Nutrient Composition of Plant-Based Substitutes for Animal-Sourced Foods: Evidence From Packaged and Processed Meats in 30 Countries

Objectives

To compare the nutrient composition of plant-based (PB) to animal-sourced (AS) items sold in 30 countries, and identify associated changes in national nutrient supply.

Methods

Using Euromonitor data from 30 countries for 2015–2021 we identified the main items contributing to PB and AS processed meat sales, and used their nutrient composition (calories, total fat, saturated fat, carbohydrate, fiber, sugar, protein, and sodium) to quantify the impacts of change in sales on the nutrient density of sales in each country and in total.

Results

Since 2014 total calories from PB meat rose a total of 3.1%/yr to 9.9 million kcal in 2021, in contrast to sales of AS processed meat which rose by about 1%/yr to 30.6 million kcal. In 2021, the energy in AS processed meat was 59% fat (22% kcal from saturated fats), 28% protein and 10% carbohydrates (3% kcal from sugar), with 0.21 g fiber, and 1.07 g sodium per 100 kcal. In comparison, energy in PB meats sold in 2021 was 41% fat (8% kcal from saturated fat), 31% protein and 13% carbohydrate (2% kcal from sugar), 2.17 g fiber and 0.56 g sodium per 100 kcal. The average annual changes in sugar were + 11% in AS and −4.1% in PB; sodium + 4% in PB and −1% in AS; and saturated fat + 12.6% in PB and −0.2% in AS.

Conclusions

Isocaloric substitution of PB meat for AS processed meats would lower density by 14% for saturated fats and 18% for total fats, while raising density by 3% for both protein and carbohydrates, raising fiber intake by 1.99 g/100 kcal, and lowering sodium intake by 0.51 g/100 kcal. Differences between items, and reformulation over time of both AS and PB products would affect these results and may explain observed changes in this period. On average since 2015, the rise of PB meats has led to healthier food composition of the processed meats category regarding these nutrients of concern. Overall, PB alternatives have a preferential nutritional profile to AS processed meat.

Funding Sources

NIFA National Needs Fellowship.

Seasonality of Child Growth: High Temperatures Coincide with Growth Faltering among Young Children in Burkina Faso

Objectives

Seasonal cycles in climatic factors such as temperature and precipitation affect the drivers of child growth and contribute to seasonal fluctuations in undernutrition. Current growth seasonality models are limited by categorical definitions of seasons that rely on assumptions about their timing and fail to consider their magnitude. In this study, we disentangle the relationship between climatic factors and growth indicators, using harmonic regression models to determine how child growth is related to peaks in temperature, precipitation, and vegetation.

Methods

Longitudinal anthropometric data collected between August 2014-December 2016 from 5,039 Burkinabè children measured monthly from age 6–28 months were linked with remotely sensed daily precipitation, vegetation, and maximum air temperature. Using harmonic regression models, we compared the magnitude and timing of seasonal peaks in climatic factors and morbidity with that of nadirs in growth velocity (cm/month, kg/month).

Results

Length and weight velocity were slowest twice a year, coinciding both times with the highest temperatures, as rains start and end, and with peak fever and diarrhea incidence.

Conclusions

Our results suggest that child health and development are more affected by high temperatures than by other aspects of climatic seasonality such as rainfall. Pathogens causing diarrheal disease and fever thrive and have more opportunities to infect children while temperatures are high, and precipitation is low. Emerging shifts in climatic conditions will pose challenges to optimal growth, highlighting the importance of programmatic and policy-level changes that optimize timing of nutrition interventions and address environmental growth-limiting conditions.

Funding Sources

United States Agency for International Development.

Role of Food Processing, Cooking Fuel, and Time Use for Meal Preparation in Diet Costs and Affordability Across 177 Countries, 2011 and 2017

Objectives

Previous studies focused on the high cost of nutrient-dense foods and low cost of convenient ultra-processed foods. This study considers fuel and time costs of meal preparation, using nutritional criteria to distinguish healthy and unhealthy processing. We identify differences in retail prices by type of food processing, revealing how processing that preserves nutritional value can help reduce the total cost of healthy diets when accounting for time and fuel burdens.

Methods

We use food prices from the International Comparison Program (ICP) for 177 countries in 2011 and 2017. We class items by processing type and food group. We use OLS regression to compute price premiums by processing type, food group, region, national income and year. For diet costs, we select least-cost items in each food group by country, year, and processing type (unprocessed; processing to maintain/enhance nutritional quality; processing to diminish nutritional quality) and add non-market costs of preparing foods using fuel prices from ICP, cooking fuel data from WHO, and recipes from the Food and Nutrient Database for Dietary Studies, Global Food Matters Database and RecipeDB.

Results

Canned, frozen, dried and cured animal foods are $1.42–4.42 on average less costly per recommended daily intake than fresh meats and fish (P < 0.001), while smoked and added-salt animal foods are $4.77–4.87 more expensive (P < 0.001). Canned plant protein foods are $2.27 more expensive on average than dry legumes/nuts/seeds (P < 0.001). Dried fruits/vegetables are $2.11–2.33 less costly than fresh (P < 0.001), while added-salt vegetables are $1.51 more expensive (P < 0.001). A healthy diet is $0.59 more costly on average if made of foods processed to diminish nutritional quality vs unprocessed foods (P < 0.001), but $0.51 less if made of foods processed to preserve nutritional quality (P < 0.001). Considering only the least expensive quintile of foods in each food group/country, average diet costs are similar across processing types.

