Household socioeconomic status, dietary diversity, and risk of stunting among South African primary school children Authors: Johanna Beckmann, Christin Lang, Rosa du Randt, Annelie Gresse, Kurt Z. Long, Ivan Müller, Madeleine Nienaber, Siphesihle Nqweniso, Uwe Pühse, Jürg Utzinger, Cheryl Walter, Markus Gerber Published: 2026 ABSTRACT BackgroundChildren in low- and middle-income countries are disproportionately affected by socioeconomic disparities, which might influence access to diverse and nutritious foods, increasing the risk of food insecurity and low dietary quality. Hence, monitoring of these conditions is crucial. This study examined whether children from households with different wealth status vary concerning food insecurity, dietary diversity, consumption of specific food groups, and risk of stunting.MethodsBaseline data were utilized from a multi-country randomized intervention trial. Primary school children (Mage= 8.3±1.4 years; 416 girls and 423 boys) from four schools in a marginalized, peri-urban area (government-defined wealth index quintile 3) in Gqeberha, South Africa, participated. Parents/guardians provided information about household socioeconomic factors and completed a 24-hour food item checklist and a food security questionnaire. Stunted children were defined as <-2 SD height-for-age compared to current World Health Organization child growth standards. Analyses of variance and χ2-tests were employed to compare children classified into five wealth quintiles.ResultsOn average, children consumed food items from 5.8±2.0 healthy food groups. One out of 12 children (8.4%) were stunted. Contrast analyses showed that children from low-resourced families were exposed to more food insecurity, consumed fewer different food items, and were more likely to be stunted than children from the highest socioeconomic categories (p < 0.05). The latter were more likely to consume cereals, roots, vitamin A rich vegetables, oil, sweets, and sweet beverages. Although stunting still exists in marginalized areas of South Africa, our findings confirm that children with higher wealth status are less likely to be stunted, but consume more food items from unhealthy food groups.ConclusionOur findings reiterate the need for improving awareness of healthy food consumption in low- resourced areas. Addressing upstream determinants such as food insecurity, poverty, and access to affordable, healthy foods is essential to reduce socioeconomic differentials in food consumption. School-based nutrition programs should be framed as complementary measures within broader government-led actions, including school feeding policies and teacher-focused nutrition education, to improve inadequate dietary quality among school-age children and address socioeconomic inequalities in stunting risk. Introduction Many low- and middle-income countries (LMICs) continue to struggle with social inequalities and poverty, thus directly affecting food security and health of household members [1]. Household socioeconomic status, income, and poverty are key factors for the health of children [2–4]. The World Food Program defines household wealth as “the value of all natural, physical and financial assets owned by a household, reduced by its liabilities” [5]. Evidence suggests that food insecurity in LMICs is shaped by basic causes, which are socioeconomic, structural, and upstream determinants, including low income, unstable employment [6], and constraints in the local food environment such as limited market availability and poor transport infrastructure [7]. At the unit of the household, the presence of children and adolescents, female-headed households, educational attainment, and reliance on social grants have been identified as factors associated with experiencing hunger [7, 8]. On the other hand, household food insecurity and low dietary diversity are linked independently or together to undernutrition [9], especially stunting [10], among young children. Although South Africa – currently classified as a middle-income country – has made improvements in reducing poverty rates and food insecurity, and lowered the prevalence of stunting, high inequalities between socioeconomic quintiles and extreme poverty remain [11, 12]. According to South African statistics from 2015 to 2023, the socioeconomic inequalities nationwide remain high, with 55.5% of the population living under the poverty line [13, 14]. A distributional analysis examining the household socioeconomic status dynamics from 2001 to 2013 in a rural South African region demonstrated that households with low socioeconomic status have greater difficulty to overcome this inequality gap; and that achieving higher socioeconomic status is faster if household wealth has already shifted toward a wealthier level [15]. Hence, there is a pressing need for the government to focus on households that are at highest risk of being affected by public health concerns. Previous studies in LMICs, including South Africa, have illustrated that an increase in socioeconomic status is linked to an increase and greater food consumption, also with higher energy, saturated fat, and cholesterol intakes [16, 17]. With regard to South Africa, representative data from all provinces showed a mean dietary diversity score of 4.02 (out of 9 food groups) at the national level, indicating poor dietary diversity [18]. Additionally, recent studies from South Africa revealed a high prevalence of food insecurity among households living in rural and peri-urban areas [19, 20]. In general, South Africa has a good food supply chain at the national level. Nevertheless, access to food is inadequate at the unit of the household, and food is not readily available everywhere [21]. For instance, households in rural areas are less likely to have sufficient food availability and have difficulties consuming different types of food that meet dietary and nutritional needs. The importance of a good food environment, such as land availability and agriculture in rural areas [22] and access to farmer- or supermarkets, has been associated with a healthier diet and better nutritional status [23]. However, as countries urbanize, transitions in supermarket availability could also result in shifts in easy access to unhealthy and high-processed foods [24]. Food security, and hence, access to nutritious food, is an important factor for good nutritional status and health, and optimal physical and mental development [25, 26]. The Food and Agriculture Organization (FAO) describes food security as a state, in which the availability of, and access to, quality, adequate food, and clean water is guaranteed at all times [27]. Households in urban areas and those with higher socioeconomic status generally have healthier diets and greater dietary diversity than households in rural areas and those with lower socioeconomic status [17]. However, a more diverse diet in these areas has been linked with a higher consumption of animal-based food, higher energy intake, and high intake of saturated fat, which might lead to overnutrition [17]. At the same time, households in marginalized areas struggling to survive economically often have to cover the daily energy requirements of its members mainly by starchy