Determinants of Growth and Development in Stunted Toddlers: A Systematic Review

¹Department of Nutrition Science, Faculty of Public Health, Hasanuddin University, Makassar, Indonesia, ²Department of Epidemiology, Faculty of Public Health, Hasanuddin University, Makassar, Indonesia

^&Corresponding author
Ruwiah Ruwiah, Department of Nutrition Science, Faculty of Public Health, Hasanuddin University, Makassar, Indonesia

Introduction    

Stunting, defined by the World Health Organization (WHO) as height-for-age more than two standard deviations below the median of the WHO Child Growth Standards, remains one of the most pressing indicators of chronic undernutrition and long-term adversity in early childhood [1-3]. Globally, it is estimated that 148.1 million children under five years of age were stunted in 2022, with the highest burden observed in South Asia and sub-Saharan Africa [4]. Stunting not only reflects poor nutritional status but also serves as a marker of cumulative exposure to adverse socioeconomic and environmental conditions during critical periods of growth and development [5,6].

The consequences of stunting extend far beyond impaired linear growth [7,8]. Evidence suggests that stunting during the first 1,000 days of life, the period from conception to a child's second birthday, can result in lasting deficits in cognitive, motor, and socio-emotional development [9]. These developmental impairments are associated with poorer school achievement, reduced economic productivity in adulthood, and an increased risk of chronic diseases such as diabetes and cardiovascular conditions later in life [10]. As such, the early identification of risk factors for stunting and its developmental consequences is a public health imperative.

Numerous determinants have been linked to stunting, including poor maternal health and nutrition [11], low birth weight [12], inadequate infant feeding practices [13], exposure to infectious diseases [14], and limited access to clean water and sanitation [15]. Socioeconomic status, parental education, and household food insecurity are also recognized as major contributors. While these determinants are well-documented, the pathways through which they influence both physical growth and developmental outcomes remain complex and context-specific [16,17]. Moreover, the interplay of these risk factors across critical periods of child development requires further investigation, particularly in resource-constrained settings.

Retrospective observational studies, including retrospective cohort, case-control, record-based analyses, and secondary analyses of existing datasets, provide a distinct perspective on determinants of stunting and developmental outcomes because they allow historical exposures and previously recorded outcomes to be examined using existing data sources [18,19]. Unlike purely cross-sectional analyses, retrospective studies based on longitudinal records may allow earlier exposures to be related to outcomes measured at later time points [7,19]. Although prospective cohort studies generally provide clearer temporal ordering, they require longitudinal follow-up and substantial methodological planning [20], while intervention studies primarily address the effects of assigned interventions rather than naturally occurring determinants. Retrospective studies can therefore provide an efficient means of examining early-life exposures and longer-term outcomes when suitable historical or routinely collected data are available [7,18,19].

While many studies have focused on the prevalence of stunting and its immediate determinants [21,22], fewer have examined determinants of both growth and developmental outcomes, and even fewer systematic reviews have specifically synthesized retrospective evidence across these domains. This focused synthesis may therefore help identify patterns of early-life exposures that are not readily captured within intervention-oriented or purely cross-sectional evidence. This systematic review aims to fill this gap by synthesizing the existing retrospective evidence on the determinants of growth and development in stunted toddlers. Specifically, it seeks to identify the maternal, environmental, biological, nutritional, and socioeconomic factors that are associated with stunting and developmental delays in children under five years of age. By focusing on retrospective observational studies, this review provides a comprehensive perspective on early-life exposures and their influence on childhood growth and developmental trajectories in vulnerable populations.

Methods     

Study design: this systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) guidelines [23]. The review focused on retrospective observational evidence examining determinants associated with growth and developmental outcomes in children with stunting. Eligible retrospective designs included retrospective cohort or longitudinal studies, retrospective case-control studies, retrospective analyses of medical records or surveillance data, and secondary analyses of previously collected observational or survey datasets. These designs were included because the objective of the review was to synthesize evidence derived from historical or pre-existing data rather than to restrict the review to a single epidemiological design. The review protocol was not prospectively registered in PROSPERO or another systematic review registry.

Search strategy: a comprehensive literature search was conducted in five electronic databases: PubMed, Scopus, Web of Science, CINAHL via EBSCO, and Embase. Searches were conducted in January 2025, with the final search performed on August 2025. The search strategy combined terms related to the population (e.g., "stunted child," "stunted children," "toddler," "under-five," and "preschool child"), outcomes (e.g., "growth," "development," "child development," and "cognitive development"), determinants or exposures (e.g., "determinant," "risk factor," "predictor," and "associated factor"), and study design (e.g., "retrospective study," "chart review," "medical record," and "secondary data"). Database-specific syntax, indexing terms, and field restrictions were applied as appropriate. No publication-date restriction was applied, and studies published in English or with an English-language version available were considered. The complete Boolean search strategies for all five databases are presented in Table 1.