Conclusions

Processing can reduce the burden of food preparation, yet processed foods are often more expensive and some are less nutritious. Distinguishing healthy and unhealthy forms of processing can help guide policies and programs that make healthy diets available and affordable for all, accounting for the time and fuel required for meal preparation.

Funding Sources

BMGF, UKAid.

Extreme Weather Events Differentially Impact Retail Food Prices: Evidence from Early Warning Systems

Objectives

This study aims to study the sensitivity of retail food prices to a suite of extreme weather events including floods, droughts, storms, heatwaves, and coldwaves. We hypothesize that extreme weather events are associated with higher retail prices of perishable food groups due to higher transportation and storage costs.

Methods

Monthly retail prices for 2000–2021 were extracted from the FAO Global Information and Early Warning System, the USAID Famine Early Warning System Network, and the WFP Vulnerability Analysis and Mapping database. All local currency units were adjusted to 2017 USD. Foods were assigned to one of eight categories and matched to food composition tables to derive units of 2017 USD/1000 kg and 2017 USD/1000 kCal as purchased. Gridded data were spatially matched to each market location and month of price observation. Temperature anomalies were extracted from the Terraclimate dataset using thresholds of 2 SD and −2 SD to define heatwave and coldwave respectively. Tropical cyclones were extracted from the International Best Track Archive for Climate Stewardship database using a minimum of at least one Category 3 event in the month of observation. Standardized Evapotranspiration Index values of −1.5 and 1.5 were used to define Drought and Flood conditions respectively. OLS regressions with fixed effects for market-region, market-month, and market-year were implemented. All effects were studied with respect to a reference category of Breads and Cereals. Additional covariates included average temperature, precipitation, and vegetation.

Results

1,346,513 observations from 2321 markets in 71 countries were included in this analysis. Heterogenous effects are observed across extreme events and food groups. Notably, storms are associated with a 26% increase of adjusted prices for Fruits and Vegetables and a 11% increase in adjusted prices of Pulses Nuts and Seeds indicating critical supply reductions.

Conclusions

Extreme weather events differentially impact perishable and non-perishable foods. Observed effects can inform nutrition-sensitive planning for climate adaptation and mitigation.

Funding Sources

This work was partially funded by the Feed the Future Innovation Lab for Sustainable Intensification through the United States Agency for International Development under Cooperative Agreement No. 720-OAA-18-LA-003 AID-OAA-L-14–00,006.

A Novel Diet-Health Index Reveals Worsening Trends and Widening Disparities in the Health Impacts of Diet in the United States

Objectives

Poor diets are associated with poor health outcomes, but existing metrics of diet quality do not directly include the health effects of diet. Using a novel international diet-health index (IDHI), we can measure diet-related health impacts from multiple dietary factors simultaneously, given a population's health status and most prevalent causes of death and disability.

Methods

We obtained individual-level data on intake of 12 dietary factors and exposure to 2 metabolic risk factors from the National Health and Nutrition Examination Survey (NHANES), 2003–2014, and computed the IDHI for 12 cardio-metabolic diseases and 15 cancers in the U.S. by sex, race, education, and income. We then compared IDHI to a modified Alternative Healthy Eating Index (mAHEI) using 10 of the 12 dietary factors and validated the indices using the National Center for Health Statistics (NCHS) linked dataset for total mortality through 2015.

Results

IDHI declined from −0.314 (95% CI: −0.323, −0.305) in 2003/04 to −0.325 (−0.334, −0.316) in 2013/14 (P = 0.007 for trend). Non-Hispanic Black Americans have persistently lower IDHI than other groups, and disparities in IDHI have widened over time by level of income (P = 0.004 for interaction), and education (P = 0.047 for interaction). IDHI was more closely correlated with the mAHEI at higher levels of diet quality, and both indices were strongly associated with total mortality.

Conclusions

The IDHI is a valid tool for measuring diet-related health impacts in the context of a population's most prevalent diseases, potentially offering tailored guidance regarding how best to reduce diet-related health disparities.

Funding Sources

No funding.

What Matters Most to Consumers in Peri-Urban Viet Nam? An Exploratory Mixed Methods Study Investigating Food Preferences and Values

Objectives

Understanding what drives consumer food choice in low- and middle-income countries with rapidly changing food environments is important in order to inform better interventions and food system policies. This study quantitatively and qualitatively assessed preferences for different food quality attributes including nutrition, energy, food safety (immediate), food safety (future), naturalness, taste, convenience, appearance, tradition, modernity/novelty, and price.

Methods

The exploratory mixed method study consisted of 264 randomly selected women aged 18–49 who participated in the quantitative survey. After participating in the quantitative survey, a subset of 40 respondents were randomly selected to attend focus group discussions on food preferences, food environments, and shopping behaviors. Food preferences and values were quantified using Best-Worst Scaling questions to elicit priorities and trade-offs among food quality attributes across four different food groups: leafy green vegetables, fruits, prepared foods, and snack foods (salty and sweet). All data collection took place in Thanh Oai District in Hanoi Province, Viet Nam.

