Nutrition transition's latest stage: Are ultra‐processed food increases in low‐ and middle‐income countries dooming our preschoolers' diets and future health?

Summary

Introduction

Rapid shifts in dietary patterns, marked by increased consumption of ultra‐processed foods (UPFs), are increasingly impacting the health and wellbeing of infants and toddlers in low‐ and middle‐income countries.

Methods

Utilizing data from the Demographic and Health Surveys, other national surveys, NCD‐RisC data and Euromonitor sales data, we examine changes in stunting and overweight/obesity prevalence alongside the latest data on UPF consumption trends.

Results

The prevalence of overweight/obesity among children and mothers is increasing rapidly while stunting rates decline slowly. Simultaneously, there is a significant increase in consumption of UPFs, especially among preschool‐aged children. Increasingly, poorer households are experiencing faster rates of increase in overweight and obesity prevalence compared to wealthier households. Results highlight the early socialization of infants and toddlers to unhealthy discretionary foods including UPFs, potentially setting the stage for long‐term dietary preferences that favour food with high sugar or excess sodium.

Conclusion

There is an urgent need to address the rapid increases in UPF consumption among infants and toddlers. Options include expanding the WHO Code on marketing to protect 0–3‐year‐olds; creating front‐of‐package warning labels focusing on products for children ages 0–3 years to remove all added sugar and limit sodium in foods and beverages they consume.

Abbreviations

BMI

body mass index

DHS

Demographic Health Surveys

LMICs

low‐ and middle‐income countries

NCDs

noncommunicable diseases

NPPM

Nutrient and Promotion Profile Model

SSBs

sugar‐sweetened beverages

UPFs

ultra‐processed foods

1. INTRODUCTION

There is no dearth of literature surrounding the shift across the globe to a stage of the nutrition transition in which processing of food has shifted towards ultra‐processing and where consumption is increasingly dominated by foods high in added sugar, sodium and unhealthy fats. ¹ , ² , ³ The evidence has focused on children aged 6 years through adulthood. The preschool period has largely been ignored except for research focused on undernutrition. This observation is made within the context and a period (over the past two decades) when the world has experienced a slow decline in stunting among children while simultaneously seeing rising obesity in children and adolescents in most low‐ and middle‐income countries (LMICs). ⁴ , ⁵ , ⁶ While there are many economic, marketing and other drivers underlying this shift, a major trend has emerged that impacts all populations: that of increased intake of ultra‐processed foods (UPFs). ⁷ UPFs are linked with an array of long‐term and short‐term health conditions, ranging from increased weight gain and risks of overweight and obesity to a vast array of disease outcomes. ⁸ , ⁹ These foods impact both stunting and obesity by appearing to slow the decline in stunting while greatly enhancing the increase in obesity. In addition, consumption of these UPFs by preschoolers increases children's long‐term tastes and preferences. ¹⁰ , ¹¹ , ¹²

The nutrition transition is a broad layout in the shifts over the many thousands of years in the ways we eat, drink, and move and their impact on our body composition and health. ¹³ , ¹⁴ This begins with a pattern (1) with hunter–gather societies where our diet was varied, energy expenditure was very high and those who reached adulthood were very tall. This was followed by famine (Pattern 2) during the period of shifting to monoculture agriculture, cereal domination and high energy expenditure with nutritional deficiencies emerging and we grew shorter; a period of receding famine (Pattern 3) when the diet was more varied and labour intensity was reduced with industrialization. The current period is pattern 4 where chronic diseases dominate, sedentarianism is high and we shift towards food dominated increasingly by UPFs which are high in saturated fat, sugar and sodium. Our goal is to shift towards a pattern of behavioural change (pattern 5) with increased purposeful activity, a diet dominated by real and minimally processed food and drinking water and other noncaloric options. With this we would see an increase in height from the reduced height of countries with mainly UPF diet to one of greater height with diets mainly of real and minimally processed food and purposeful activity.

