Nutrition and Development of Children in Foundational Learning Spaces in Johannesburg: A Cross-Sectional Study of Dietary Diversity and Nutritional Status

Abstract

Background: Foundational learning spaces in South Africa, designed to nurture growth and development, continue to grapple with malnutrition, a persistent barrier to the health, cognitive potential, and wellbeing of preschool-aged children, amidst a nutrition transition. Aim: This study assessed dietary diversity, nutritional status, and their associations among children aged 2–5 years attending funded Early Learning Centres (ELCs) in Johannesburg (Region C). Methods: Using systematic random sampling across 33 nutrition-funded ELCs in Region C, we assessed the nutritional status of children aged 2–5 years with WHO Anthro software (z-score cut-offs for undernutrition: stunting, underweight, thinness; overnutrition: overweight, obesity). Dietary diversity scores (DDSs) were derived from a 24 h recall of 16 food groups, classified by primary nutrient contributions (some groups spanning multiple classes), and categorized as low (≤8) or normal (≥9). Associations between DDS and nutritional indicators were analyzed using Poisson regression to estimate adjusted prevalence ratios (aPRs). Results: Despite structured feeding practices, all ELCs reported inadequate nutritional funding, prompting calls for dietitian support. While 27% sourced groceries from wholesalers, most relied on supermarkets and spaza shops; all had cooking infrastructure, but only 12% had food gardens, and 88% expressed interest in establishing them to improve dietary diversity. The mean DDS was 9.47 (±1.07), and 83% of children had a normal DDS (≥9), with common consumption of cereals (100%), vitamin A-rich vegetables (100%), oils (100%), and leafy greens (96%), but limited intake of protein-rich foods like eggs (7%), legumes (19%), and fish (37%). A dual burden of malnutrition was observed: 31% of children were stunted and 30% were overweight or obese. Multivariable analysis showed that boys had significantly lower odds of stunting compared to girls (aPR = 0.38; 95%CI: 0.01–0.74), while younger age (aPR = 0.61; 95%CI: 0.37–0.85) and low DDS (aPR = −0.15; 95%CI: −0.29–−0.06) were independently associated with increased risk of stunting. Age was positively associated with underweight (aPR = 1.27; 95%CI: 0.58–1.96), and thinness was strongly associated with boys (aPR = 17.00; 95%CI: 15.12–18.74), with a marginal association with age. Conclusions: Integrated nutrition strategies are critical to addressing the dual burden of stunting and being overweight in urban ELCs. Strengthening funding, professional dietetic support, and promoting food gardens can enhance dietary diversity and child health outcomes.

1. Introduction

Nutrition in early childhood is critical for lifelong health, learning, and development [1]. In low- and middle-income countries (LMICs), children under five face high rates of undernutrition and poor diet quality, with dietary diversity emerging as a key predictor of micronutrient adequacy and developmental outcomes [2,3]. Dietary diversity has been widely used as a proxy indicator for micronutrient adequacy in children, particularly in LMICs [2,3]. Studies have shown that higher dietary diversity scores (DDS) are positively associated with improved intake of essential vitamins and minerals, especially in non-breastfed children aged 2–5 years [4]. While DDS does not capture nutrient density or food processing levels, it remains a practical and validated tool for assessing diet quality in resource-limited settings [5]. Inadequate dietary diversity, marked by reliance on starchy staples and limited intake of nutrient-dense foods, has been linked to stunting, underweight, wasting, and micronutrient deficiencies [5,6,7,8]. Children exposed to inadequate diets during critical growth periods face increased risks of infections, developmental delays, impaired immune and cognitive function, and chronic diseases later in life [9,10,11]. In Sub-Saharan Africa (SSA), structural challenges such as poverty, food insecurity, and inadequate health and sanitation services continue to drive high rates of child malnutrition [12], compounded by low caregiver nutrition knowledge, poor household food environments, and inconsistent feeding practices in early learning settings [13,14].

In South Africa, Early Learning Centres (ELCs) are foundational spaces for nurturing children aged 0–5 years, supporting their cognitive, emotional, physical, and social development [15,16]. These centres are especially vital for children from disadvantaged backgrounds, offering structured environments that promote school readiness and social integration [17,18]. However, many ELCs face challenges such as inadequate funding, limited resources, and untrained staff, which can compromise the quality of care and education [19,20]. Although national programmes like the Integrated Nutrition Programme (INP) and the National School Nutrition Programme (NSNP) aim to improve child nutrition, their implementation in ELCs remains inconsistent, particularly in low-income communities [21,22,23]. Evidence shows that children attending ELCs with diverse and balanced meal plans have better nutritional status and are less likely to suffer from stunting, underweight, or micronutrient deficiencies [23,24,25].