Results

Food groups differ in the relative importance of each food quality attribute. Nutrition and food safety related attributes (naturalness, immediate and future food safety) were the most important for the green leafy vegetable and fruit food groups. While convenience and taste were most important for prepared foods and snack foods, respectively, with food safety (immediate) being the second most important in both food groups. Across all food groups price was consistently the least important attribute. Focus group discussions reinforced these results with additional insights particularly regarding trade-offs between nutrition and food safety that consumers face.

Conclusions

In rapidly changing food environments such as peri-urban Viet Nam, a wide range of food attributes drive consumer decisions towards – or away from – each food group. With Best-Worst Scaling, we generated insights about food choice in a Southeast Asian environment. Our results can inform behavior change interventions and guide food system policies such as labeling and food safety in Viet Nam.

Funding Sources

N/A.

High-Frequency Repeated Measures of Over 5,000 Infants Aged 6–27 Months Reveals Pattern of Growth Faltering in Rural Burkina Faso

Objectives

Linear growth faltering, failure to achieve one's potential for length or height at each age, is associated with increased risk of mortality and morbidity. Understanding the age-related patterns of growth faltering can inform timing of interventions and policies for prevention; however current knowledge is limited by cross-sectional data. We examine the longitudinal characterization of the timing of growth faltering among young children.

Methods

Using longitudinal anthropometric data originally collected as part of a food-aid cost-effectiveness trial, we investigated individual growth curves of 5,039 children aged 6–27 months in Burkina Faso (108,580 total measurements) to determine whether growth faltering occurs through intermittent episodes of slower growth, or continuously slow growth. We visualized individual growth curves and used two-stage regressions to evaluate growth curve smoothness at

different levels of attained length. To do this, we first obtained model fit and diagnostic parameters from individual regressions of each child's length on their age. Parameters extracted included R2, as a metric of curve smoothness, initial length at 6 months, and average velocity. We then regressed these parameters on individual-level attained length at study end.

Results

Growth faltering manifests as both continuously lower growth velocity throughout the observation period and greater heterogeneity in growth velocity amplitude. Children with lower attained length start smaller (Quintile 1 of attained length-63.1 cm initial length; Quintile 5-68.4 cm) and stay on their initial trajectories, continuously growing slower than their taller counterparts. In addition, a 0.01 increase in the R2 (smoothness) from regression of a child's length on their age is associated with a 3.10 cm increase in attained length (95% CI: 2.80, 3.41), showing that smoother growth is also associated with higher attained length.

Conclusions

Children who experience the most extreme growth faltering are likely to be less resilient to systematic growth-limiting conditions as well as episodic insults to their growth. Future research should investigate ways of improving environmental conditions to support growth.

Funding Sources

Financial support provided by the Food Aid Quality Review project, funded by the United States Agency for International Development.

Stringency of Movement Restrictions Linked With Higher Retail Food Prices but Not Overall Prices for All Consumer Goods During COVID-19 Pandemic

Objectives

The COVID-19 crisis has disrupted economies and health systems across the globe and has brought substantial challenges for food systems. Government responses key to minimizing disease spread have included a number of movement restriction policies (e.g., school closures, stay at home measures, etc.). Such policies have impacted food consumption and purchasing behaviours and have harmed much of the face-to-face type of labour required for food retailing, which in turn may have impacted the affordability of diets. We use evidence from 133 countries to investigate the association between stringency in movement restriction policies and retail price levels for food and other consumer goods.

Methods

We use the International Monetary Fund's monthly national consumer price index (CPI), and food and non-alcoholic beverage index (FCPI), as well as a ratio of FCPI to CPI—the food price index (FPI)—for 133 countries from January 2017 to November

2020. Data on stringency in movement restriction policies and COVID-19 cases and mortality per million were obtained from the Oxford COVID-19 Government Response Tracker and the European Centre for Disease Prevention and Control, respectively. Fixed effects regression models were used to estimate the association between stringency of COVID-19 movement restrictions and monthly differences in 2020 retail price levels (from average 2017–2019 levels) of foods and other consumer goods, while controlling for pandemic severity in each country and month.

Results

Regression models yielded a positive and significant relationship between stringency level and FCPI level (coeff. = 1.24–1.91; 95% CIs: 0.25–2.79) or FPI level (coeff. = 1.24–2.14; 95% CIs: 0.60–2.53). Alternatively, stringency level was either negatively associated with CPI level (coeff. = −0.57; 95% CI: −1.06 – −0.07) or not significantly associated with CPI level.

Conclusions

Government response stringency in COVID-19 movement restriction policies is linked with higher retail food price levels. Governments could consider implementing these policies alongside other measures (e.g., food assistance) to mitigate negative spillovers into the domain of food security and nutrition.

Funding Sources

This work was supported by the Bill & Melinda Gates Foundation and UKAid through the Foreign, Commonwealth & Development Office of the UK.

The Economics of Feeding Infants: Least-Cost Nutrient Adequate Diets From 6 to 23 Months of Age Using Breastmilk and Locally Available Foods in 31 Low-Income Countries

Objectives

Meeting infant needs requires complementing breastmilk with the gradual introduction of solid foods after 6 months. Existing studies have limited evidence on the availability and cost of complementary foods needed during the 6–23 month window when millions of children experience irreversible harm from inadequate diets.