The long‐term shifts have been documented often, including in a Lancet series on the double burden of malnutrition. ⁴ The overweight/obesity shifts have also been documented for all ages. ¹⁵ Having contained the pandemic, calls to address extant challenges of rising overweight–obesity, human‐made and climate‐related tragedies have become louder. Using nationally representative data from the Demographic Health Surveys and other data sets, we examine the most recent patterns and trends.

We express concern that many studies on UPFs, including those conducted in LMICs, fail to adequately address the rapid rise in UPF consumption among infants and young children in these regions. UPFs are not only often high in added sodium, sugar and saturated fats but have also undergone manufacturing processes by which key components have been completely decomposed, often to the cellular level, then reconstituted with many colours, flavours and other additives to appear food‐like. ¹⁶ While the bulk of the research has focused on food consumed by adults and children, there is increasing scrutiny of infant and toddler foods, most of which contain excessive levels of sugar and in some cases sodium. ¹⁷ , ¹⁸ , ¹⁹

UPFs' impact on infants and toddlers is critical. While there has been a slow decrease in stunting, the emergence of obese stunted preschoolers is a new phenomenon related to increases in UPFs as shown by dozens of diet surveys and two publications. ⁶ , ²⁰ , ²¹ Taste preferences established in this time are linked with profound impacts on long‐term eating patterns. ²² , ²³ , ²⁴ The intake of UPFs—many with excessive sugar and sodium content and use of refined carbohydrates—impact our microbiome and influence long‐term eating patterns. ²⁴ , ²⁵ Not only are there potential health impacts such as on the gut microbiome but profound impacts on long‐term diets ²⁶ and health. ²⁷ , ²⁸

In this paper, after a brief section discussing the data sets used, we will first document the long‐term trends in stunting and overweight–obesity across LMICs then present data on UPF trends overall, then focused on consumption among 0–4‐year‐olds and their mothers. This is followed by a discussion. ²⁴

2. MATERIALS AND METHODS

2.1. Data sets used

2.1.1. Demographic and health surveys (DHS) and other national surveys

The DHS data sets derive from nationally representative samples of mothers and preschooler child ages 0–4 years. ²⁹ These data are available publicly for countries defined at different times by the World Bank as LMICs. We use data for countries which were collected in the 1990s and this past decade for many analyses. Starting in about 2016, the DHS also began to ask questions about junk food intake from mothers and their preschoolers. We do not have precise definitions of junk food, but the interview instructions suggest they are all UPFs as they noted to be foods high in unhealthy nutrients. We use these data for all countries with such data available through 2023. The DHS used highly standardized measurements for weight and height of all preschoolers ages 0–4 years and their mothers. These data are all national representative surveys with standardized anthropometric collection with the exception of the China survey. ³⁰ Table S1 provides the countries, sample sizes and survey years utilized in this study. All these data are based on real measures of weight and height and hence body mass index (BMI).

2.1.2. NCD‐RisC data

The approach used by this team of scholars is to find whatever data for representative samples—be they small or large, and create a complex simulation to generate for all age groups in a country BMI data. ¹⁵ In other words, these data are estimates based on sophisticated simulation models using whatever data they could find from small samples to national ones.

2.1.3. Euromonitor data

Euromonitor International, a commercial market research entity, uses data on sales for each commodity to show trends in sales for major corporations. Small‐scale local companies are often not included. The trends tend to fit dietary and food purchase trends when compared; however, they are lower than the actual trends. ³¹ More recently, Euromonitor created a nutrition module within their Passport platform, which obtains macronutrient levels of key nutrients in the products in their global database. ³²

2.1.4. Anthropometric measurements

All the DHS data sets along with the Chinese and Indonesian databases we used had standardized protocols to measure weight and height. ³³ We calculated BMI as weight in kilograms divided by height in meters squared (kg/m²). We defined overweight according to the WHO definition of BMI ≥25 kg/m². ³⁴ , ³⁵ We adjusted BMI for women ages 15.0–17.9 according to the International Obesity Task Force definitions.