Despite South Africa’s commitment to improving child nutrition through national programmes, stunting remains a persistent and complex public health challenge, particularly in children aged below five years [26,27,28]. Recent evidence suggests that stunting is increasingly observed in urban settings, including among children attending ELCs [29,30]. This observed paradoxical rise in stunting, alongside increasing rates of overweight and obesity, reflects the dual burden of malnutrition [28,31,32], where the causes remain multidimensional, including poor dietary diversity, inadequate institutional feeding practices, and limited caregiver nutrition knowledge [12,24,33]. Additionally, this phenomenon is not isolated but indicative of a broader nutrition transition occurring nationally and across LMICs, marked by increasing consumption of ultra-processed foods and a gradual erosion of traditional, nutrient-rich diets [34,35,36].

Although stunting among children in South Africa declined slightly from 25% in 2008 to 23% in 2017 [29], SSA data shows that the prevalence has surged to over 50%, driven by persistent maternal and household risk factors such as poor maternal nutrition, low education, poverty, and lack of medical insurance [37,38]. However, the lack of empirical data on dietary patterns and nutritional outcomes in foundational learning spaces hinders the development of targeted interventions. Therefore, this study assessed dietary diversity, nutritional status, and their associations among children aged 2–5 years attending funded ELCs in Johannesburg (Region C). The findings will contribute to targeted strategies to improve dietary diversity and child health, supporting progress toward Sustainable Development Goals (SDGs) 2 (Zero Hunger) and 3 (Good Health and wellbeing).

2. Materials and Methods

2.1. Study Design and Conceptual Frameworks

This study used a cross-sectional analytical design to examine the relationship between dietary diversity and nutritional status among children aged 2–5 years attending ELCs in Region C of the City of Johannesburg Metropolitan Municipality, Gauteng Province, South Africa. The study was conceptually informed by the UNICEF and WHO frameworks on the determinants of child malnutrition, which delineate immediate causes (e.g., inadequate dietary intake and disease), underlying factors (e.g., food insecurity, caregiving practices, and access to health services), and structural determinants (e.g., poverty, education, and governance) [39,40,41]. These frameworks guided the selection of variables and analytical strategy, enabling a multidimensional assessment of nutritional risk in foundational learning spaces.

2.2. Study Setting and Population

Region C of the City of Johannesburg Metropolitan Municipality, Gauteng Province, South Africa, includes suburbs such as Roodepoort, Florida, and Bram Fischerville. These areas encompass several densely populated communities where socioeconomic challenges, including food insecurity and limited access to quality early childhood nutrition services, have been documented. The study population comprised children aged 2–5 years enrolled in 33 funded ELCs that receive nutrition funding from the Department of Basic Education as the inclusion criteria, in addition to children not being sick or diagnosed with any condition that affects appetite, as well as parental consent. These centres serve as critical platforms for foundational learning and early childhood development, making them ideal for assessing the intersection of nutrition and educational readiness [15,18,42]. The study specifically targeted children aged 2–5 years because this age range represents a critical window for physical, cognitive, and emotional development [16,43]. During this period, nutritional interventions have the greatest potential to influence long-term health outcomes, including growth, immunity, and school readiness [39,44]. Moreover, ELCs in South Africa are primarily designed to serve children within this age group, making them ideal platforms for assessing institutional feeding practices and their impact on child nutrition. This focus aligns with global frameworks from WHO and UNICEF, which prioritize early childhood as a key stage for addressing malnutrition and promoting optimal development [40].

2.3. Sample Size, Sampling Procedure and Recruitment

A systematic random sampling approach was applied across 33 nutrition-funded ELCs in Region C, targeting children aged 2–5 years. Selection followed an odd-number sequence from classroom registers (e.g., 1st, 3rd, 5th), with recruitment conducted mid-week (Tuesday–Thursday) to capture both institutional and home dietary intake. All parents or caregivers of eligible children were approached for written informed consent prior to sampling. Although the initial target was 15 children per centre, participation varied due to consent and attendance. Initially, we aimed to select a fixed proportion of children per ELC; however, variability in enrolment size, parental consent, and attendance made this infeasible. Consequently, the number of children sampled per centre varied, reflecting real-world constraints. Of 258 identified children, 26 did not participate due to non-consent or absence, resulting in a final sample of 232 (response rate: 90%), with minimal refusal rates across centres. According to Department of Social Development records, approximately 600 children were enrolled across 33 centres. Using the Raosoft Sample Size Calculator [45], a minimum sample of 235 was estimated based on this population (95% confidence level, 5% margin of error, 50% response distribution). The sample size achieved was determined by consent availability and attendance, resulting in uneven distribution across centres. Children from non-funded ELCs or outside the specified age range were excluded to maintain sample homogeneity. This approach ensured ethical compliance and alignment with study objectives.

2.4. Data Collection Instruments and Procedures

Data were collected by the two research assistants under the supervision of the main researcher, using a structured, interviewer-assisted questionnaire adapted from validated tools in maternal and child nutrition research [23,24,46]. The questionnaire was designed to capture a comprehensive range of variables, including demographic variables of children, dietary intake, ELC infrastructure, and administrative practices. To accommodate the linguistic diversity of the study population, the instrument was translated into Sepedi and IsiZulu, Two predominant local languages in the study region. Forward and backward translations were performed by independent bilingual translators fluent in both English, Sepedi and isiZulu, and the final version was approved by an expert committee to ensure semantic equivalence and cultural appropriateness [47].

