesity among adults residing in rapidly epidemiologically transitioning communities in Sonipat, North India.


The analysis is based on data from a representative population-based cross-sectional survey under a community intervention study entitled UDAY among 6208 participants aged ≥30 years, residing in rural and urban areas Sonipat. Participants were selected using a multistage cluster random sampling technique. Data on socio-demographics, physical activity, and consumption of fruits and vegetables were obtained using an interviewer-administered questionnaire and body size was measured by anthropometry. The location of food outlets was geo-coded using a handheld device (Garmin). Food outlets were classified as either healthy (stores selling cereals, pulses, dairy, poultry and meat, fruits & vegetables) or unhealthy food outlets (stores selling fried, sweet, salty and processed foods). The density of food outlets was calculated as the number of shops/1000 households per primary sampling unit. Access was categorized into tertiles as low, medium and high as per the following definition ≤5 shops, 6 to 45 shops and >45 shops per 1000 households respectively for healthy food outlets. Similarly, for unhealthy food outlets the definition used was 0, 1 to 8 and ≥9 shops per 1000 households. Overweight/obesity was defined as adults with a body mass index (BMI) >25kg/m2. The associations were studied using logistic regression.


Participants who had information on the food environment (n=2847) were analyzed for the current analysis. The mean age of the participants was 48.5±SD 12.3 years and 53.4% were women. Of these 10.6% (13.1% men, 8.5% women) were underweight (BMI<18.5kg/m2) and 46.7% (39.8% men, 52.7% women) were overweight/obese. Of the total, 23.4% had low, 45.2% had moderate and 31.4% had high access to healthy foods while 25.3% had low, 40.1% had moderate and 34.6% had high access to unhealthy foods. The odds of overweight/obesity among adults with low access to healthy foods was 1.5 (95%CI: 1.2 to 1.9) and 1.6 (1.3 to 1.9) for those with high access to unhealthy foods. Residing in urban location (OR=2.4, 2.0 to 2.8), being woman (OR=1.6, CI: 1.4 to 1.9), consuming sweet fruits in higher frequency (OR=1.02, 1.01 to 1.30) and belonging to the highest wealth quintile/tertile (OR=1.8, CI: 1.7 to 2.0) were associated with overweight/obesity. The association between healthy as well as unhealthy food access and overweight/obesity remained unchanged when controlled for residence (urban/rural), sex and frequency of fruit consumption. However, the association between access to unhealthy food and overweight/obesity became insignificant when further adjusted for the wealth index.


Our analysis indicates that access to foods impacts overweight/ obesity and thus highlights the role of the food environment. This warrants policy interventions to reshape food environments to promote healthier consumption and control overweight/obesity.


Food and Agriculture Organisation of the United Nations – FAO (2016) Influencing food environments for healthy diets, Rome, Available at:
World Health Organization G. WHO. Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Consultation; 1995.

Drivers of food choice in overweight and normal weight mother-child dyads in Malawi

Chrissie Thakwalakwa1

Valerie Flax2

John Phuka3

Lindsay Jaacks4

1Centre for Social Research, Chancellor College, University of Malawi, Zomba, Malawi
2RTI International, Research Triangle Park, NC, USA
3College of Medicine, University of Malawi, Blantyre, Malawi
4Harvard University, Boston, MA, USA


As the prevalence of overweight and obesity increases among mothers and children in Malawi and other sub-Saharan countries (1, 2), research is needed to understand the factors that drive their food choices. The aim of this study was to measure a variety of a priori drivers of food choice among mothers and children in dyads with different weight status and assess their relationship to macronutrient intake and to the frequency of food group consumption.


We enrolled 274 mother-child dyads in Lilongwe and Kasungu districts into three groups based on standard body size cutoffs: overweight mother, overweight child (n=74); overweight mother, normal weight child (n=120); and normal weight mother, overweight child (n=80). Mothers were 18-45 years and children were 6-59 months. Trained research assistants collected quantitative food frequencies for mothers and children, from which we derived 11 food groups (grains, roots/tubers, vegetables, fruits, meat/eggs, fish, dairy, legumes and nuts, oil/fat, snacks, and sweets), total daily energy, and percent of energy from carbohydrates, fat, and protein. We specified 20 drivers of food choice a priori, including dyad type, rural/urban residence, mother’s taste preferences, household food insecurity access scale (HFIAS) (3), female autonomy (4), mother’s body size preferences for herself and her child (5), mother and child morbidity in the past 2 weeks, household assets, monthly food expenditures, and other socioeconomic variables. We tested for differences in demographic and socioeconomic characteristics across the three types of mother-child dyads using chi-square or Kruskal-Wallis H tests. We conducted univariate analyses of predictors in relation to the macronutrients and food groups, then entered covariates associated with the outcome at p<0.10 into multivariate models and presented findings with p<0.05.