We assessed children's anthropometric statuses by comparing data with the WHO Child Growth Standards (WHO, 2006). ³⁶ We used the WHO igrowup macro to calculate z‐scores and excluded children with invalid z‐scores for ages years. ³⁷ Of note, if the z‐scores for a child fall outside ranges stipulated by the WHO growth standards, they are typically excluded because they are considered invalid due to being biologically improbable or erroneous. ³⁷ We excluded two countries, Benin and Pakistan, from this study entirely due to their high percentages of invalid z‐scores.

We calculated annualized change in prevalence by dividing the absolute change in levels by the number of years between surveys to provide a comparable measure of positive or negative change for all measures of malnutrition in countries for which we have 2 years of data. Our measure of undernutrition was stunting (HAZ <−2) for children aged 0–4 years.

Some might argue from a high‐income perspective that we should consider only obesity as our measure of overweight. However, extensive epidemiological research significantly associates BMIs of 22 or 23 with increased risks of noncommunicable diseases (NCDs) and the risk of becoming overweight (BMI ≥25 kg/m²) across LMICs. ³⁸ , ³⁹ , ⁴⁰ We also acknowledge the role of poor dietary quality as a common determinant of the double burden of malnutrition and potentially an element contributing to other dimensions of poor health, independently of the anthropometric variables.

2.1.5. Data sharing

The DHS and NCD‐RisC data are available online and are free to use. Euromonitor data from Passport International are proprietary and accessed via paid subscription but are available through thousands of universities, research institutes and libraries globally.

3. RESULTS

3.1. Long‐term trends in stunting and overweight

The DHS data shows the remarkable slow and steady decline in stunting prevalence among LMICs between surveys in the 1990s and their most recent survey. WHO growth standards are used for these analyses which includes delineation of implausible scores. ³⁷ , ⁴¹ There are major declines in most countries with a few exceptions in sub‐Saharan Africa, as shown in Figure 1. Table S1 provides the basic sample size of all the DHS surveys and Table S2 shows the changes in prevalence of stunting. Only two out of 70 countries—Mozambique and Sierra Leone—do not have a long‐term decline in stunting.

FIGURE 1.

Stunting prevalence among children ages 0–4 years in low‐ and middle‐income countries, 1990–2000s and 2000–2010s. Source: The Demographic and Health Surveys (DHS) Program.

For overweight trends, we do not utilize the DHS aged 0–4 years measures of overweight as we are measuring adiposity at such a young age, given the complexity of rebound growth at young ages, which may be quite misleading. ⁴² , ⁴³ , ⁴⁴ Rather, we use the NCD‐RisC simulated figures for children and adolescents shown in Figure 2. We combine overweight and obesity as this is so much more relevant in countries with such low BMI thresholds for onset of many NCDs.

FIGURE 2.

Overweight + obesity prevalence among children ages 5–19 years in 1990 and 2022. Source: NCD‐RisC.

This increase in overweight and obesity is indicative of a shift to the stage of the nutrition transition wherein our eating, drinking and movement patterns shift and NCDs increase significantly. ¹⁴ We cover later the dietary shifts involved from both UPF sales and DHS survey results. We show in Table S3 that in all DHS countries except Kazakhstan and Sierra Leone, overweight and obesity have increased among women of child‐bearing age (15–49 years).

3.2. Are these changes affecting the poor or the rich?

We utilize the DHS data, as these include measures of household assets that create a wealth index that has been used by many scholars over time to indicate socioeconomic status. Figure 3 shows that the countries in the darker colour, with annualized growth rates of overweight/obesity, are in the lowest quintile wealth group, whereas those with negative growth and the lighter colour are countries where the top quintile of the wealthy household's experience more adult growth in overweight/obesity (see also Table S4). This highlights that countries in East Asia and the Pacific, Latin America and the Middle East predominantly have higher overweight/obesity growth among the poor. In contrast in South Asia and sub‐Saharan Africa, it is still a situation where the highest‐wealth quintile is still experiencing higher overweight/obesity prevalence than lower‐income individuals. This is only suggestive of what might impact children in the long‐term as we do not have causal data.