Validation procedures were employed to ensure the reliability and accuracy of the instrument. Content validity was established through expert review, ensuring that the questionnaire items adequately captured the constructs of interest and were sufficient to measure the intended domains [47]. Face validity was confirmed through expert judgement, verifying that the questionnaire appeared to measure what it was designed to assess. A pilot study was conducted among 38 mothers at a non-participating facility to evaluate the feasibility, clarity, and contextual relevance of the instrument. This sample size aligns with literature recommendations for pilot testing in field-based studies [48]. Two research assistants fluent in Sepedi and isiZulu were trained in standardized interviewing techniques and anthropometric measurement protocols prior to the pilot phase. Their performance was assessed during the pilot to ensure consistency and accuracy in administering the questionnaire and recording measurements. Following the pilot, minor revisions were made to improve clarity of wording and layout, while the core content remained unchanged. This study adhered to the STROBE guidelines to ensure methodological transparency, completeness, and reproducibility [49].

2.4.1. Anthropometric Measurements

Anthropometric measurements were conducted by the principal investigator using calibrated equipment, following standardized protocols outlined by the World Health Organization and the Centers for Disease Control and Prevention for assessing child growth and nutritional status [44,50]. These measurements included weight, height/length, and mid-upper arm circumference (MUAC), which were used to derive standardized nutritional indicators such as weight-for-age (WAZ), length or height-for-age (LAZ/HAZ), weight-for-length (WLZ), and BMI-for-age (BAZ). Weight was recorded to the nearest 0.1 kg using a digital scale, and height/length to the nearest 0.1 cm using a stadiometer. MUAC was measured using a non-stretchable tape at the midpoint of the left arm. Nutritional indicators were calculated using WHO Anthro software (version 3.2.2), including weight-for-age (WAZ), length or height-for-age (LAZ/HAZ), weight-for-length (WLZ), MUACZ and BMI-for-age (BAZ). Nutritional indicators were derived using WHO Anthro software, with cut-offs defined as follows: stunting (LAZ/HAZ < −2 SD), underweight (WAZ < −2 SD), wasting (WLZ < −2 SD), overweight (WLZ > +2 SD), obesity (WLZ > +3 SD), and for mid-upper arm circumference-for-age z-scores (MUACZ), acute malnutrition was defined as MUACZ < −2 SD, moderate acute malnutrition (MAM) as −3 SD ≤ MUACZ < −2 SD, and severe acute malnutrition (SAM) as MUACZ < −3 SD [44,50].

2.4.2. Dietary Diversity Assessment

Dietary diversity was assessed using a 24 h recall method, capturing all foods consumed from breakfast to supper. Caregivers reported ELC meals, while parents/guardians provided home intake.

The 24 h recall method was selected for this study due to its practicality and validation in institutional settings such as ELCs. It minimizes recall bias and is feasible for caregivers and ELC staff who may not reliably recall food intake over longer periods [2,5]. Moreover, the study focused on institutional dietary intake rather than habitual household consumption, making the 24 h recall more appropriate for capturing structured feeding practices within ELCs.

The DDS tool was adapted from FAO and WHO guidelines to reflect local dietary patterns and institutional feeding practices [4,46]. Dietary diversity was assessed using a 16-food group DDS [4,46] with each food group scored to indicate variety and potential nutrient adequacy. To provide a nutrient-based perspective, food groups were classified into broader categories according to their primary contributions: carbohydrate-rich, protein-rich, fat-rich, vitamin-rich, mineral-rich, fibre-rich, and sugary/processed foods/other. Some food groups spanned multiple nutrient classes due to mixed composition. DDS was categorized as low (≤8) or normal (≥9). While DDS captures the diversity of food groups consumed, it does not account for portion sizes, nutrient density, or the level of food processing [2,5].

2.4.3. ELCs Infrastructure and Administrative Data

Information on monthly food expenses, transport costs, food sourcing patterns, and infrastructure was collected to contextualize dietary diversity and nutritional outcomes. Data included the type of food premises used (e.g., supermarkets, spaza shops, wholesalers), distance travelled for procurement, and the presence of cooking facilities and food gardens. Centres also reported on structured grocery lists, menu planning, and child-friendly meal practices. Additional details captured were the proportion of centres with food gardens and those expressing interest in establishing gardens to improve dietary diversity.

2.5. Data Analysis

All data were cleaned, coded, and analyzed using STATA version 18 (StataCorp LLC, College Station, TX, USA) [51]. Nutritional status indicators (LAZ/HAZ, WAZ, WHZ/WLZ, BAZ, and MUACZ) were calculated using WHO Anthro software (v3.2.2) based on WHO Child Growth Standards, excluding biologically implausible z-scores (±5 SD) as outliers.