foods, which do not meet the requirements for an adequate nutritious diet. Children in such households are more prone to developing stunting or impaired growth and development due to poor nutrition [17, 28], illustrating that household wealth is related with stunting and undernutrition [29]. Although stunting still remains a public health concern in South Africa, an overall decrease has been described in the recent past with 21.4% of children < 5 years being stunted in 2017 [30]. Additionally, Labadarios et al. showed provincial differences in South Africa and their association with mean dietary diversity [18]. The Eastern Cape province, one of the nine studied areas, was found to have a low level of dietary diversity with the most consumed food groups being cereals/roots, meat/fish, dairy products, and vegetables. Eggs and vitamin A rich fruits and vegetables were the least consumed food groups. This study also found that households with lower socioeconomic status have a higher probability of having low dietary diversity [18]. Inadequate diets that lack diversity could lead to contrasting adverse outcomes, such as overnutrition characterized by consuming highly processed food that is high in fat and energy, or undernutrition by consuming a nutrient-low diet resulting in severe impairments in physical and cognitive development [31]. An increased understanding of the role of socioeconomic factors in determining dietary diversity in marginalized and peri-urban areas is important. The present study aimed at providing a descriptive overview on food insecurity and consumption of food items from different healthy and unhealthy food groups and number/percentage of stunted children in marginalized areas of Gqeberha (formerly known as Port Elizabeth) in the Eastern Cape province of South Africa. Furthermore, the present study explored how household socioeconomic factors in these areas is associated with food insecurity, dietary diversity, consumption of food items from specific food groups, and prevalence of stunting. We hypothesized that children in the lowest household wealth quintile would experience higher levels of food insecurity [17], consume a lower number of food items, and report a less varied diet than their peers in the highest wealth quintile [18]. Moreover, we expected that the risk of childhood stunting would be higher among children from households with lower socioeconomic status compared to children from better-off households [28]. Methods Study design and population This study used baseline data from a multi-country randomized intervention trial, which explored the effects of a school-based physical activity and multi-micronutrient supplementation intervention on growth, health, and wellbeing of school children in three African countries (Côte d’Ivoire, South Africa, and Tanzania), which took place over a 2-year period. Schools were eligible to participate if they were public (no fee paying), had available facilities for physical education, and were not registered in any other clinical trial or nutritional intervention. The detailed protocol of the multi-site trial, including sample size calculation, has been described elsewhere [32]. Data collection for this study was conducted between February and April 2019 in the regions of New Brighton, Zwide, Gelvandale, and Bethelsdorp, located approximately 20 km from the city Gqeberha in the Eastern Cape province of South Africa. Gqeberha is unique from other areas as it is one of only two large cities in the Eastern Cape, and one of the most disadvantaged regions in South Africa. Gqeberha differs from most other South African cities in its economic structure and in the severity of poverty and service challenges, partly due to its large industrial area [19]. The study population consisted of 839 primary school children aged 6–12 years recruited from 41 classes at grade levels 1–4 from four public primary schools. South African public schools are ranked in government-defined wealth index quintiles (1 = poorest; 5 = least poor), based on financial support provided according to the wealth of the community surrounding the school [33]. The schools in this study were classified as quintile 3 schools. The inclusion criteria for participants were that they had written informed parental consent, were not included in another research project, and were not acutely ill at the time of data assessment. The study was registered as a cluster randomized control trial in the ISRCTN registry. The study was registered on 9 August 2018 in the ISRCTN registry (ISRCTN29534081; http://www.isrctn.com/ISRCTN29534081; last accessed on 7 June 2022). Measures Socioeconomic status A self-administered structured questionnaire provided in English and Afrikaans was distributed to the parents/guardians to collect information on sociodemographic variables, including age, sex, housing characteristics, and types of household goods owned. Detailed information was provided to the parents/guardians during a parental meeting before the questionnaire was distributed. During this meeting, parents/guardians were invited to ask questions if need be. After completion of the questionnaire, the parents returned the questionnaire to the teacher in a sealed envelope via their child. Household socioeconomic factor scores were derived using principal component analysis (PCA) to generate a wealth index, which was utilized as a concrete operationalization of socioeconomic status [5, 34]. Two housing characteristics (access to running water inside the house and bathroom/shower inside the house) and eight household material goods (cell phone, fan, electric oven, fridge, freezer, radio, TV, and washing machine) were employed in the generation of household wealth index. Items were included in the PCA if the following criteria were fulfilled: (i) less than 2% of missing values; (ii) in PCA between-item correlations of ≥ 0.10 and < 0.90; and (iii) cross-tabulations pointing toward a positive relationship between the independent (quintiles) and dependent variables (household characteristics/material goods). Assets were excluded if they were possessed by ≥ 95% or ≤ 5% of the sample population. In brief, eigenvalues were built from the transformed variables using the factor functions. The first factor (component 1) was utilized to generate weights for each included variable, which had an eigenvalue of 3.3 and explained 33.3% of variance (Kaiser-Meyer-Olkin measure of sampling adequacy: 0.83, Bartlett’s test of sphericity: approximately χ2 [45] = 1756, p < 0.001). This approach has been described elsewhere [35]. Wealth index quintiles were subsequently built with quintile 1 representing children with the lowest household socioeconomic factors and quintile 5 representing the highest household socioeconomic factors. Food insecurity A self-administered questionnaire was distributed to the parents/guardians to assess household food insecurity and consumption of specific food items. To assess household food insecurity, we used the household hunger scale (HHS) from Food and Nutrition Technical Assistance III Project supported by FAO [36], an experience-based measure of severe food deprivation developed for use in food-insecure settings. The HHS includes three