Methods

Using food prices from the World Bank on 374 items potentially used for infant foods and available for sale in 31 low-income countries, food composition data, and dietary reference intakes (DRIs), we compute least-cost diets for children between 6–23 months. We further used data for food yield and nutrient retention factors and introduced volume constraints. DRIs differ for children aged 6–8 months, 9–11 months, and 12–23 months. We used lower and upper

bounds on 25 different nutrients plus fiber and sodium. Our primary specification (Scenario I) considers only total protein and fats, while a more restrictive Scenario II also considers 11 essential amino acids and fatty acids.

Results

The cost of complementary foods to replace breastmilk is $1–2/day at 6–8 months when small quantities of high-cost ingredients are needed, then declines to below $/1day at 9–11 months when lower-cost foods can be used before rising above $1/day again as children grow. Increased breastfeeding substitutes for complementary foods to a limited degree, as shown by the small decline in cost as breastmilk intake rises, and there is a clear threshold beyond which sufficiently nutrient-dense ingredients are very expensive or entirely unavailable. Average intake recommendation levels of breastmilk measured in low-income countries are close to that which makes complementary feeding least expensive, except in Scenario II for the youngest children.

Conclusions

We find that sufficient nutrients from complementary foods for infants from 6 to 23 months are not affordable for many households in these countries, implying a need for targeted assistance, especially in settings where household incomes are below the global poverty line of $1.90/person/day.

Funding Sources

This study was funded by the Bill & Melinda Gates Foundation through the project Advancing Research on Nutrition and Agriculture (AReNA), Phase II (Investment ID: OPP1177007), jointly with the project Food Prices for Nutrition (INV-016158) funded by the Bill & Melinda Gates Foundation and UKAid.

Measuring Consumer Preferences in Changing Food Environments: A Methods Review

Objectives

Measuring consumer preferences for different food quality attributes in low- and middle-income countries (LMICs) is increasingly important for interventions and policies to better address poor nutrition and health outcomes in the context of rapidly changing food environments. Despite the importance of measuring preferences, limited research has been conducted in LMICs to develop a better understanding of what matters most to consumers. This study reviews existing methods for measuring preferences and proposes a way forward for the nutrition public health community to address this important gap.

Methods

Relevant papers were identified in PubMed using pre-selected Mesh terms and by searching reference lists of key review articles. Approaches identified span the fields of marketing, economics, psychology, and nutrition public health. The papers reviewed used different methods to measure preferences of various types of food attributes.

Results

In marketing and economics, the term conjoint analysis is used to describe a category of methods that measure the stated preference of respondents by asking them to rate, rank, or choose between competing alternatives. Within conjoint analysis, several different methods exist including discrete choice experiences, ranking conjoint analysis, and best-worst scaling and each can be used to elicit preferences about observable and unobservable attributes of foods (e.g., price, taste etc.). Within the field of psychology, several techniques have been used including the Food Choice Questionnaire and Food Choice Values. Other approaches include qualitative interviews, pile sorting, and Likert scale-based instruments. Each method has strengths and weaknesses but in general, those from marketing and economics have the benefit of resulting in a ranked choice, in contrast to Likert scales and pile sorting which can be difficult to interpret and cognitively burdensome.

Conclusions

Most methods have been primarily developed, validated, and used in high-income countries with much less application in LMICs. Further research is required to adapt and develop preference elicitation methods for LMICs to better measure food preferences in the context of rapidly evolving food environments.

Funding Sources

N/A.

The Cost of Nutritious Diets for Individuals and Households in Rural Malawi

Objectives

The affordability of healthy diets is a key concern for policymakers and an active area of research. To the best of our knowledge, ours is the first study to examine the cost of nutritionally adequate diets for all population age and sex subgroups at subnational disaggregation over a 10-year monthly time series as well as to extend the least-cost diet framework from individuals to household units who consume shared meals.

Methods

We use household survey panel data matched with local food composition data, human nutrient requirements (EARs, ULs and AMDRs), and monthly market food prices. The food price data include a standardized list of 51 items comprising those foods accounting for the vast majority of household food spending, covering all food groups. Household data are from the Integrated Household Panel Survey (IHPS) 3-round panel (2010–2017). Using linear programming, we identify the combination of available foods and quantities that meet the specified nutrient requirements at the lowest total cost.

Results

Local markets can supply an adequate diet most of the time for some groups but much less often for breastfeeding women and young children, and not at al for children 6–36 months. Where possible, we find that the least-cost individual diet costs 32% of current per capita food spending. For whole households, the local market can meet aggregate nutrient needs 30% of the time. Where possible, it costs 1.8 times household food spending or all total expenditure. Food group analysis reveals households spend more on cereals, meat and roots/tubers, compared to the least-cost adequate diet, but may be under-spending on legumes, milk, fats, and vitamin A-rich fruits.

Conclusions

Finding a healthy diet in Malawi's local markets is not always possible for all types of individuals or families of varying composition. Children 6–36 months require immediate attention, including increasing access to complete complementary foods. Guiding households to make smart shifts in their food spending could improve diet quality but is insufficient alone given that adequate shared meals are unaffordable for most families. Year-round access to nutritious, affordable diets will require an increase in availability and lower cost of nutrient-dense foods and economic growth to raise incomes.