Anthropometric indices: Normality was assessed using the Shapiro–Francia and skewness–kurtosis tests. These indices were non-normally distributed; therefore, descriptive statistics are presented as medians and interquartile ranges (IQR: 25th–75th percentile). Group comparisons for median anthropometric indices (LAZ/HAZ, WAZ, WHZ/WLZ, BAZ, and MUACZ) were performed using non-parametric tests: Wilcoxon rank-sum for sex and Kruskal–Wallis for age categories. Sex was categorized as boys and girls, and age into three groups: younger (2 years), middle (3 years), and older (4–5 years). Dietary diversity scores (DDS): DDS was approximately normally distributed; descriptive statistics are presented as means and standard deviations (SD). Group comparisons for means were performed using independent sample t-test for sex and one-way ANOVA for age categories. Categorical variables: Nutritional status categories (undernutrition: stunting, underweight, thinness; overnutrition: overweight, obesity) and DDS classifications by sex and age are presented as frequencies (n) and percentages (%). Associations were tested using chi-square tests; Fisher’s exact test was applied where expected cell counts were <5. Multivariable analysis: Associations between DDS and nutritional status indicators were assessed using Poisson regression to estimate adjusted prevalence ratios (aPR), rather than odds ratios, given the high prevalence of outcomes such as stunting and overweight/obesity. Variables with p < 0.25 in univariate analysis were included in the multivariable model to avoid excluding potential confounders. aPRs were estimated using a generalized linear model (GLM) with a Poisson distribution and log link, applying robust standard errors to correct for overdispersion. Results are presented as medians (IQR), means (SD), frequencies (n), percentages (%), prevalence ratios (PRs), and aPRs with 95% confidence intervals (CIs). Statistical significance was set at p < 0.05.

3. Results

3.1. Characteristic of ELCs

This section describes the administrative, infrastructural, and food-related characteristics of ELCs that participated in the study.

The study included 33 funded ELCs in Region C of Johannesburg, all reporting monthly food expenses exceeding R10,000 (≈$582.40). These centres maintained structured grocery lists, balanced menus, and child-friendly meals. Despite these efforts, all reported that nutritional funding was inadequate to sustain healthy meals, prompting a collective call for dietitian support. Most centres spent under R1000 (≈$58.24) on food transport, and while 27% sourced groceries from wholesalers, the majority relied on supermarkets and spaza shops. All had cooking infrastructure, but only 12% had food gardens, typically growing vitamin A-rich and dark green leafy vegetables. The remaining 88% expressed interest in establishing gardens to improve dietary diversity.

Table 1. Administrative and infrastructure indicators across ELCs in region C, Johannesburg.

Questions	Category	Frequency (n)	Proportion (%)
What are the average monthly expenses for food?	>R10,000 (≈$582.40)	33	100
What is your average monthly cost for transport when must you buy food?	<R1000 (≈$58.24) R1000–R5000 (≈$58.24–$291.20)	32 1	97 3
Do you have a specific food grocery list?	Yes	33	100
Where do you buy your groceries?	Wholesalers Supermarkets/spaza	9 24	27 73
Do you have a balanced diet menu?	Yes	33	100
Is food appealing enough to encourage children to consume all food items on the menu?	Yes	33	100
Is the nutritional funding enough to sustain healthy meals?	No	33	100
Do you need help from dietitians?	Yes	33	100
Do you have access to a food garden?	No Yes	29 4	88 12
What diverse food groups are planted in the food garden?	Group 2 (Vit. A rich vegetables/tubers) Group 4 (Dark green leafy vegetables)	3 1	9 3
Would you like access to a food garden to diversify your food?	Yes	33	100
Do you have electricity or gas to cook food?	Yes	33	100
Do you believe that you have an efficient system?	Yes	33	100

summarizes these findings and overall, the findings highlight that while ELCs had structured food systems, they faced financial constraints and lacked food gardens, indicating a need for external support to enhance dietary quality.

3.2. Characteristics of Children

Table 2. Age, anthropometric, and nutritional indicators of children.

Variables	Median	IQR	Minimum	Maximum
Age (months)	48	36; 48	24	60
Weight (kg)	16.1	14.5; 18.3	10.3	23.5
Length/height (cm)	95.7	91.2; 100.4	75.4	115.5
LAZ/HAZ	−0.82	−2.37; 0.97	−4.64	3.79
WAZ	0.51	−0.50; 1.51	−3.64	4.37
BAZ	1.51	0.57; 2.15	−1.88	4.30
WHZ/WLZ	1.47	0.53; 2.11	−1.89	4.44
MUACZ	0.40	−0.41; 1.15	−1.58	3.25

presents the demographic and nutritional profiles of the 232 children aged 2–5 years enrolled in funded ELCs across Region C, Johannesburg. The sample comprised 118 boys and 114 girls, with a median age of 48 months (IQR: 36; 48). Median anthropometric measurements included a weight of 16.1 kg, and a height of 95.7 cm. Nutritional indicators showed mild linear growth faltering (median LAZ/HAZ:−0.82), while weight-for-age scores (median WAZ: 0.51) were generally within normal ranges. However, elevated BMI-for-age (BAZ: 1.51) and weight-for-length (WLZ: 1.47) suggested a trend toward overweight. MUACZ ranged from−1.58 to 3.25, with a median of 0.40, indicating overall adequate nutritional status. These findings highlight a dual concern: while most children had sufficient weight-for-age, signs of linear growth faltering and increased BMI-related indicators highlight emerging risks of overnutrition.