occurrence questions: (i) “Was there ever no food to eat of any kind in your house because of lack of resources to get food?” (ii) “Did you or any household member go to sleep at night hungry because there was not enough food?” (iii) “Did you or any household member go a whole day and night without eating anything because there was not enough food?” with frequency of occurrence follow-up questions referring to the previous 30 days. Possible answers range from 0 (never) to 2 (often) [36]. In this study, the sum of these food access questions was defined as 0–1 reflecting little to no household hunger, 2–3 reflecting moderate household hunger, and 4–6 reflecting high household hunger. Throughout this manuscript, we refer to the HHS-derived categories as ‘food insecurity’. Consumption of food items and dietary diversity In low resourced settings, the dietary diversity score (DDS) is a valid and commonly used indicator [37–39], which serves as a proxy for diet quality reflecting the individual probability of healthy micronutrient food sources in the diet [40]. To measure dietary diversity, we assessed children’s food item consumption during the past 24 h applying a self-administered structured food item checklist, being a fast, reliable and inexpensive method [41]. Parents/guardians were asked to consult with their child before completing the questionnaire. On behalf of the children, parents/guardians were asked to report each food item (independently of quantity) that the child consumed retrospectively. In total, information on 143 food items of the following food group categories [1–16] was collected based on FAO recommendations [39, 42]: (i) cereals (e.g., maize and rice); (ii) white roots and tubers (e.g., potatoes); (iii) vitamin A rich vegetables (e.g., carrot and pumpkin); (iv) green, leafy vegetables (e.g., spinach and lettuce); (v) vegetables (e.g., tomato and pepper); (vi) vitamin A rich fruits (e.g., mango and peach); (vii) fruits (e.g., wild fruits and melon); (viii) organ meat (e.g., liver and kidney); (ix) flesh meat (e.g., beef and pork); (x) eggs; (xi) fish (e.g., dried fish and tuna); (xii) legumes, nuts and their products (e.g., beans, peas); (xiii) milk and milk products (e.g., cheese and cow milk); (xiv) oil (e.g., oil and butter); (xv) sweets (e.g., honey and chocolate); and (xvi) spices, condiments, and beverages (e.g., sweetened juice drinks and sweetened soda). The literature indicates that regardless of the number of recalls, dietary diversity is indicative of what the nutritional and socioeconomic status of study participants may be [37, 41, 43]. Whereas the applicability of the resulting DDS assessed by the specific 24–hour recall food checklist, as well as the number of food items and the 1–day administration period applied in the present study were not specifically tested for validity in this local setting. The food items used in the present study are based on the recommended food items of the FAO and slightly modified through commonly consumed food items among South African families [39, 42]. In the present study, two measures of dietary diversity were considered according to the FAO “Guidelines for measuring household and individual dietary diversity” [44]. First, we calculated the total number of consumed food items (possible range from 0 to 143 items with higher values reflecting higher food item consumption). Second, we also calculated the “woman dietary diversity score” (WDDS) [44, 45]. The WDDS score ranges from 0 to 9 and includes the following aggregated healthy food groups, with 0 reflecting no consumption of any food group and 9 reflecting consumption of all healthy food groups with at least one food item per group: starchy samples (cereals and white roots and tubers); vitamin A rich fruits and vegetables; green, leafy vegetables; other vegetables and fruits; organ meat; meat, and fish; eggs; legumes (including legumes, nuts, and their products); and milk (including milk and milk products). To calculate the WDDS, one point was given for each healthy food group consumed and lowest dietary diversity was defined as WDDS ≤ 3. Nutritional status The nutritional status of each child was assessed using body mass and height measurements, which were carried out by trained nurses and researchers. Children’s body mass (kg) was measured to the nearest 0.1 kg using Tanita scale (Tanita MC-580, Tanita Corp.; Tokyo, Japan), wearing only light sports clothing; and height was measured to the nearest 0.1 cm using a stadiometer with the child’s shoulders against a stadiometer. Based on these measurements, we used the classification of stunting, which is defined as being below − 2 SD of mean height-for-age-score (HAZ) of a reference population as recommended by the World Health Organization (WHO) [46]. Statistical analysis Descriptive statistics (M, SD, min, max, 95% CI) were calculated for all main study variables: food insecurity, consumption of food items from different food groups, and dietary diversity (as assessed with the WDDS). Furthermore, the percentage of stunted children was computed. Normal distribution of data was tested via Kolmogorov-Smirnov tests. To explore possible between-wealth quintile differences in food insecurity, dietary diversity, and consumption of food items from different food group categories, separate univariate analyses of covariance (ANCOVAs) were carried out, adjusting for age and sex as potential confounders. To interpret between-wealth group differences, the effect size (eta square) was calculated, referring to Cohen’s [47] classifications (η2 ≥ 0.01: small effect, η2 ≥ 0.06: moderate effect, η2 ≥ 0.14: large effect). To examine specific differences between any two groups, contrast analyses were performed to compare children from higher wealth quintile with children from lower wealth quintile, again referring to Cohen’s [47] classifications (d ≥ 0.2: small effect; d ≥ 0.5: medium effect; and d ≥ 0.80: large effect). All analyses were performed with SPSS 26 for Mac (IBM Corporation; Armonk, USA), and p-values of < 0.05 were considered as statistically significant. Results Sample characteristics A total of 1369 children were enrolled. Parents/guardians of 908 children completed the parental survey about household socioeconomic factors. 