FIGURE 3.

The shifting burden of overweight and obesity from higher‐ to lower‐wealth populations in sample countries. ^aPositive difference indicates higher annualized growth in overweight/obesity prevalence for the lowest wealth quartile. Source: All data are from the Demographic and Health Surveys (https://dhsprogram.com/) except for China (China Health and Nutrition Survey), Indonesia (Indonesian Family Life Survey), Mexico (National Survey of Health and Nutrition), Brazil (Brazilian National Health Survey) and Vietnam (Vietnam Living Standards Survey).

3.3. Ultra‐processed food sales and consumption

Using Euromonitor data, we can look at aggregate sales trends. In Figure 4, we see the growth in sugar‐sweetened beverages (SSBs) and non‐essential food sales (see also Tables S5 and S6). These are aggregate data. In every country there is growth in both categories of food, which all would be classified as ultra‐processed.

FIGURE 4.

Total sales volumes of (A) sweetened beverages^a and (B) non‐essential foods^b in 2004 and 2022. ^aSweetened beverages include non‐cola carbonates, regular cola carbonates liquid and powder concentrates (not reconstituted), juice drinks (up to 24% juice), nectars (25%–99% juice), ready‐to‐drink coffees and teas, sports and energy drinks and Asian specialty drinks. Not all countries have data for all categories, and not all categories distinguish between sugar‐sweetened and ‘light’ or ‘diet’ drinks. Included are low‐ and middle‐income countries for which Euromonitor had data for most sugar‐sweetened beverage categories; countries with modelled data were excluded. ^bNon‐essential foods include cakes, pastries, chocolate and sugar confectionery, chilled and shelf‐stable desserts, frozen desserts, ice cream, sweet biscuits, snack bars, processed fruit snacks, salty snacks, savoury biscuits, popcorn and other savoury snacks. Source: Euromonitor International Limited 2024 © All rights reserved.

Euromonitor sales data also show that the amount of sugar sold from selected milk‐based products is increasing rapidly. Figure 5 illustrates this increase in billions of grams or 1 million kg of sugar sold. Figure 6 shows global growth in the retail value of the global infant and toddler feeding sector, including for baby foods, toddler milks and infant formulas. In LMICs, the increase may be slightly overestimated as infant numbers increased in Africa while much less so in Asia.

FIGURE 5.

Sugar sales from infant and toddler foods. ^aLow‐ and middle‐income countries. LMIC, low‐ and middle‐income countries. Source: Euromonitor International Limited 2024 © All rights reserved.

FIGURE 6.

Retail value^a of the global infant and toddler feeding section. ^aRetail Value RSP (the retail selling price—i.e., sales at end price to consumer—including retailer and wholesaler mark‐ups and sales tax—except in the United States and Canada—and excise taxes). ^bLow‐ and middle‐income countries. LMIC, low‐ and middle‐income countries. Source: Euromonitor International Limited 2024 © All rights reserved.

We shift to the DHS data to show results from survey questions asking about the food consumed the day before the survey for select focused questions. These are new data introduced to the DHS only in the past decade. Figure 7 shows the large proportion of preschoolers ages 6–23 months who consumed SSBs the day before the survey (see also Table S7). Additional questions on unhealthy foods were asked for the children 6–23 months and for women aged 15–49 years (shown in Figure 8, Table S8). During this crucial period of development for toddlers, these results are quite high and fit the increasing sales trend shown above.

FIGURE 7.

Percent of children ages 6–23 months who consumed sweet beverages^a yesterday. ^aBeverages include sweet/flavoured milk and yogurt drinks, sweet/flavoured soy milks, fruit juice and fruit‐flavoured drinks, chocolate‐flavoured drinks, sodas, malt drinks, sports drinks, energy drinks, sweetened tea, coffee, herbal drinks and other sweetened liquids. Base map © 2024 Mapbox © OpenStreetMap.

FIGURE 8.