3.3. Comparison of the Nutritional Indicators of Children

3.3.1. Comparison by Sex

Table 3. Comparison of the proportions of nutritional indicators of children by sex.

Variables	All	Boys	Girls
	n (%)	n (%)	n (%)	p-Value
LAZ/HAZ				0.088
Normal Stunting   Moderate   Severe Tallness	125 (56) 69 (31) 31 (14) 38 (17) 25 (11)	56 (50) 42 (39) 16 (14) 26 (23) 13 (12)	69 (64) 27 (25) 15 (14) 12 (11) 12 (11)
WAZ				0.376
Normal Underweight Growth problem	182 (78) 13 (6) 37 (16)	90 (77) 9 (8) 18 (15)	92 (80) 4 (3) 19 (17)
BAZ				0.161
Normal Thinness Overweight risk Overweight Obesity	81 (36) 2 (1) 73 (33) 44 (20) 23 (10)	33 (29) 2 (2) 42 (38) 23 (20) 12 (11)	48 (43) 0 31 (28) 21 (19) 11 (10)
WHZ/WLZ				0.233
Normal Acute malnutrition   MAM   SAM Overweight Obesity	161 (72) 2 (1) 1 (0.5) 1 (0.5) 36 (16) 25 (11)	78 (69) 2 (2) 1 (1) 1 (1) 22 (19) 11 (10)	83 (75) 0 0 0 14 (13) 14 (13)
MUACZ				0.426
Normal Acute malnutrition   Overnutrition	208 (90) 1 (0.5) 22 (9.5)	103 (88) 1 (1) 13 (11)	105 (92) 0 9 (8)

Malnutrition (poor nutritional status) was classified according to the WHO growth standards indicated in Section 2.5.

compares the proportions of nutritional indicators between boys and girls. Overall, 31% of children were stunted, with boys showing a higher prevalence (39%) than girls (25%), though not statistically significant (p = 0.088). Tallness (LAZ/HAZ > 2 SD) was observed in 11% of children, equally across sexes. Underweight affected 6% of the sample, while 16% had growth concerns (WAZ > 2 SD). Overnutrition was notable: 33% were at risk of overweight, 20% were overweight, and 10% were obese. Girls had a higher proportion of normal BAZ (43%) than boys (29%). Based on WHZ/WLZ, 29% had overnutrition, and acute malnutrition was rare (0.5–2%), with no sex differences. In summary, while sex-based differences in nutritional status were observed, particularly higher stunting and lower normal BAZ among boys, most were not statistically significant, suggesting broadly similar nutritional patterns across sexes.

3.3.2. Comparison of Age Groups

Table 4. Comparison of the proportions of nutritional indicators of children by age.

Variables	2 Years	3 Years	4–5 Years
	n (%)	n (%)	n (%)	p-Value
LAZ/HAZ				≤0.0001 *
Normal Stunting   Moderate   Severe Tallness	13 (39) 2 (6) 0 2 (6) 18 (55)	49 (78) 9 (14) 3 (5) 6 (10) 5 (8)	63 (51) 58 (47) 28 (23) 30 (24) 2 (2)
WAZ				≤0.0001 *
Normal Underweight Growth problem	19 (51) 0 18 (49)	49 (77) 0 15 (23)	114 (87) 13 (10) 4 (3)
BAZ				0.436
Normal Thinness Overweight risk Overweight Obesity	16 (44) 0 10 (28) 9 (25) 1 (3)	22 (37) 0 16 (27) 12 (21) 9 (15)	43 (33) 2 (2) 47 (37) 23 (18) 13 (10)
WHZ/WLZ				0.629
Normal Acute malnutrition   MAM   SAM Overweight Obesity	27 (72) 0 0 0 6 (17) 3 (8)	39 (65) 0 0 0 13 (22) 8 (13)	95 (74) 2 (2) 1 (1) 1 (1) 17 (13) 14 (10)
MUACZ				≤0.0001 *
Normal Acute malnutrition   Overnutrition	23 (63) 13 (36) 0	59 (92) 5 (8) 0	126 (96) 4 (3) 1 (1)

Malnutrition (poor nutritional status) was classified according to the WHO growth standards indicated in Section 2.5. * Indicates a statistically significant difference.

presents a comparison of the proportion of nutritional indicators across age group (2, 3, and 4–5 years). Significant age-related differences were observed for LAZ/HAZ, WAZ, and MUACZ (p ≤ 0.0001). Stunting was most prevalent among 4–5-year-olds (47%), while underweight was only observed in this group (10%). Growth concerns (WAZ > +2 SD) were highest among 2-year-olds (49%). Although BAZ differences were not statistically significant (p = 0.436), overweight risk was highest in the 4–5-year group (37%), and obesity was most common among 3-year-olds (15%). WHZ/WLZ showed no significant variation by age (p = 0.629), though overweight and obesity were more frequent in younger children. MUACZ revealed the highest rate of acute malnutrition among 2-year-olds (36%), decreasing with age. Overall, age-related disparities in nutritional status were evident, with younger children more affected by acute malnutrition and growth concerns, while older children showed higher rates of stunting and overweight risk.