69 parents/guardians failed to complete or return the food item checklist and food security questionnaire. Hence, the final sample consisted of 839 children (416 girls and 423 boys). Univariate ANOVAs and χ2-tests showed that children included versus those excluded from the present analyses did not differ significantly with regard to sex, age, height, and stunting (all p > 0.05). The mean age of the children was M = 8.3±1.4 years, the mean height was M = 124.9±9.1 cm, and the mean body mass was M = 25.7±7.1 kg. In total, 8.4% (n = 67, 95% CI: 6.6–10.6%) of the children were classified as stunted (overall 45 children had missing data). Most children lived in Reconstruction and Development Program (RDP) houses (38.1%, 95% CI: 34.7–41.8%) or council houses (24.5%, 95% CI: 21.4–27.7%), with access to electricity (98.3%, 95% CI: 97.1–99.0%) and a flush toilet inside the house (91.5%, 95% CI: 89.5–93.3%). On average, each household had M = 2.4±1.2 bedrooms, and M = 6.4±2.7 people lived in single households. Distribution of household material goods and characteristics based on wealth quintiles In our sample, 80.8% (95% CI: 78.3–83.7%) of the households had a TV, 80.5% (95% CI: 77.8–83.0%) had a cell phone, 72.9% (95% CI: 70.2–76.3%) had a fridge, 60.4% (95% CI: 57.1–63.8%) had access to running water inside the house, 57.0% (95% CI: 53.6–60.4%) had a bathroom/shower inside the house, 50.7% (95% CI: 47.4–54.3%) had washing machine, 49.5% (95% CI: 46.4–53.3%) had an electric oven, 25.6% (95% CI: 22.2–28.2%) had a freezer, and 16.2% (95% CI: 13.5–18.6%) had a fan. Figure 1 shows that between the five wealth quintiles, there were large differences with regard to the number of household material goods/positive household characteristics, F(4,834) = 592.6, p < 0.001, η2 = 0.740. Contrast analyses showed that these differences between all quintiles were statistically significant (p < 0.001), with effect sizes varying between d = 0.82 (Q2 vs. Q3) and d = 5.50 (Q1 vs. Q5). Fig. 1Between-wealth quintile differences in total number of possessed material goods and positive household characteristics in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019 Between-wealth quintile differences in total number of possessed material goods and positive household characteristics in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019 Table 1Descriptive statistics for the total sample for food insecurity, dietary diversity, and consumption of food items from different food group categories in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019 N MSDMinMaxSkewKurtn (%)with low dietary diversity295% CIFood insecurity7881.841.18061.200.86Dietary diversityTotal number of consumed food items183930.3922.6601261.271.71Number of consumed healthy food groups (WDDS)28395.841.9908–0.940.31113 (13.5)11.0–15.8Consumption of food items from different food group categories1. Cereals8394.143.120222.01 (0.10)7.30(–0.51)802 (95.6)94.4–97.02. Roots8390.870.4802–0.340.89683 (81.4)78.5–84.23. Vitamin A rich vegetables8391.251.15040.72–0.27577 (68.8)65.8–71.84. Green, leafy vegetables8390.550.90041.762.54288 (34.3)31.0–37.55. Vegetables8394.013.300181.332.13765 (91.2)89.3–93.16. Vitamin A rich fruits8390.931.77051.250.88422 (50.3)47.0–54.07. Fruits8393.863.600161.181.13711 (84.7)82.3–87.18. Meat8390.680.84031.190.78405 (48.3)44.8–51.79. Flesh meat8394.794.620251.542.63777 (92.6)90.9–94.310. Eggs8390.660.4701–0.67–1.56552 (65.8)62.7–69.011. Fish8390.791.07051.572.56394 (47.0)43.7–50.412. Legumes8391.761.840122.27 (1.10)7.92 (1.02)643 (76.6)73.7–79.413. Milk8392.751.99090.710.27737 (87.8)85.6–90.014. Oil8391.621.68081.091.13548 (65.3)62.1–68.615. Sweets8393.402.75090.57–0.86719 (85.7)83.4–87.916. Sweet beverages8391.881.81071.160.73753 (89.7)87.7–91.7Descriptive statistics are defined as: Number (N), mean (M), standard deviation (SD), minimum (Min), and maximum (Max). Skewness and Kurtosis values in brackets represent scores after log-transformation. 1Possible range from 0 to 143 items. 2Woman dietary diversity score (WDDS) is based on (healthy) food group categories 1 (cereals) to 13 (milk) with a total of 9 aggregated food groups (possible range from 0 to 9 points). Low dietary diversity is defined as consuming less than 4 (out of 9) aggregated healthy food groups based on the WDDS [44] Table 1Descriptive statistics for the total sample for food insecurity, dietary diversity, and consumption of food items from different food group categories in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019 N MSDMinMaxSkewKurtn (%)with low dietary diversity295% CIFood insecurity7881.841.18061.200.86Dietary diversityTotal number of consumed food items183930.3922.6601261.271.71Number of consumed healthy food groups (WDDS)28395.841.9908–0.940.31113 (13.5)11.0–15.8Consumption of food items from different food group categories1. Cereals8394.143.120222.01 (0.10)7.30(–0.51)802 (95.6)94.4–97.02. Roots8390.870.4802–0.340.89683 (81.4)78.5–84.23. Vitamin A rich vegetables8391.251.15040.72–0.27577 (68.8)65.8–71.84. Green, leafy vegetables8390.550.90041.762.54288 (34.3)31.0–37.55. Vegetables8394.013.300181.332.13765 (91.2)89.3–93.16. Vitamin A rich fruits8390.931.77051.250.88422 (50.3)47.0–54.07. Fruits8393.863.600161.181.13711 (84.7)82.3–87.18. Meat8390.680.84031.190.78405 (48.3)44.8–51.79. Flesh meat8394.794.620251.542.63777 (92.6)90.9–94.310. Eggs8390.660.4701–0.67–1.56552 (65.8)62.7–69.011. Fish8390.791.07051.572.56394 (47.0)43.7–50.412. Legumes8391.761.840122.27 (1.10)7.92 (1.02)643 (76.6)73.7–79.413. Milk8392.751.99090.710.27737 (87.8)85.6–90.014. Oil8391.621.68081.091.13548 (65.3)62.1–68.615. Sweets8393.402.75090.57–0.86719 (85.7)83.4–87.916. Sweet beverages8391.881.81071.160.73753 (89.7)87.7–91.7Descriptive statistics are defined as: Number (N), mean (M), standard deviation (SD), minimum (Min), and maximum (Max). Skewness and Kurtosis values in brackets represent scores after log-transformation. 1Possible range from 0 to 143 items. 2Woman dietary diversity score (WDDS) is based on (healthy) food group categories 1 (cereals) to 13 (milk) with a total of 9 aggregated food groups (possible range from 0 to 9 points). Low dietary diversity is defined as consuming less than 4 (out of 9) aggregated healthy food groups based on the WDDS [44] Descriptive statistics for the total sample for food insecurity, dietary diversity, and consumption of food items from different food group categories in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019 7.30 (–0.51) Descriptive statistics are defined as: Number (N), mean (M), standard deviation (SD), minimum (Min), and maximum (Max). Skewness and Kurtosis values in brackets represent scores after log-transformation. 1Possible range from 0 to 143 items. 2Woman dietary diversity score (WDDS) is based on (healthy) food group categories 1 (cereals) to 13 (milk) with a total of 9 aggregated food groups (possible range from 0 to 9 points). Low dietary diversity is defined as consuming less than 4 (out of 9) aggregated healthy food groups based on the WDDS [44] Descriptive statistics Table 1 provides an overview of the descriptive statistics for food insecurity, dietary diversity, and consumption of items from different food group categories. The mean for food insecurity was 1.84±1.18, reflecting a low insecurity level. The mean number of different food items consumed during the last 24 h was 30.4±22.7 items. Cereals, flesh meat, vegetables, fruits, and sweet food items were the most frequently consumed food groups. Regarding the WDDS, on average, 5.8±1.99 out of 9 healthy, aggregated food groups were consumed, displaying a medium healthy food group consumption. Based on Kolmogorov-Smirnov tests, none of the assessed dietary variables were normally distributed. Severe non-normality (skewness and kurtosis values of ≥|2| and ≥|7|, respectively) [48] was found in two variables (cereals and legumes). Hence, these two variables were log-transformed before inferential statistics were calculated. Table 2Differences between wealth quintiles in food insecurity, dietary diversity, and consumption of food items from different food group categories in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019Quintile 1 (n = 167)Quintile 