Funding Sources

Bill & Melinda Gates Foundation, CANDASA project.

Temporal Patterns in Linear and Ponderal Growth Velocity Among Children 6–23 Months in Burkina Faso

Objectives

Although studies have found co-occurrence of stunting (low height-for-age) with wasting (low weight-for-height), and found that wasting often precedes stunting, none has investigated links between velocities of linear and ponderal growth over time. Understanding temporal dependencies and growth trajectories in both height and weight can help determine optimal timing of interventions to prevent wasting and stunting.

Methods

We re-analyzed monthly height and weight data from a cohort of 6112 children aged 6–23 months in a supplementary feeding trial in Burkina Faso, conducted from July 2014 to September 2016. We built mixed effects models with multiple lag periods to determine links between linear growth trajectory (cm gained/month) and ponderal growth trajectory (kg gained/month), and vice versa, controlling for age, season, study arm, and morbidities. Lag periods included growth velocity measurements taken one (t-1) to four (t-4) months prior to the outcome velocity.

Results

On average in these data, ponderal growth is positively associated with future linear growth, but linear growth is followed by lower future weight gain. Doubling the average ponderal growth velocity (0.18 kg/month) is associated with 3%–0.2% higher linear growth velocity one to four months later, while doubling the average linear growth velocity (0.95 cm/month) is associated with 16%–3% lower ponderal growth velocity one to four months later. In all cases, the magnitude of a growth velocity's association with future growth velocities decreases as the time interval increases. These results are all significant at the 5% level.

Conclusions

We find that episodes of faster linear growth are associated with lower subsequent ponderal growth. This finding demonstrates temporal dependencies and suggests that, for infants in Burkina Faso, linear growth spurts may not be accompanied by sufficient increases in dietary intake to avoid a slowdown in weight gain. To improve child growth, programs should combine strategies that address both height and weight.

Funding Sources

United States Agency for International Development, Office of Food for Peace.

The Cost of Recommended Diets: Development and Application a Food Price Index Based on Food-Based Dietary Guidelines (P10-033-19)

Objectives

Food price data is routinely collected by governments but has not been utilized to its full potential for tracking the cost of nutritious diets. Food prices typically are used to monitor the cost of a basket of goods purchased by consumers, which may bear little relation to recommended diets. National food-based dietary guidelines (FBDG) attempt to provide specific, actionable recommendations for diets that meet nutritional needs and protect long-term health, based on expert evidence review and in-country dialogue. The objective of this research is to show how governments can use their food price data to track the cost of meeting FBDG.

Methods

The Cost of Recommended Diets (CoRD) is an estimate of the minimum cost of meeting FBDG. It requires data on food prices and quantitative FBDG. Food price data can be sourced from national statistical organizations that track inflation using a Consumer Price Index (CPI); from ministries of food/agriculture/trade; and from national household budget surveys. CoRD is calculated by identifying the least-cost 2–3 foods (by edible portion) in each food category contained in FBDG (e.g., starchy staples, protein foods, dairy, fruits, vegetables, oils), and summing the mean cost of obtaining the average gram amounts of each group. We demonstrate the results of CoRD using data from Ghana. A stakeholder dialogue was conducted with official food price data collectors in Ghana in 2016–2018 on the applicability of CoRD within existing monitoring systems.

Results

Using data from the Ghana Statistical Service and the Ghana Ministry of Food and Agriculture (MoFA), we find that meeting FBDG for vegetables and fruits is relatively expensive compared to starchy staples and the cheapest forms of protein foods, and that consumers typically underspend on fruits and vegetables compared to the expenditure shares required to obtain recommended amounts. MoFA has made changes in their food price monitoring system to enable tracking of CoRD.

Conclusions

The CoRD indicator is a policy-coherent metric of food prices that measures the cost of meeting FBDG. It rests on existing data and can be incorporated into national food price monitoring systems.

Funding Sources

Funding for this work is provided by the Bill and Melinda Gates Foundation and DFID, partially through an IMMANA (Innovative Metrics and Methods for Agriculture and Nutrition Actions) grant.

Global Inequality in National Food Supplies and Diet-related Health Outcomes, 1970–2010 (P04-094-19)

Objectives

The global food system has changed rapidly over recent decades, with large shifts in agricultural supply conditions, international trade, food manufacturing and nutrient availability. These shifts have contributed to both positive and negative impacts on health and wellbeing with some countries catching up while others have fallen further below international norms. We aim to quantify long-run changes in the food supply and diet-related health disparities across countries, to illustrate how inequality has changed from the 1970 s to the 2010 s.

Methods

Data were drawn from publicly available sources, using food supply from the United Nations Food and Agriculture Organization (FAO) and diet-related health outcomes from the NCD Risk Factor Collaboration (NCD-RisC) and the Joint Monitoring Programming, a collaboration of the World Bank, the World Health Organization, and UNICEF. We used techniques from economics (Lorenz curves, Gini coefficients, and Pen's Parade) to generate novel visualizations and metrics for global inequality and disparities in national food supply and diet-related health outcomes across countries.