3.4. Dietary Diversity and Food Group Consumption Among Children

This section presents dietary diversity scores (DDS) and food group consumption patterns among children, disaggregated by sex and age groups.

DDS (overall mean: 9.47 ± 1.07) were assessed across sex and age groups using a context-specific scoring system. Most children (83%) had normal DDS (≥9), with girls showing slightly higher scores than boys, though not statistically significant (

Table 5. Comparison of means and proportions of DDSs among children by sex groups.

DDS	All	Boys	Girls	p-Value
Mean	9.47 ± 1.07	9.35 ± 1.09	9.60 ± 1.05	0.073
Normal	191 (83)	94 (80)	97 (85)	0.343
Low	40 (17)	23 (20)	17 (15)

).

In

Table 6. Comparison of means and proportions of DDSs among children by age groups.

DDS	2 Years	3 Years	4–5 Years	p-Value
Mean	9.28 ± 1.11	9.70 ± 1.08	9.41 ± 1.05	0.108
Normal	26 (72)	57 (89)	108 (82)	0.102
Low	10 (28)	7 (11)	23 (18)

, age-related differences were more pronounced: 2-year-olds had the highest prevalence of low DDS (28%), compared to 11% among 3-year-olds and 18% among 4–5-year-olds. The results show that while dietary diversity was generally adequate, younger children, particularly 2-year-olds, were more likely to have limited food variety, highlighting early-life nutritional vulnerabilities.

In

Table 7. Proportions of each food group consumed by children across nutrient classes.

Nutrient Class	Food Groups	Proportions (%)
Carbohydrate-rich foods	Cereals, carbohydrates rich foods White roots and tubers	100 72
Protein-rich foods	Flesh meats Organic meats Fish Eggs Legumes/nuts/seeds Milk and milk products	74 46 37 7 19 84
Fat-rich foods	Oils and fats Milk and milk products Legumes/nuts/seeds	100 84 19
Vitamin-rich foods	Vitamin A, vegetables and tubers Dark green leafy vegetables Vitamin A-rich fruits Other fruits Other vegetables	100 96 68 0 0
Mineral-rich foods	Dark green leafy vegetables Milk and milk products Fish	96 84 37
fibre-rich foods	Cereals, carbohydrates rich foods Dark green leafy vegetables Legumes/nuts/seeds Other fruits Other vegetables	100 96 19 0 0
Sugary/Processed foods/other	Sweets Spices, condiments, and beverages	77 100

, the food groups are classified into broader nutrient categories based on their primary nutritional contributions, with some food groups classified in more than one nutrient class. Carbohydrate-rich foods were universally consumed, with 100% of children eating cereals and 72% consuming white roots and tubers. Protein-rich foods showed varied intake: milk and milk products (84%), flesh meats (74%), and fish (37%), while eggs (7%) and legumes/nuts/seeds (19%) were least consumed. Fat-rich foods were well represented, with 100% consuming oils and fats, and moderate intake of milk products (84%) and legumes/nuts/seeds (19%). Vitamin-rich foods included vitamin A vegetables and tubers (100%), dark green leafy vegetables (96%), and vitamin A-rich fruits (68%), but no children consumed other fruits or vegetables. Mineral-rich foods such as leafy greens (96%), milk products (84%), and fish (37%) were moderately consumed. Fibre-rich foods followed a similar pattern, with high intake of cereals (100%) and leafy greens (96%), but low intake of legumes (19%), and no consumption of other fruits or vegetables. Sugary and processed foods were common, with 77% consuming sweets and 100% consuming spices, condiments, and beverages. Data reflect adequate consumption of staple and micronutrient-rich foods, but highlight significant gaps in protein diversity, fruit intake, and fibre-rich sources, suggesting areas for targeted nutritional improvement.

3.5. Association Between Nutritional Indicators and Dietary Diversity

This section summarizes multivariable regression findings on associations between nutritional outcomes and key predictors among children in ELCs. Variables included in the model (sex, age, and DDS) were selected based on univariate p-values < 0.25. Girls, younger age, and high DDS are reference categories. While individual-level factors were adjusted for, site-level influences such as ELC infrastructure and feeding systems were not assessed and warrant future investigation (

Table 8. Association on nutritional indicators with age, sex, and DDS.