2 (n = 163)Quintile 3 (n = 170)Quintile 4 (n = 164)Quintile 5 (n = 175)Model 1: UncontrolledModel 2: Controlled for age, and genderMSDMSDMSDMSDMSDFη2Fη2Food insecurity2.16abc1.261.92d1.231.77ae1.121.88bf1.211.48cdef0.987.34***0.0367.34***0.036Dietary diversityTotal number of consumed food items126.97ab19.6030.2324.0728.2720.3633.82a23.5232.71b24.832.74*0.0132.73*0.013Number of healthy food groups (WDDS)25.961.995.831.995.811.955.971.905.642.130.790.0040.800.004Consumption of food items from different food group categories1. Cereals3.41abcd2.334.14a3.474.02be2.864.77ce3.474.39d3.234.47**0.0214.47**0.0212. Roots0.72abcd0.510.93a0.490.92b0.410.89c0.510.90d0.455.77***0.0275.86***0.0273. Vitamin A rich vegetables1.04ab1.051.211.111.221.131.40a1.201.37b1.232.75*0.0132.73*0.0134. Green, leafy vegetables0.500.750.540.950.510.810.580.950.621.030.530.0030.530.0035. Vegetables3.44ab3.054.033.333.752.834.24a3.554.57b3.603.01*0.0143.01*0.0146. Vitamin A rich fruits0.861.020.911.200.81a1.071.09a1.240.971.321.450.0071.400.0077. Fruits3.633.123.883.583.42a3.134.35a3.844.034.171.690.0081.650.0088. Meat0.670.860.670.880.650.780.740.840.650.860.310.0010.320.0029. Flesh meat4.31a4.094.644.844.41b4.265.49ab5.135.094.651.910.0091.890.00910. Eggs0.680.470.650.480.650.480.650.480.650.480.150.0010.140.00111. Fish0.720.920.731.110.700.980.901.180.881.111.370.0071.360.00612. Legumes1.87a1.601.98b2.121.45ab1.471.701.931.812.022.010.0102.300.01113. Milk2.52a1.802.742.062.65b1.973.09ab2.102.752.011.820.0091.850.00914. Oil1.31ab1.591.50cd1.781.49ef1.541.92ace1.751.86bdf1.674.15**0.0204.14**0.02015. Sweets2.93ab2.783.352.583.16cd2.643.84ac2.823.70bd2.843.12*0.0153.13*0.01516. Sweet beverages1.50ab1.691.84c1.841.74d1.702.05a1.822.25bcd1.934.35**0.0204.35**0.021abcd Mean values are significantly different between quintiles if they have the same superscript letters, based on contrast analyses (p < 0.05). 1Possible range from 0 to 143 items. 2Woman dietary diversity score (WDDS) is based on (healthy) food group categories 1 (cereals) to 13 (milk) with a total of 9 aggregated food groups (possible range from 0 to 9 points). Low dietary diversity is defined as consuming less than 4 (out of 9) aggregated healthy food groups based on the WDDS [44]. *p < 0.05. **p < 0.01. ***p < 0.001 Table 2Differences between wealth quintiles in food insecurity, dietary diversity, and consumption of food items from different food group categories in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019Quintile 1 (n = 167)Quintile 2 (n = 163)Quintile 3 (n = 170)Quintile 4 (n = 164)Quintile 5 (n = 175)Model 1: UncontrolledModel 2: Controlled for age, and genderMSDMSDMSDMSDMSDFη2Fη2Food insecurity2.16abc1.261.92d1.231.77ae1.121.88bf1.211.48cdef0.987.34***0.0367.34***0.036Dietary diversityTotal number of consumed food items126.97ab19.6030.2324.0728.2720.3633.82a23.5232.71b24.832.74*0.0132.73*0.013Number of healthy food groups (WDDS)25.961.995.831.995.811.955.971.905.642.130.790.0040.800.004Consumption of food items from different food group categories1. Cereals3.41abcd2.334.14a3.474.02be2.864.77ce3.474.39d3.234.47**0.0214.47**0.0212. Roots0.72abcd0.510.93a0.490.92b0.410.89c0.510.90d0.455.77***0.0275.86***0.0273. Vitamin A rich vegetables1.04ab1.051.211.111.221.131.40a1.201.37b1.232.75*0.0132.73*0.0134. Green, leafy vegetables0.500.750.540.950.510.810.580.950.621.030.530.0030.530.0035. Vegetables3.44ab3.054.033.333.752.834.24a3.554.57b3.603.01*0.0143.01*0.0146. Vitamin A rich fruits0.861.020.911.200.81a1.071.09a1.240.971.321.450.0071.400.0077. Fruits3.633.123.883.583.42a3.134.35a3.844.034.171.690.0081.650.0088. Meat0.670.860.670.880.650.780.740.840.650.860.310.0010.320.0029. Flesh meat4.31a4.094.644.844.41b4.265.49ab5.135.094.651.910.0091.890.00910. Eggs0.680.470.650.480.650.480.650.480.650.480.150.0010.140.00111. Fish0.720.920.731.110.700.980.901.180.881.111.370.0071.360.00612. Legumes1.87a1.601.98b2.121.45ab1.471.701.931.812.022.010.0102.300.01113. Milk2.52a1.802.742.062.65b1.973.09ab2.102.752.011.820.0091.850.00914. Oil1.31ab1.591.50cd1.781.49ef1.541.92ace1.751.86bdf1.674.15**0.0204.14**0.02015. Sweets2.93ab2.783.352.583.16cd2.643.84ac2.823.70bd2.843.12*0.0153.13*0.01516. Sweet beverages1.50ab1.691.84c1.841.74d1.702.05a1.822.25bcd1.934.35**0.0204.35**0.021abcd Mean values are significantly different between quintiles if they have the same superscript letters, based on contrast analyses (p < 0.05). 1Possible range from 0 to 143 items. 2Woman dietary diversity score (WDDS) is based on (healthy) food group categories 1 (cereals) to 13 (milk) with a total of 9 aggregated food groups (possible range from 0 to 9 points). Low dietary diversity is defined as consuming less than 4 (out of 9) aggregated healthy food groups based on the WDDS [44]. *p < 0.05. **p < 0.01. ***p < 0.001 Differences between wealth quintiles in food insecurity, dietary diversity, and consumption of food items from different food group categories in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019 abcd Mean values are significantly different between quintiles if they have the same superscript letters, based on contrast analyses (p < 0.05). 1Possible range from 0 to 143 items. 2Woman dietary diversity score (WDDS) is based on (healthy) food group categories 1 (cereals) to 13 (milk) with a total of 9 aggregated food groups (possible range from 0 to 9 points). Low dietary diversity is defined as consuming less than 4 (out of 9) aggregated healthy food groups based on the WDDS [44]. *p < 0.05. **p < 0.01. ***p < 0.001 Between-quintile differences in food insecurity, dietary diversity, and consumption of food items from different food group categories Table 2 provides a summary of the M and SD, separately for the five quintiles. The ANCOVAs showed significant main effects for wealth with regard to food insecurity, total number of consumed food items, and mean consumed food items from the following food group categories: cereals, roots, vitamin A rich vegetables, vegetables, oil, sweets, and sweet beverages. Contrast analyses (Table 3) highlighted that those children in the highest wealth quintile were exposed to lower food insecurity (p < 0.001), and that they consumed significantly more food items (p < 0.05) than those children in the lowest wealth quintile. Specifically, children in the highest wealth quintile consumed significantly more food items from the food group categories cereals, roots, vitamin A rich vegetables, vegetables, oil, sweets, and sweet beverages (p < 0.01). Based on Cohen’s [47] standards, all effects were small or medium. Table 2 further shows that descriptively, children classified in quintile 4 (not quintile 5) had the highest scores for total number of consumed food items, cereals, vitamin A rich vegetables, vitamin A rich fruits, fruits, meat, fish, milk, oil, and sweets. Contrast analyses also showed that mean scores were significantly different if compared to quintile 1. Table 3Differences in food insecurity, dietary diversity, and consumption of food items from different food group categories between the lowest and highest wealth quintile in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019Contrast betweenquintile 5 and 1T p dFood insecurity-5.25< 0.001-0.60Dietary diversityTotal number of consumed food items12.350.0190.26Number of healthy food groups (WDDS)2-1.500.133-0.16Consumption of food items from different food group categories1. Cereals3.110.0020.342. Roots3.590.0000.383. Vitamin A rich vegetables2.700.0070.294. Green, leafy vegetables1.230.2210.135. Vegetables3.190.0010.346. Vitamin A rich fruits0.850.3910.097. Fruits1.040.2980.118. Meat-0.270.782-0.039. Flesh meat1.550.1210.1810. Eggs-0.610.544-0.0711. Fish1.400.1610.1612. Legumes-0.760.446-0.0813. Milk2.080.2790.1214. Oil3.060.0020.3415. Sweets2.600.0100.2716. Sweet beverages3.85< 0.0010.41Cohen’s d should be interpreted as follows: d < 0.2 (negligible effect); d ≥ 0.2 (small effect); d ≥ 0.5 (medium effect); and d ≥ 0.80 (large effect). 