Increasing unhealthy food consumption: Percent of children and women who reported consuming unhealthy foods^a yesterday. ^aUnhealthy foods are a group of sentinel food types that include sweet foods such as chocolates, candies, cakes, cookies, vishetti, sweet biscuits and ice cream and fried and salty foods such as chips, bagia, mandaazi, fried potatoes, fried cassava, fried sweet potatoes and instant noodles. Source: The Demographic and Health Surveys (DHS) Program.

4. DISCUSSION

This paper shows that not only is the overweight/obesity level of women of childbearing age and children increasing, but the rate of increase is increasingly greater among lower socioeconomic status households. Linked to this is the rapid increase in the sales of UPFs and beverages and early introduction of SSBs to infants and young children in LMICs. We also highlighted the large increase in sales of infant/toddler foods and the significant proportion of UPF intake among toddlers aged 6–23 months and among mothers. The excessive sugar and sodium from these foods are not needed. We agree that not only 0 to 23‐month‐olds but also a year older infants and toddlers should have no added sugar in their diets aside from the lactose in milk and infants' sodium needs should be limited to about 370 mg total, while for those aged 1–3 years, the maximum is about 700–800 mg. ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ All of these can be linked not only to increased risk of overweight and obesity but also at very early ages to increased risk of stunting though the only studies on stunting and UPF intake is cross‐sectional. ⁴⁹

Without a doubt, such a rising trend of UPF consumption among infants, toddlers and women of childbearing age is a growing public health concern, particularly in terms of excessive sugar and sodium intake. ² , ³ , ⁷ , ⁴⁸ , ⁵⁰

Moreover, UPFs are aggressively marketed, affordable and widely available in LMICs, making them a dietary staple for many resource‐constrained households. This shift towards UPFs perpetuates chronic undernutrition and stunting, as noted by UNICEF ⁵ and Baker et al. ⁵¹ Emerging evidence underscores this connection, with studies by Tzioumis and Adair ⁶ and Hoffman et al. ²¹ identifying the rising nutrition transition—characterized by UPFs—as a significant contributor to persistent stunting in LMICs. Dietary patterns established in early childhood are crucial, as they significantly influence long‐term health outcomes. The widespread consumption of UPFs during these formative years threatens not only current health but also the future potential of children in these regions.

There is evidence that dietary patterns established in early childhood are crucial, as they significantly influence long‐term dietary and health outcomes. ²⁶ , ⁵² , ⁵³ Poor dietary habits rooted in early childhood can lead to a higher propensity for NCDs such as obesity and type 2 diabetes later in life. Concerns thus got to be legitimately raised about the problematic observation of early socialization of infants and toddlers to unhealthy discretionary foods including UPFs. For instance, the 2022 DHS conducted in Ghana showed that 31% of children aged 6–23 months were given a sweet beverage during the previous day. ⁵⁴

High consumption of UPFs is associated with negative health impacts, including poor nutritional profiles and elevated risks for various metabolic disorders. ⁸ The World Health Organization (2023) ⁵⁵ specifically recommends that infants and toddlers aged 0–24 months should not consume added sugars, including those found in many UPFs, to avoid the development of a lasting preference for sweet tastes. Similarly, the WHO has established guidelines with limited sodium intake for these ages. ⁵⁶

Even more concerning is the potential impact that prenatal exposure to excessive sugars and sodium can have on foetal development and predisposition to health issues, as maternal nutrition plays a critical role in shaping future health risks. ⁵⁷ This concern extends to the breastfeeding period, when nutrients ingested by the mother can be passed to the infant, potentially affecting their metabolic programming and appetite regulation. Given these findings, there is a pressing need for improved dietary guidelines for mothers and stringent regulations on infant and toddler food products to protect and promote the health of future generations.

Our results are backed up by a series of papers examining infant and toddler diets in various LMICs. One cross‐sectional study of preschoolers in Nepal found that over 25% of calories in toddlers' diets were from UPFs and that high UPF intake was linked with stunting. ⁴⁹ Other studies by Pries et al. ²⁰ document the shift in infant feeding towards increased UPF consumption in a set of different LMICs. Before the last decade, few individuals studied this shift.