Outcome	Predictor	PR (95% CI)	p-Value	aPR (95% CI)	p-Value
Stunting
	Sex (boys vs. girls) Age (older vs. younger) DDS (low vs. high)	0.42 (0.03–0.82) 0.64 (0.40–0.88) −0.21 (−0.37–−0.05)	0.036 * ≤0.0001 * 0.010 *	0.38 (0.01–0.74) 0.61 (0.37–0.85) −0.15 (−0.29–−0.06)	0.042 * ≤0.0001 * 0.041 *
Underweight
	Sex (boys vs. girls) Age (older vs. younger)	0.78 (−0.37–1.93) 1.30 (0.60–2.00)	0.182 ≤0.0001 *	0.72 (−0.40–1.85) 1.27 (0.58–1.96)	0.207 ≤0.0001 *
Thinness
	Sex (boys vs. girls) Age (older vs. younger)	16.81 (15. 46–18.16) 1.27 (−0.16–2.69)	≤0.0001 * 0.081	17.00 (15.26–18.74) 1.12 (−0.03–2.27)	≤0.0001 * 0.056
Overweight/Obesity
	Sex (boys vs. girls) DDS (low vs. high)	0.25 (−0.11–0.59) −0.07 (−0.23–0.90)	0.166 0.394	0.22 (−0.12–0.57) −0.06 (−0.23–0.10)	0.204 0.443

* Indicates a statistically significant association (p < 0.05).

).

Boys had a significantly lower risk of stunting compared to girls (aPR = 0.38; 95%CI: 0.01–0.74). Younger age (aPR = 0.61; 95%CI: 0.37–0.85) and low DDS (aPR = −0.15; 95%CI: −0.29–−0.06) were associated with higher stunting risk. Age was positively associated with underweight (aPR = 1.27; 95%CI: 0.58–1.96), while thinness was strongly linked to boys (aPR = 17.00; 95%CI: 15.12–18.74). No significant associations were observed for overweight/obesity.

4. Discussion

This study aimed to determine the relationship between dietary diversity and nutritional status among preschool-aged children in Johannesburg’s Early Learning Centres (ELCs).

To begin with ELCs, the findings showed that all ELCs reported funding constraints, despite structured feeding practices. Financial constraints might lead to ELCs using fixed meal plans that overlook age- and sex-specific nutritional needs, particularly the higher energy and nutrient requirements of older children and boys [9,44]. Such non-individualized feeding approaches can exacerbate nutritional vulnerabilities in low-resource settings [19]. Urban ELCs face additional challenges, including limited space for food gardens, reliance on processed foods, and inconsistent caregiver training [20]. These are compounded by weak procurement systems, inadequate infrastructure, and limited access to nutrition professionals, all of which hinder the provision of nutrient-dense, age-appropriate meals [24,52]. Township-based ELCs also struggle with transport limitations, lack of bulk storage, and proximity issues, making local supermarkets and spaza shops more accessible. Similar procurement constraints have been observed in small-scale childcare centres, which often rely on nearby outlets due to affordability, infrastructure gaps, and regulatory exclusion [53,54]. Furthermore, the low proportion of ELCs sourcing food from wholesalers may reflect logistical and infrastructural constraints such as lack of bulk storage, transport limitations, and distance from wholesale outlets. Most childcare services have been reported to rely on supermarkets due to affordability and accessibility, with limited use of farmers’ markets or wholesalers [53].

Additionally, in these ELCs, older children have increased energy and nutrient requirements due to accelerated growth and development, and boys typically require more nutrients than girls, as reflected in sex-specific growth curves [44]. If ELCs provide fixed meal portions without adjusting for age or sex, children with higher needs may be at risk of undernutrition. For instance, as discussed below, this study found that underweight was more prevalent among older children, suggesting that current feeding practices may not be adequately tailored. While all ELCs expressed the need for dietitian support, it remains unclear whether they currently have the capacity to adjust meal plans based on individual nutritional needs. Integrating trained nutrition professionals into ELCs could facilitate such adjustments.

In the context of these findings, DDS remains a widely recognized proxy for micronutrient adequacy [3,4]. Its simplicity and predictive value make it suitable for institutional settings such as ELCs, despite limitations in accounting for portion size, nutrient density, and food processing [2,5]. Importantly, dietary diversity does not always equate to diet quality; a high DDS may include ultra-processed, energy-dense foods that are nutrient-poor. Therefore, DDS should be interpreted alongside indicators of nutrient density and food processing to avoid overestimating diet quality [34]. These considerations highlight the need for complementary strategies to improve the intake of protein-rich and micronutrient-dense foods within structured feeding programmes.

Furthermore, these findings on consumption of food groups by children in ELCs indicate the importance of site-level factors, such as infrastructure, procurement systems, and access to nutrition professionals in shaping dietary outcomes. Evidence from South Africa shows that ELCs with structured nutrition support and dietitian involvement are better positioned to deliver age-appropriate, nutrient-dense meals [55]. Moreover, integrating food gardens into ELCs has been shown to improve dietary diversity and micronutrient intake among preschool children [56,57]. Community-based nutrition education for caregivers also plays a critical role in reinforcing healthy feeding practices at home, complementing institutional efforts [57]. These multi-level interventions are essential for addressing the dual burden of malnutrition in early childhood settings.