1Possible range from 0 to 143 items. 2Woman dietary diversity score (WDDS) is based on (healthy) food group categories 1 (cereals) to 13 (milk) with a total of 9 aggregated food groups (possible range from 0 to 9 points). Low dietary diversity is defined as consuming less than 4 (out of 9) aggregated healthy food groups based on the WDDS [44] Table 3Differences in food insecurity, dietary diversity, and consumption of food items from different food group categories between the lowest and highest wealth quintile in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019Contrast betweenquintile 5 and 1T p dFood insecurity-5.25< 0.001-0.60Dietary diversityTotal number of consumed food items12.350.0190.26Number of healthy food groups (WDDS)2-1.500.133-0.16Consumption of food items from different food group categories1. Cereals3.110.0020.342. Roots3.590.0000.383. Vitamin A rich vegetables2.700.0070.294. Green, leafy vegetables1.230.2210.135. Vegetables3.190.0010.346. Vitamin A rich fruits0.850.3910.097. Fruits1.040.2980.118. Meat-0.270.782-0.039. Flesh meat1.550.1210.1810. Eggs-0.610.544-0.0711. Fish1.400.1610.1612. Legumes-0.760.446-0.0813. Milk2.080.2790.1214. Oil3.060.0020.3415. Sweets2.600.0100.2716. Sweet beverages3.85< 0.0010.41Cohen’s d should be interpreted as follows: d < 0.2 (negligible effect); d ≥ 0.2 (small effect); d ≥ 0.5 (medium effect); and d ≥ 0.80 (large effect). 1Possible range from 0 to 143 items. 2Woman dietary diversity score (WDDS) is based on (healthy) food group categories 1 (cereals) to 13 (milk) with a total of 9 aggregated food groups (possible range from 0 to 9 points). Low dietary diversity is defined as consuming less than 4 (out of 9) aggregated healthy food groups based on the WDDS [44] Differences in food insecurity, dietary diversity, and consumption of food items from different food group categories between the lowest and highest wealth quintile in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019 Cohen’s d should be interpreted as follows: d < 0.2 (negligible effect); d ≥ 0.2 (small effect); d ≥ 0.5 (medium effect); and d ≥ 0.80 (large effect). 1Possible range from 0 to 143 items. 2Woman dietary diversity score (WDDS) is based on (healthy) food group categories 1 (cereals) to 13 (milk) with a total of 9 aggregated food groups (possible range from 0 to 9 points). Low dietary diversity is defined as consuming less than 4 (out of 9) aggregated healthy food groups based on the WDDS [44] Between-quintile differences in stunting As shown in Fig. 2, the percentage of stunted children among the three higher wealth quintiles was lower compared to children in the two lowest wealth quintiles, χ2(4,794) = 10.11, p < 0.05. A separate comparison between children from the lowest and highest quintile highlights a significant between-group difference in socioeconomic status, χ2(4,327) = 5.09, p < 0.05, in that quintile 1 children were more likely to be stunted than children in quintile 5. Fig. 2Percentage of stunted children by wealth quintile in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019 Percentage of stunted children by wealth quintile in a sample of school-age children from marginalized, peri-urban primary schools in Gqeberha, South Africa in 2019 Discussion The present study examined wealth differences in dietary diversity and food consumption in the Gqeberha region of South Africa, playing particular emphasis on marginalized, peri-urban areas. The key findings are that among school-age children living in a peri-urban area, differences in household wealth status are associated with differences in food security and consumption of food items from different food group categories. Specifically, higher household wealth status was related to higher food security, and consumption of more unhealthy food items. Children from wealthier households were less likely to be stunted than children from households with lower wealth. Moreover, children from families with the highest wealth consumed a greater number of both healthy (e.g., cereals, roots, and vegetables) and unhealthy food items (e.g., oil, sweets, and sweetened beverages) than children from the lowest wealth quintile. As a result of the rapid socioeconomic and nutrition transition in South Africa, there is a double burden of malnutrition of stunting and obesity among school-age children and this is of considerable public health relevance. Although dietary behavior is a complex phenomenon and depends on multiple factors [17], we hypothesized that children from wealthier households would consume a more diverse diet and a higher number of food items compared to their peers from less wealthy households. These hypothesized relationships were partly supported by our data, as we observed that household socioeconomic differences are associated with food variety, dietary diversity, and food item consumption. While no differences in the absolute number of consumed healthy food groups was observed, number of food items from different food group categories differed significantly between wealth quintile groups, indicating that children from higher wealth quintiles consume at the same time more food items from healthy and unhealthier food group categories. The studied community is marked by a low resourced area, which means similar food access and availability for all participants. A systematic review on socioeconomic determinants of dietary patterns in LMICs reported similar results where, on one hand, higher socioeconomic status had positive effects on food security and overall food consumption, and on the other hand, the risk for a high-energy and high-fat diet was increased for middle-income countries [17]. Several possible explanations are offered for discussion. First, in our study, children from higher wealth households consumed more animal-based food items, sweets, and sweetened beverages, which can be considered as luxury items. Indeed, FAO described income and food prices as the two most influencing factors driving dietary diversity in LMICs [49]. Additionally, higher income has been linked to higher educational attainment [12], and both factors are associated with enhanced food purchase and nutritional status [50]. In contrast, in marginalized communities, low socioeconomic households still struggle to obtain access to sufficient food and fulfill basic energy requirements. This often results in the consumption of a diet that is strongly based on high energy, starchy foods (particularly cereals), and energy-dense foods. For many households, starchy food