Dyads with overweight mothers had larger household sizes (p=0.03) and dyads with overweight children more frequently purchased special foods for children (p=0.02). Normal-weight mother, overweight-child dyads had the greatest food insecurity (p=0.001) and spent the least on food (p=0.02). Dyads with normal weight versus overweight children consumed fewer calories from carbohydrates (mothers p<0.001, children p<0.01), fewer grams/day of grains (mothers p<0.001), more calories from fat (both p<0.001), and more grams/day of roots/tubers (children p=0.03) and oil/fat (children p=0.01). Household assets positively predicted mothers’ calories and mothers’ and children’s meat/eggs and oil/fat consumption. Higher HFIAS negatively predicted mothers’ and children’s snack consumption and mothers’ calories and fruit consumption. Taste preferences positively predicted mothers’ and children’s calories and grain and vegetable consumption, and children’ legume/nut and fish consumption. Purchase of special foods for children and amount spent on the special foods positively predicted mothers’ snack and children’s sweet consumption and negatively predicted children’s vegetable consumption. Amount spent on household food positively predicted children’s dairy consumption. Mothers’ preference for a normal body size negatively predicted mothers’ grain consumption and the perception that an overweight/obese body size was healthy for a child negatively predicted children’s dairy consumption and positively predicted children’s sweet consumption.


Our findings show that dietary intake and food group consumption differed in dyads with a normal weight versus overweight child. Dyads with higher socioeconomic status and discretionary funds for special foods for children consumed more calories, meat/eggs, dairy, oil/fat, snacks, and sweets, whereas households with greater food insecurity consumed fewer snacks and fruits. Maternal taste preferences and body size preferences were related to some food groups, but female autonomy and mother/child morbidity were not consistent predictors of food choice. These findings can be used to tailor messages promoting a healthy diet in the context of the dual burden in Malawi.


Jaacks LM, Slining MM, Popkin BM. Recent underweight and overweight trends by rural-urban residence among women in low- and middle-income countries. J Nutr. 2015;145:352-7.
Tzioumis E, Kay MC, Bentley ME, Adair LS. Prevalence and trends in the childhood dual burden of malnutrition in low- and middle-income countries, 1990-2012. Public Health Nutr. 2016;19:1375-88.
Coates J, Swindale A, Bilinksy P. Household food insecurity access scale (HFIAS) for measurement of food access: indicator guide, version 3. Washington, DC: FHI 360, Food and Nutrition Technical Assistance III Project; 2007.
Agarwala R, Lynch SM. Refining the measurement of women's autonomy: an international application of a multi-dimensional construct. Social Forces. 2006;84:2069-90.
Croffut SE, Hamela G, Mofolo I, Maman S, Hosseinipour MC, Hoffman IF, et al. HIV-positive Malawian women with young children prefer overweight body sizes and link underweight body size with inability to exclusively breastfeed. Matern Chi Nutr. 2018;14:12446.

Rural transformation and the double burden of malnutrition among rural youth in low and middle-income countries

Giacomo Zanello1

Elisabetta Aurino2

C.S. Srinivasan1

Cristina Cirillo3

Kelvin Balcombe1

Suneetha Kadiyala4

1University of Reading

2Imperial College London

3UNICEF Office of Research

4London School of Hygiene and Tropical Medicine


After years of neglect, nutrition among young people is now firmly on the international health and development agendas (1, 2, 3). However, little is known about the nutritional behaviours and outcomes among youth in rural areas. Even less is currently known about the way in which economic and social changes in rural areas are contributing to shape nutrition among youth (4). Based on these premises, we examine the association between the trends in rural transformation and emergence of double-burden of malnutrition, with a focus on thinness and overweight/obesity among adolescents and youth in LMICs.