Results

We computed global inequality in food supplies as available dietary energy from all fruits and vegetables, all animal source foods, and all foods other than starchy staples, and inequality in diet-related outcomes such as overweight and obesity, underweight, stunting, raised blood pressure, and diabetes. Results showed greater equality in food supplies and diet-related health outcomes converging at higher prevalence levels, while stunting and underweight became more unequal and more highly concentrated in a few high-burden countries. For example, the Gini coefficient for energy from fruit and vegetables (0.36 to 0.27), energy from animal source foods (0.51 to 0.33), and overweight and obesity (0.39 to 0.27) decreased indicating greater equality, while the Gini coefficient for underweight increased (0.40 to 0.47) demonstrating greater disparity.

Conclusions

Convergence in availability of harmful dietary components and negative health outcomes occurred faster than convergence in beneficial nutrients and positive outcomes. This research can help to inform the design of policies and programs aimed at achieving the UN Sustainable Development Goals and other targets for global food systems.

Funding Sources

Wellcome Trust.

Heterogeneity in the Cost of Nutrient Adequacy by Age, Sex, Pregnancy or Lactation Status and Other Influences on Individual Requirements in Malawi (OR21-04-19)

Objectives

Market prices for nutrient-dense foods may influence adequacy of intake, but previous studies have focused on estimated average requirements (EARs) of a target group. The aim of this study was to measure sensitivity in the cost of nutrient adequacy (CoNA) to changing requirements over the life cycle from infancy through adolescence, pregnancy or lactation, and other potential influences such as phytates in food that limit absorption of zinc.

Methods

Using nutrient composition and monthly prices for 55 foods at 29 regional markets in Malawi over 10 years (2007- 2017), we computed the least-cost diets at each place and time that meet nutrient requirements for 23 age-sex groups. We applied four levels of nutrient requirements from dietary reference intake (DRI) data, using EARs and RDAs as well as upper limits (ULs) and acceptable macronutrient distribution ranges (AMDRs). We also considered the effect of phytate on the absorption of zinc in each food. All costs account for inflation using 2011 purchasing-power parity exchange rates (USD/day).

Results

The cost of nutrient adequacy shifts over the life cycle, peaking for males aged 14–18 (median CoNA of 1.39 with EARs and 1.72 with RDAs). For all groups, CoNA also varies across the distribution of nutrient requirements, for example, among females aged 19–30 the median CoNA is 1.00 to meet only EARs, rising to USD 1.16 to also stay within UL and AMDR limits and avoid risks from excess consumption. In Malawi, some foods have high phytate levels whose anti-nutrient effect on zinc absorption also raises costs, for example, median CoNA for a lactating woman aged 19–30 is 1.43 in our benchmark case with EARs, ULs and AMDRs, and USD 1.59 if phytate is considered, rising to 1.86 if EARs are replaced by RDAs.

Conclusions

We found high variation in the cost of meeting nutrient needs over the life cycle with large differences for males and females from infancy through pregnancy and lactation, as well as other influences on nutrient requirements for each group such as accounting for the effect of phytates on zinc absorption, and differences across the distribution of requirements from EARs to RDAs. Results will be useful for programs to target populations at risk of inadequate intake, and for research on the affordability of nutritious diets.

Funding Sources

Bill & Melinda Gates Foundation (OPP1182628).

Supporting Tables, Images and/or Graphs

Sex-specific Sibling Rivalry in Child Health: Evidence from over 190,000 Twins (P18-098-19)

Objectives

This study measured ‘sibling rivalry' in child health, testing the degree to which nutritional and other resources were preferentially allocated to male instead of female children. We used twin data to control for maternal health and other circumstances, and used mortality, birthweights, heights, and weights to test whether any differences between twins started in utero or only emerged later in life.

Methods

We extracted all twin births recorded in 245 nationally representative surveys conducted in 73 low- and middle-income countries between 1990 and 2016 for the Demographic and Health Surveys (DHS). We used logistic regression to test for mortality differences (N = 191,782 twins ever born), and OLS regression to test for differences between twins < 5 yrs in birthweight (N = 10,769), height-for-age z score (N = 18,124), and weight-for-height z score (N = 17,600).

Results

We found a large negative impact on birthweight and survival of having a male co-twin. This effect occurred entirely among boys rather than girls. Males with a twin brother had lower odds of survival (OR = 0.88; 95% CI [0.83, 0.94]) and lower birthweights in grams (coefficient estimate = −0.05; CI [−0.08, −0.02]) compared to males with a twin sister. Results were robust to different model specifications. We found null effects for heights and weights. The estimated 'male twin rivalry' effect on birthweight and mortality was similar in magnitude to having a mother with below-median schooling, or two quintiles lower wealth.

Conclusions

The sex of a child's siblings has a significant impact on their health, which has important implications for intervention and research to reduce gender-based disparities. We found that having a male co-twin reduced birthweight, demonstrating that sibling rivalry begins in utero, and found that it also reduced later-life survival, implying that biological differences in gestational health may be worsened by gender biases. Both results were found among boys but not girls, which is consistent with other evidence of greater sensitivity of male fetuses and infants to their environments. Sibling rivalry among twins is just one aspect of gender bias, revealing how efforts to help both boys and girls fulfill their genetic potential requires attention to both sex-specific biology and gender bias in behavior.

Funding Sources

None.