items are the most affordable choice, and deemed most suitable to achieve fast saturation for their children [51]. As a consequence, a preference for foods with lower nutritional value appeared among lower socioeconomic groups, which can be referred to the lower costs of these food items [51]. Thus, one possible approach to overcome socioeconomic-based differences in nutritional status and deficiencies is food fortification. In line with this, in South Africa, the food system is addressing multi-micronutrient deficiencies by fortifying maize meal with vitamin and mineral mix and salt with iodine [52]. Even though food fortification is just one method, these products might reach most households in most areas, although not providing sufficient daily nutrient intake for all. Furthermore, school-based nutritional and health interventions is another approach addressing nutritional inequalities and all forms of malnutrition [53]. In the long run, however, food fortification and school-based nutrition intervention seem to be a suitable approach to address some deficiencies independently of socioeconomic status [54]. Second, research also supports the assumption that priorities of food choice and consumption differ depending on the area in which a child grows up [55]. For example, families living in urban areas reported a higher food security and availability compared to rural living families [17]. A recent study of Zimbabwean children and their families reported that households located in more urban communities and/or of higher wealth status had higher food security and availability [56]. Indeed, car ownership can influence the accessibility to larger supermarkets, which may contain a larger variety of specific food items. In line with this suggestion, greater access to cars and larger supermarkets should be interpreted cautiously, as such changes in the food environment may simultaneously facilitate access to healthy foods and increase exposure to ultra-processed foods and sugar-sweetened beverages, with potentially adverse implications for diet quality [57, 58]. However, as reported previously, only 15.6% of households own a car in our study sample [35]. Thus, it might be more difficult for low-income households to reach out-of-town supermarkets. Hence, to increase the availability of healthy food items among marginalized neighborhoods, future interventions could target accessibility to food markets or greater agroecological potential to reduce dependence on limited access to certain food groups [22]. Third, the situation is of public health concern given the recent transition toward physically inactive lifestyles and a tendency to eat more processed food observed in South Africa and elsewhere [59]. Indeed, younger children who consume a diet which is high in dense energy and fat, and low in fiber are at higher risk of being overweight and obese at a later age [60]. In high-income countries, higher socioeconomic households are more likely to be health-conscious and educated, and so consume healthier diets with more fruits and vegetables [61]. This is not comparable to the present study population, and highlights that families in the present area are living in marginalized areas. Children from the wealthiest quintile households did not necessarily have a healthier diet. This finding may be due to families from wealthier quintiles have adopted a more westernized, high-energy diet. This possible explanation is in line with other studies from other LMICs [28] showing that the risk of becoming overweight is particularly high among children from wealthier families, explained by the fast shift of dietary and lifestyle changes. The determinants of food choice are multifactorial, with the food environment [62], food prices [49], and education [29] having all been identified as key drivers of unhealthy dietary patterns among marginalized communities. While food prices are largely shaped by broader market forces and are not easily modified at the local level, health awareness, education initiatives, and policy regulations such as sugar taxes [63] may represent feasible strategies to address the adverse consequences of the ongoing dietary and lifestyle transition, to prevent the increasing burden of non-communicable diseases (NCDs) [59]. One potential approach is the implementation of school-based programs that provide healthy food options at the schools, such as fruits, either free of charge or at prices comparable to less healthy alternatives. Early interventions targeting primary school children and their parents may strengthen awareness of the long-term consequences of unhealthy food choices for physical and mental wellbeing as well as cognitive performance [31]. A conceptional framework of women’s food environment proposed by O’Meara et al. (2025) highlights the influence of social, household, and individual factors as determinants for personal food environment [62]. Given that behavior change [64] and dietary intake are strongly shaped by social contexts [7], upstream strategies that modify food environments and target key institutional settings such as schools can favorably play an important complementary approach to slowing the nutrition transition toward unhealthy diets. School-based interventions that combine provision of healthy meals with teacher-focused nutrition training and positive role modeling may be particularly effective, as evidence indicates that involving teachers and the broader school environment can positively influence children’s eating behaviors [65]. Despite the increasing prevalence of overnutrition, our analyses revealed that stunting is still of concern in peri-urban settings of South Africa. Specifically, stunting was more prevalent among children from households with lower wealth compared to their richer counterparts. These findings accord well with a previous review by Reyes et al. (2020), who highlighted the within-country socioeconomic inequalities in LMICs (using wealth deciles) across 49 countries [28]. The review found a clear association between lower socioeconomic status and higher prevalence of underweight. Similarly, a longitudinal study in South Africa [50] and a cross-sectional study in Nigeria [66] described the associations of socioeconomic status with nutritional status, in which prevalence of stunting and underweight were decreased for each category of increased socioeconomic status. It is important to understand when considering the broader environment that stunting results from multi-causal factors which explain continued growth impairment [67]. In summary, the social disparities in nutritional status observed in our study represent the difficulties of low resourced areas to ensure adequate growth in school children’s life. To tackle social disparities, South Africa has one of the most developed social protection systems in sub-Saharan Africa, and these programmes play a central role in supporting child nutrition. Longitudinal evidence indicates that children who receive the Child Support Grant early in life and without interruption improved linear growth and had a reduced likelihood of stunting, suggesting a sustained protective effect against undernutrition [68]. Another social protection system is the