We combined historical data series from the NCD-RisC collaboration on nutritional status of adolescents and young people (10-19 years old) between 1976 to 2016 with data from FAOSTAT to provide proxy measures of rural transformation and structural shifts in dietary patterns at the population level. The full dataset includes 72 LMICs. We examine global historical trends in nutritional status indicators and associate them with trends in rural transformation variables using Preston curves. Following the methodology used in (5), using a non-parametric approach Preston curves allow us to statistically capture the empirical cross-sectional relationship between the rural transformation indicator of interest and the outcome variables (nutritional status indicators) for different time periods (1976 - 2016). The estimates are computed using Epanechnikov kernel local polynomial regressions (6). Long-term trends in adolescent nutritional status are examined alongside variables capturing rural transformation and diet. Agriculture value added per-worker (constant 2010 US$) was used to capture rural transformation. We report five sets of results. We first explore the relationship between level and speed of rural transformation and nutritional status of adolescents. We then split the analysis by sex, age groups, and geographical regions. We finally look at the relationships between rural transformation and dietary indicators.


Our study is the first one to track four decades of changes in BMI of youth and assess the correlations of these changes with rural transformation processes. Our analysis shows that indicators of rural transformation have important implications for the dual burden: at lower levels of rural transformation, countries with fastest speed of rural transformation saw steepest increases in overweight and obesity, especially among younger age groups. Trends seem to suggest that more developed economies tend to be correlated with a shift from normal weight to overweight. Countries with low level of rural transformation but transforming at fast speed are also experiencing rapid shifts in foods available, with larger increase of fat and food supply and sugar intakes.


Given the positive and negative impacts of rural transformation processes on the nutrition transitions of rural youth, policies need to focus on making these processes youth and nutrition sensitive. This means focusing not only on food supply and food trade, but how and what food is distributed and marketed and by whom, shaping food environments and thus food acquisition patterns of various segments of the population. In areas where investment by the agri-food industry is seen to be having deleterious effects of rural youth nutrition and health, strong public interest regulation may be needed.


Bundy, D. A. P., Silva, N. de, Horton, S., Patton, G. C., Schultz, L., Jamison, D. T., … Al., E. (2017). Investment in child and adolescent health and development: Key messages from Disease Control Priorities. The Lancet, 2423–2478.

Patton, G. C., Sawyer, S. M., Santelli, J. S., Ross, D. A., Afifi, R., Allen, N. B., … Viner, R. M. (2016). Our future: A Lancet commission on adolescent health and wellbeing. The Lancet, 387(10036), 2423–2478.

Christian, P., & Smith, E. R. (2018). Adolescent Undernutrition: Global Burden, Physiology, and Nutritional Risks. Annals of Nutrition and Metabolism, 72(4), 316–328. 

Vargas-Lundius, R., & Suttie, D. (2014). Investing in young rural people for sustainable and equitable development. Rome (Italy).

Masters, W. A., Rosenblum, N. Z., & Alemu, R. G. (2018). Agricultural transformation, nutrition transition and food policy in Africa: Preston Curves reveal new stylised facts. Journal of Development Studies, 54(5), 788-802. [6] Henderson, D. J., & Parmeter, C. F. (2015). Applied nonparametric econometrics. Cambridge University Press.

How does the food environment influence household food purchase patterns and nutritional status? Empirical evidence from food vendor mapping in peri-urban Dar es Salaam, Tanzania.

Ramya Ambikapathi1,2

Gerald Shively3

Germana Leyna4

Dominic Mosha5

Alli Mangana6

Crystal Patil7

Morgan Boncyk1

Savannah Froese1

Cristiana Edwards1

Patrick Kazonda6

Japhet Killewo6

Mary Mwanyika-Sando5

Nilupa S. Gunaratna2

1Department of Nutrition Science, Purdue University, USA

2Department of Public Health, Purdue University, USA

3Department of Agricultural Economics & International Programs in Agriculture, Purdue University, USA

4Tanzania Food and Nutrition Centre, Tanzania

5African Academy of Public Health, Tanzania

6Department of Epidemiology and Biostatistics, Muhimbili University of Health and Allied Sciences

7Department of Women, Children & Family Health Science, University of Illinois at Chicago, USA


We study the relationship between the local food environment (FE) and the food consumption patterns and nutritional status of adults in peri-urban Tanzania. In Africa, the food environment contains a high density of informal vendors, which creates challenges to characterize the FE. We present a protocol and electronic tool developed as part of the Diet, Environment, and Choices of positive living (DECIDE) study. We mapped food vendors in a peri-urban settlement of Dar es Salaam and compare several FE metrics, including new indicators, dispersion and dominance, to better understand drivers of household food purchases and nutrition outcomes.