At a global level, Baker et al. ¹⁷ have documented the ways the infant food industry has found ways around the WHO–UNICEF Code. The original code focused only on marketing of infant formula. As Baker et al. have shown, global formula companies got around the code by making toddler and follow‐up milks look identical in packaging to infant formulas and advertised these products to the extent that sales of formula, as well as these substitutes, grew rapidly in LMICs. ¹⁷ , ⁵⁸ , ⁵⁹ , ⁶⁰ Others have written on this subject. ⁶¹ , ⁶² , ⁶³ Essentially, Baker and others call for an expansion of the WHO code to cover all breast‐milk substitutes and all food fed to infants aged 0–24 months.

There is a vast literature from higher income countries documenting the profound impact of early infant diets on long‐term food preferences. Mennella is one of the leaders in this work. ³⁰ , ³¹ , ³² This literature suggests that the less sugar and sodium we feed to our infants and toddlers, the healthier will be their subsequent eating patterns and in turn, these eating patterns are associated with increased risk of many NCDs ranging from becoming overweight/obesity to diabetes, hypertension and many other NCDs.

The key policy options are twofold: (1) expand the WHO code to ages 0–2 years and (2) in all front‐of‐package labelling approaches, give the worst scores or warning signs to all foods sold to infants 0–24 months that contain any added sugar and for sodium at levels set for 0–3‐year‐olds by each country or WHO. ⁵⁶ This is particularly pertinent right now for the large number of countries instituting front‐of‐package warning labels. Each of these countries should create warning labels for infants/toddlers.

Additional policy options are discussed. One potential approach is the utilization of the WHO's Nutrient and Promotion Profile Model (NPPM) developed for foods intended for infants and young children. ⁶⁴ This model provides a robust framework for evaluating the nutritional quality of foods and regulating their marketing practices to protect young populations from unhealthy food exposures. This model should be used to set clear, enforceable standards for the nutrient contents of infant and toddler foods, explicitly banning added sugars and limiting sodium. Additionally, the NPPM should guide the creation of labelling mandates, and marketing restrictions that prevent the promotion of nutritionally inadequate foods to parents and caregivers. The policy could be operationalized through the development of legislation that mandates compliance with NPPM standards for all foods marketed to children under 24 months or require all infant and toddler food products to adhere to NPPM standards, with mandatory labelling on packaging that confirms compliance.

To ensure the effectiveness of these policies, regular monitoring and compliance checks should be conducted by national food safety authorities, as well as academia and civil society organizations. ⁶⁵ However, such partnerships and interventions need relevant contextual challenges including the peculiar heterogeneity of national food environments and their variegated political economies. ⁶⁶ Collaborations with international organizations such as UNICEF and the WHO can provide additional support and resources for implementing these measures effectively.

5. CONCLUSION

Infant and toddler diets in LMICs have changed over time. Children are increasingly being fed UPF at very young ages, and child overweight/obesity is increasing faster than stunting is declining. Not only will this impact long‐term eating patterns but also increase the risks of a wide array of NCDs for these children as they age. Global action on this topic is essential; however, major current healthy food policies such as taxes on UPFs and warning labels on food have ignored this critical age group. We need strong food policy action to rectify this omission. Consideration of expansion of the WHO Code on marketing of breast‐feeding substitutes should be expanded to include infant and toddler foods and beverages, as should front‐of‐package labels, especially warning labels.

AUTHOR CONTRIBUTIONS

BP and AL created the broad outline of this study. BP oversaw data analysis, drafted the text and conceived the tables and figures. AL carefully reviewed and edited the text. All the authors approved the final version of the paper.

CONFLICT OF INTEREST STATEMENT

No conflict of interest was declared.

Supporting information

Data S1.

Popkin BM, Laar A. Nutrition transition's latest stage: Are ultra‐processed food increases in low‐ and middle‐income countries dooming our preschoolers' diets and future health? Pediatric Obesity. 2025;20(5):e70002. doi: 10.1111/ijpo.70002