National School Nutrition Programme, which reaches approximately nine million learners and provides a daily meal at school. Evaluations report improvements in attendance and learning outcomes, while also documenting uneven implementation quality across provinces [69]. In the Eastern Cape, where unemployment is high and poverty remains widespread, these programmes contribute substantially to safeguarding basic food access but often do not ensure adequate dietary diversity or fully prevent chronic undernutrition [70]. International evidence, including evaluations of Brazil’s Bolsa Família programme, similarly demonstrates improvements in child growth and household food security, with reductions in stunting among the most vulnerable populations. However, emerging data also indicate rising overweight and obesity in less vulnerable groups, reflecting a growing double burden of malnutrition [68]. Taken together, these findings suggest that while social protection measures mitigate extreme deprivation, they do not, in isolation, resolve structural inequalities or fully protect against the risks associated with the nutrition transition. Cash transfers and school feeding programmes can improve children’s nutritional status, but their long-term effectiveness depends on integration with broader policies addressing food environments, income generation, and local economic development in order to sustainably reduce inequalities in child nutrition. Strengths and limitations Our study provides insights in the association between socioeconomic factors, dietary diversity, and stunting among primary school children from low resourced areas in the Gqeberha region. A distinct strengths of this study is that possessed household assets and housing characteristics were utilized to assess wealth quintiles, which is widely used in LMICs research and can be easily implemented in household surveys. Assets tend to approximate longer-term living standards and permanent income [71]. The asset-based wealth index represents one important dimension of socioeconomic status, complementary to income and education. Furthermore, our study population lives in a peri-urban setting that has been described as one of the poorest areas in Gqeberha in the Eastern Cape province of South Africa, making PCA a useful method to characterize this narrowly distributed population and to distinguish between poor and the very poorest households [72]. Another strength is the use of a context-specific wealth index that incorporates locally relevant material possessions, consistent with evidence that in very poor settings, the absence of a single key item can signal extreme deprivation [73]. Moreover, wealth quintiles were derived by a PCA approach, which is considered as gold standard to create a wealth index and has been widely used by the World Bank and the United Nations to monitor socioeconomic development in LMICs [5]. Importantly, evidence of the validity of similar socioeconomic measures has been documented in previous studies [5, 74]. Despite these strengths, our study is not without limitations. Although the DDS has been applied in several settings [40] and is a validated tool for cost-effectively measuring food intake [39], the results in the present study need to be interpreted with caution. Even though the present DDS is similar to other studies [37, 43], the parent-reported consumption of food items evaluated by the food item checklist is relatively high considering that households in low resourced settings are more likely to be food insecure than households with greater resources [17]. There are different ways to measure DDS [37, 43]. Especially, the recall period should be considered. Thus, while a 24–hour recall is able to distinguish between a basic diet and a better diet [37, 41], focusing on one single day might result in unusually high or low food item consumption and should therefore be interpreted with caution. The results may also indicate a tendency towards overreporting of food items. One explanation for this possible overreporting of certain foods is that they were all included in the same mixed dish. In previous research, social desirability has been identified as another factor that influences estimation of food item consumption and should be considered when describing health risk estimates [75], especially when instruments are self-administered. In our study, parents may have wanted to portray a rather good picture of the food intake in the home of their child. The fact that the language used was not the first language of the participants may also have played a role, but most of the parents/guardians understand English well, and were familiar with the names of the various foods. Hence, we believe that this factor was not a major reason for the rather high number of reported food items. We also acknowledge that even though our food item checklist captures a range of consumed food items, precise information about portion size is missing. In other words, information about food items alone does not allow a conclusion about children’s overall food and energy intake. Information of micronutrient status would be one approach to gain deeper insights about the source of misreporting of food group consumption [76]. Another approach to minimize potentially unmeasured effects on food insecurity and related inequalities would be to assess and adjust for school meal provision. To better identify explanatory factors underlying wealth differences future research should include data on income, employment, household crowding, and migrant status to better reflect a peri-urban population wealth. Another limitation is that our study focused exclusively on quintile 3 public schools in peri-urban areas. Hence, further research is needed in private schools, in rural schools and other public schools (quintiles 1–2 and 4–5) to further establish the association between socioeconomic factors, food security, and dietary diversity. Finally, since our analyses are based on cross-sectional data, causal interpretations of the associations between wealth status, food insecurity and dietary diversity is not possible. Hence, with respect to the currently ongoing multi-micronutrient supplementation intervention, it will be interesting to see whether children with lower wealth status will benefit in particular from such intervention measures. Conclusions Our data highlight that children from wealthier families consume a higher total number of food items (including both healthy and unhealthy food items), whereas children in the lowest wealth quintile consume less nutrient-dense foods and are more likely to be stunted than their peers in the highest wealth quintile. Children living in a low-resourced area in marginalized and peri-urban communities should be the focus of broader government-led programs concerned with nutrition education, school feeding policies, food availability, and improving food consumption choices. Our findings reiterate the need for improving awareness of healthy dietary behaviors. In a local context, we recommend expanding appropriate policies at a scale to prioritize nutrition-related programs to reduce existing inequalities and to identify children at risk of under- and overnutrition.