Turning to nutritional outcomes, the coexistence of undernutrition and overnutrition highlights a dual burden of malnutrition in this population. Stunting remains the most persistent form of undernutrition, aligning with national trends of slow progress despite interventions [29]. Its association with younger age, male sex, and low DDS indicate the critical role of diet quality in supporting linear growth [2,3]. These patterns call for integrated strategies that address both inadequate and excessive nutrient intake within institutional feeding programmes.

Building on this, underweight was observed primarily among older age. These results reflect cumulative nutritional deficits and sex-based disparities in vulnerability [13,38]. These findings align with evidence from informal settlements and rural South Africa, where underweight remains one of the concerns despite national nutrition programmes [26,30]. In SSA, underweight is often linked to food insecurity, poor maternal nutrition, and limited access to health services [13,38]. Comparable trends have been reported in Kenya [52], Nigeria [58], and Ethiopia [59], where urban poor children face similar nutritional challenges. These patterns call for integrated strategies that address both inadequate and excessive nutrient intake within institutional feeding programmes.

In contrast, overnutrition affected nearly one-third of children. However, there was no significant association between DDS and overweight/obesity suggesting that dietary quantity and energy density, rather than diversity, may be driving weight gain [22]. These rates mirror findings from Vhembe District and Mpumalanga, where early childhood obesity is rising [24,28]. This reflects the nutrition transition occurring in South Africa [36], Southern Africa [35] and in LMICs at large [34], characterized by the rise of ultra-processed food consumption and a gradual decline in traditional, nutrient-rich dietary patterns [34].

Taken together, these findings reflect the paradoxical coexistence of stunting and overweight/obesity (CSO), a pattern consistent with evidence from other South African provinces where CSO prevalence is alarmingly high [28,31]. Structural inequities, limited caregiver nutrition knowledge, and gaps in institutional feeding practices have been identified as key drivers of this phenomenon [26,27,33]. Addressing CSO requires integrated interventions that simultaneously tackle undernutrition and overnutrition [28,60,61]. These priorities align with South Africa’s National Food and Nutrition Security Plan (2022–2027), which emphasizes early childhood nutrition and calls for multisectoral collaboration to strengthen outcomes in foundational learning environment.

Although the study had some limitations, it provides valuable insights into the nutritional realities of children in ELCs, using validated tools aligned with WHO and STROBE standards. Further strength lies in its focus on institutional feeding environments, a relatively underexplored domain in urban South Africa. However, the relatively small sample size may limit generalizability and reduce statistical power for detecting subtle associations. The cross-sectional design also limits causal inference. While full participation was achieved among eligible children, reliance on caregiver consent, particularly the absence of mothers due to work obligations, may have introduced selection bias and constrained household-level data collection. The absence of household data restricts understanding of broader nutritional determinants, although the study’s primary focus was on ELCs. Systematic random sampling was applied within each centre; however, maintaining a fixed proportion of participants per ELC was not possible due to variability in enrolment size, parental consent, and attendance. This may have introduced sampling imbalance and affected representativeness at the centre level, even though the overall sample achieved a 90% response rate. DDS, while validated, does not fully capture nutrient density, portion size, or food processing, and may overestimate diet quality when dietary variety includes energy-dense, nutrient-poor foods. Future studies should combine DDS with nutrient profiling or biochemical assessments. The exclusion of biologically implausible z-scores (>±5 SD) enhanced data validity but may have underestimated extreme malnutrition prevalence. Importantly, structural determinants at the site level, such as ELC infrastructure, food procurement systems, and feeding practices were not assessed. These factors can significantly influence meal quality, food availability, and cost-efficiency, and their omission limits the ability to fully capture contextual drivers of dietary diversity and malnutrition. Future research should integrate these variables to inform comprehensive, system-level interventions. The infrastructure and administrative questionnaire, while developed using literature, yielded limited contextual depth; future studies should refine this tool to capture more detailed indicators of food systems, staffing, and feeding practices.

5. Conclusions

This study highlights the coexistence of stunting and overweight/obesity among preschool-aged children in Johannesburg ELCs (Region C), reflecting a persistent dual burden of malnutrition shaped by inadequate dietary diversity, institutional feeding constraints, and limited caregiver engagement. Stunting was significantly associated with younger age, male sex, and low DDS, emphasizing the importance of diet quality for linear growth, whereas overweight showed no link to DDS, suggesting that energy-dense, ultra-processed foods may drive excess weight gain. These patterns point to structural inequities within early learning environments and call for integrated, context-sensitive interventions. Key priorities include strengthened nutrition governance and funding, incorporation of dietitians for age-appropriate meal planning, promotion of food gardens to improve micronutrient intake, and caregiver-focused nutrition education. Institutionalizing routine growth monitoring, dietary audits, and procurement training, alongside infrastructure investment, can optimize food sourcing and cost-efficiency. Future interventions should also consider site-level determinants within ELCs, such as infrastructure and feeding systems to ensure sustainable improvements. Collectively, these strategies align with South Africa’s National Food and Nutrition Security Plan and global nutrition targets, offering a pathway to reduce malnutrition and advance child nutrition equity.