Supporting Tables, Images and/or Graphs

Nutrient Adequacy at the Household Level and the Cost of Nutritious Diets in Malawi (P10-055-19)

Objectives

Nutrient requirements are defined for individuals, but meals are often shared and food consumption is typically measured at the household level. Prior studies of nutrient adequacy using household data have estimated requirements in terms of adult equivalents. We introduced a nutrient-by-nutrient approach to capture differences in household composition, and used this measure to test whether a household's nutrient adequacy was associated with the market cost of nutritionally adequate diets.

Methods

We used panel data on food consumption from 1398 rural Malawian households from the Malawi Integrated Household Survey Panel with monthly prices for 53 foods at the nearest market. Both datasets are collected by the National Statistics Office (NSO), matched at the market-month level. We defined household nutrient needs as the highest density of each nutrient (quantity per kilocalorie) required by any household member over 6 months, summed over daily energy requirements for each individual in the household. From local prices and food composition data we calculated a least-cost, nutritionally adequate diet for the mean household, and computed its level at each location. From observed consumption and nutrient needs we calculated household nutrient adequacy ratios (HNARs) and mean adequacy ratios (HMAR), and tested their association with the local market cost of nutrient adequacy (CoNA), controlling for seasonality, volatility of diet cost, and household size.

Results

In each round of the survey, between 36% and 59% of households reached their nutrient requirements (HMAR = 1). We found no significant association with the level, seasonality or volatility in market costs, but found that greater household size is associated with lower odds of meeting nutrient adequacy.

Conclusions

Our novel approach recognizes food sharing at the household level, typical in developing countries, thereby advancing the potential to conduct nutrition-related analyses with existing household survey data. Our initial application found no association with market prices suggesting that other factors may be more important predictors of nutrient adequacy. Future analyses will investigate dietary patterns and cost of individual nutrients to further explain the odds of meeting household needs.

Funding Sources

Bill & Melinda Gates Foundation via the CANDASA project.

Supporting Tables, Images and/or Graphs

Menu Descriptions and Nutrient Content of Food Items in U.S. Restaurants (FS02-01-19)

Objectives

This study concerns how the description of foods on restaurant menus relates to their nutrient content as disclosed on company websites. We aimed to test halo effects, regarding how claims about some desirable features might be associated with the presence of other attributes.

Methods

We used item descriptions and nutrient data for food items (n = 92,949) at the top-selling restaurant chains (n = 92) from 2012 through 2017 in the United States, compiled by the MenuStat project. We classified items into 4 types (mains, appetizers, desserts, sides) and claims into 3 groups using 29 search terms based on consumer interests in health (e.g., “nutritious”), product sourcing (e.g., “local” or “organic”), and vegetal items (vegetarian or vegan). Nutrient data focus on 4 dietary recommendations to limit sodium (mg), trans-fat (g) and saturated fats (% of energy), and to increase fiber (g). We also report calories per item (kcal) and its share from carbohydrates, protein and total fat (%). We used multiple regression to test whether nutrient content was associated with menu claims, controlling for year and restaurant brand, the item being marked as “shareable”, on a kid's menu, or regional and limited-time offerings. Methods and hypotheses were preregistered on As-Predicted.com.

Results

Contrary to our prediction, nutrient content was more often aligned with U.S. dietary guidelines when their description did include claims. With 3 claim types, 4 food types and 4 recommendations we test 48 possible cases. In 25 (52%) we found alignment between claims and nutrient recommendations, e.g., main dishes with health-related claims had 2% less calories from saturated fat (P < 0.01) and 142 mg less sodium (P < 0.01). In 3 of 48 cases (7%), claims were contrary to recommendations, all of which were desserts with sourcing claims which had more sodium, more trans-fat and more saturated fat than other desserts (all P < 0.01). In 20 of 48 cases (42%) there was no significant difference between items with and without claims.

Conclusions

Items described as vegetarian/vegan or with sourcing and health claims had nutrient contents that were more often aligned with dietary guidelines than other items. Menu labeling that communicates meal content more directly, such as nutrient fact panels, could inform choice and build trust in restaurant meals.

Funding Sources

None.

A Novel Method to Measure Resilience in Nutrition: Application to Diets and Body Mass Index of Rural Women in Nepal and Bangladesh (FS01-02-19)

Objectives

This study aimed to demonstrate feasibility of a novel method for measuring resilience in dietary diversity (DD) and body mass index (BMI) of rural women of reproductive age (WRA) in Nepal and Bangladesh. Resilience is defined as the population's ability to recover from adverse shocks, measured relative to statistical mean reversion.

Methods

We used regionally representative surveys from Nepal (n = 2187) and Bangladesh (n = 1715) collected annually in Nepal (four panels, 2013–2016) and every 6 months in Bangladesh (3 panels, 2016–2017), for BMI (kg/m2) and DD using 10 food groups (7-day qualitative diet recall, Nepal) and 6 food groups (24-hour diet recall, Bangladesh). We estimated where i is the woman, y is the outcome of interest, and z is a vector of controls. ∆yi, t+j denotes change in y between t + j-1 and t + j. Declinedi,t+1 equals 1 if the change between t and t + 1 is negative, zero otherwise. β2 measures the degree of reversal in decline controlling for mean reversion (β1) and other differences (age, age2, age cube, and socioeconomic status). β2 = 0 is our benchmark of mean reversion around the trend, i.e., recoveries after decline are not significantly different from declines after recovery.