The DECIDE study is an observational cohort study that aims to characterize food choice and environment among families with persons living with human immunodeficiency virus (PLHIV) using qualitative, geo-spatial, and quantitative methods. To conduct a census of informal and formal food vendors, we used a participatory research approach with enumerators residing in the study area. Using a photo transect survey and group discussion, we developed a new tablet-based tool to capture georeferenced food vendor typology (formal – restaurants, shops; semi-formal – umbrella vendors; informal - hawkers), gender, and availability of specific foods and food groups from 6,627 food vendors. Each food vendor questionnaire took 1-2 minutes. We further collected food purchase patterns and sourcing of 49 foods in the last 7 days, dietary intake, body mass index (BMI), and waist to hip ratios (WHR) from adults in 232 PLHIV households in the study area. We compared FE metrics such as density (count), distance, and diversity (measure of evenness and richness) of food vendor typology, as well as new metrics, i.e., dominance (extent to which few food vendor types dominate the landscape) and dispersion (spatial evenness), borrowed from landscape ecology in relation to household food purchase patterns and nutritional status.


Out of 6,627 food vendors, 39% were formal (301 restaurants & 2289 shops), 44% semi-formal (1526 semi-structure food vendors, 1526 umbrella food vendors), and 17% informal (mobile hawkers). Among all food vendors, 40% sold cooked foods. Fruits and vegetables were primarily sold through informal and semi-formal vendors, while meat, grains, and legumes were sold through formal vendors. 73% female with a median age of 40 (Interquartile range [IQR] 33, 47) and median BMI of 23 (IQR 21,27) and WHR of 0.85 (IQR 0.8,0.9). Typical households had purchased three different vegetables in the last seven days (IQR 0,5). Household purchase of rice, sugar, unpolished maize, tomato sauce, cassava, meat and fish were positively associated with BMI and WHR, while purchase of watermelon, yam, bananas, green beans, and palm oil were negatively correlated. Household had a median density of 90 food vendors within 500 meters (IQR: 70-131); of these only 17% offered fresh produce. Despite high correlation among food environment metrics, diversity and dominance showed the only significant associations with household food purchase patterns. Vegetable purchases were negatively correlated with higher diversity of food vendors (r= -0.15, p-value<0.0001), but positively corelated with dominance of one type of food vendor (r=0.13, p-value=0.0004).


Economic, nutrition, demographic, and agricultural transitions are occurring rapidly in Africa. Despite the growth of supermarkets, informal food vendors play an important role in the peri-urban FE. Here, we illustrate that people source fruits and vegetables from informal food vendors. We compared several FE metrics: diversity of food vendor typology and dominance of one type of vendor were significantly correlated with vegetable purchase patterns. Further spatial analysis with the type of foods will identify pathways for healthy and unhealthy food consumption. FE interventions should acknowledge the role informal vendors can play in the consumption of healthy foods.


Leyna, G.H., et al., Profile: The Dar Es Salaam Health and Demographic Surveillance System (Dar es Salaam HDSS). Int J Epidemiol, 2017. 46(3): p. 801-808.

Ambikapathi, R., et al., Food purchase patterns indicative of household food access insecurity, children's dietary diversity and intake, and nutritional status using a newly developed and validated tool in the Peruvian Amazon. Food Secur, 2018. 10(4): p. 999-1011.

O'Neill, R.V., et al., Indices of landscape pattern. Landscape Ecology, 1988. 1(3): p. 153-162.

Turner, C., et al., Food Environment Research in Low- and Middle-Income Countries: A Systematic Scoping Review. Adv Nutr, 2019.

Reardon, T., et al., Rapid transformation of food systems in developing regions: Highlighting the role of agricultural research & innovations. Agricultural Systems, 2019. 172: p. 47-59.

Login or Join ANH Academy to be part of the conversation