Results

Our technique revealed significant (P < 0.01) resilience of DD in Nepal (Figure 1, Panel A). Among 1682 women in the Terai region, 47% (n = 784) experienced an initial decline in DD, and 61% of that initial decline was recovered. Mean reversion removed only 25% of the initial change among those who gained. We found no significant resilience in Bangladesh (Figure 1, Panel B) or for BMI in either country.

Conclusions

The resilience of DD in Nepal could reflect food aid responses to the 2015 earthquake or other interventions, while lack of resilience in BMI could reflect time lags, measurement errors or limits on the speed and timing of weight change. Lack of resilience of DD in Bangladesh could reflect shorter time intervals. Future work will apply this method to test for differences in resilience associated with exposure to programmatic interventions.

Funding Sources

Support provided by Feed the Future Innovation Lab for Nutrition, funded by the United States Agency for International Development (USAID).

Supporting Tables, Images and/or Graphs

A Novel Method to Measure Resilience in Nutrition: Application to Diets and Body Mass Index of Rural Women in Nepal and Bangladesh (FS01-02-19)

Objectives

This study aimed to demonstrate feasibility of a novel method for measuring resilience in dietary diversity (DD) and body mass index (BMI) of rural women of reproductive age (WRA) in Nepal and Bangladesh. Resilience is defined as the population's ability to recover from adverse shocks, measured relative to statistical mean reversion.

Methods

We used regionally representative surveys from Nepal (n = 2187) and Bangladesh (n = 1715) collected annually in Nepal (four panels, 2013–2016) and every 6 months in Bangladesh (3 panels, 2016–2017), for BMI (kg/m2) and DD using 10 food groups (7-day qualitative diet recall, Nepal) and 6 food groups (24-hour diet recall, Bangladesh). We estimated where i is the woman, y is the outcome of interest, and z is a vector of controls. ∆yi, t+j denotes change in y between t + j-1 and t + j. Declinedi,t+1 equals 1 if the change between t and t + 1 is negative, zero otherwise. β2 measures the degree of reversal in decline controlling for mean reversion (β1) and other differences (age, age2, age cube, and socioeconomic status). β2 = 0 is our benchmark of mean reversion around the trend, i.e., recoveries after decline are not significantly different from declines after recovery.

Results

Our technique revealed significant (P < 0.01) resilience of DD in Nepal (Figure 1, Panel A). Among 1682 women in the Terai region, 47% (n = 784) experienced an initial decline in DD, and 61% of that initial decline was recovered. Mean reversion removed only 25% of the initial change among those who gained. We found no significant resilience in Bangladesh (Figure 1, Panel B) or for BMI in either country.

Conclusions

The resilience of DD in Nepal could reflect food aid responses to the 2015 earthquake or other interventions, while lack of resilience in BMI could reflect time lags, measurement errors or limits on the speed and timing of weight change. Lack of resilience of DD in Bangladesh could reflect shorter time intervals. Future work will apply this method to test for differences in resilience associated with exposure to programmatic interventions.

Funding Sources

Support provided by Feed the Future Innovation Lab for Nutrition, funded by the United States Agency for International Development (USAID).

Supporting Tables, Images and/or Graphs

Nutrient Composition and Mycotoxin Contamination of Commercially Sold Premixed Cereals for Complementary Feeding of Infants in Malawi: Test Results for 94 Samples of 8 Brands (P11-079-19)

Objectives

High quality fortified premixed cereals can help caregivers meet infants’ nutrient needs after 6 months of exclusive breastfeeding, as complementary foods alongside breastmilk especially where family foods have low nutrient density. Such cereals are widely used in nutrition assistance programs, but commercially-sold brands are rarely subject to quality assurance using international standards for composition and labeling. Previous work found low and variable nutrients and high mycotoxin levels in a few samples. This study aimed to provide the first combined test for nutrient and mycotoxins in multiple samples of all premixed cereal brands for sale in a country, using Malawi as an example of a country with high burdens of child malnutrition and an active market for commercial premixed cereals.

Methods

We purchased a convenience sample of 94 packages from all 8 brands available in markets representing the Central and Southern regions of Malawi, and conducted tests for macronutrients, iron and zinc as well as aflatoxins and fumonisin, using standards set by the Association of Official Analytical Chemists. Test results were compared to labeled values and the few available standards, and interpreted in the light of qualitative findings from semi-structured interviews using snowball sampling with Malawian stakeholders.

Results

Premixed cereals generally met moisture, ash, and iron standards set by the Malawi Bureau of Standards (MBS), but just 35%, 56% and 39% of samples were compliant with the MBS standards for zinc, protein, and fat respectively, and only 63% of samples met Codex standards for energy. Most of the locally-made products exceeded national and international standards for aflatoxin and fumonisin. Labeled values deviated significantly from test results (P < 0.05) for all nutrients except protein and zinc. Key informants identified several challenges in achieving high and uniform quality premixed products in Malawi including reliability and independence of test results, cost of testing, and quality control.

Conclusions

Test results showed low and variable nutrient content, discordance with labeled values and mycotoxin contamination above Malawian standards, revealing the need for introduction of trustworthy quality assurance standards to improve quality and labeling practices for complementary foods in Malawi.

Funding Sources

This project was supported by a USDA award to Tufts University.

Supporting Tables, Images and/or Graphs