Inclusion and exclusion criteria: studies were included if they examined children who experienced stunting during the first five years of life, assessed at least one determinant or risk factor related to growth or developmental outcomes, and used a retrospective observational approach. Eligible designs included retrospective cohort or longitudinal studies, retrospective case-control studies, retrospective record-based cross-sectional analyses, and secondary analyses of existing observational datasets. Studies were excluded if they focused exclusively on non-stunted populations or children outside the predefined age criteria, did not assess determinants or risk factors, did not report relevant growth or developmental outcomes, used prospective or interventional designs, or were purely cross-sectional studies without retrospective ascertainment or analysis of previously collected data. Reviews, editorials, case reports, and non-original research were also excluded.

Study selection and data extraction: data extraction was performed by two authors using a structured template developed in Microsoft Word. The extraction form was not formally piloted before use. Extracted variables included study characteristics (author, year, country, study design, sample size, and age range), determinant categories, exposures, outcomes assessed, and key findings, including reported effect estimates where available. The extracted data were subsequently reviewed and discussed by the full author team, and the final dataset was approved by all authors through consensus. No external reviewers were involved in the data extraction process. A formal inter-reviewer agreement statistic was not calculated.

Methodological quality assessment: methodological quality was assessed using the Newcastle-Ottawa Scale (NOS) for observational studies. The NOS evaluates three domains: selection (maximum 4 points), comparability (maximum 2 points), and outcome/exposure assessment (maximum 3 points), with a maximum total score of 9 points. Studies scoring 7-9 were classified as good quality, those scoring 5-6 as moderate quality, and those scoring below 5 as low quality. The NOS was retained as the sole methodological quality assessment tool because the included studies were observational studies evaluating determinants and exposures rather than interventions.

Assessment of heterogeneity and data synthesis: heterogeneity across the included studies was assessed qualitatively in terms of clinical, methodological, and statistical characteristics. Clinical heterogeneity was examined based on differences in study populations, participant age, geographical setting, exposure type, timing of exposure, and growth or developmental outcomes. Methodological heterogeneity was assessed according to differences in retrospective study design, sampling approach, data source, exposure and outcome ascertainment, and adjustment for potential confounders. Statistical heterogeneity was considered by examining the variation in effect measures, analytical approaches, covariate adjustment, and reporting formats across studies. Because the studies differed substantially in exposures, outcome definitions and time points, study designs, and reported effect measures, a statistically meaningful pooled effect estimate could not be derived. Therefore, formal statistical heterogeneity measures such as Cochran's Q and I² were not calculated, and a narrative synthesis was undertaken. Findings were grouped thematically into maternal, environmental, nutritional, biological, and socioeconomic determinants, while reported effect estimates such as odds ratios, risk ratios, regression coefficients, confidence intervals, and p-values were presented where available.

Results     

The database searches identified 2,804 records. After the removal of 2,277 records before screening, 527 records remained for title and abstract screening. Of these, 498 records were excluded because they did not meet the review criteria, leaving 29 reports for full-text eligibility assessment. Following full-text evaluation, 16 reports were excluded because they did not meet one or more predefined eligibility criteria, including an ineligible study design, an ineligible population or age range, absence of relevant growth or developmental outcomes. Consequently, 13 studies met all eligibility criteria and were included in the final qualitative synthesis and data extraction. The study-selection process is presented in the PRISMA 2020 flow diagram (Figure 1).

Study characteristics: thirteen retrospective studies were included in this review, conducted between 2017 and 2025. Most were based in Indonesia, with others from Rwanda, Iran, South Africa, and Japan. Study designs included retrospective cohort, retrospective case-control, retrospective cross-sectional, and secondary data analyses of national surveys. Sample sizes varied considerably, ranging from 51 to 33,599 participants, with age ranges from birth to five years. Data sources included maternal and child health (MCH) records, household surveys, and national-level health databases such as the Demographic and Health Survey in Rwanda. The detailed characteristics of the included studies, including study design, sample size, determinants, and outcomes, are presented in Table 2 and Table 2.1.

Summary of studies quality assessment: methodological quality was assessed using the Newcastle-Ottawa Scale (NOS). Of the 13 included studies, seven were classified as good quality and six as moderate quality; no study was classified as low quality. Studies rated as good quality generally demonstrated clearer participant selection, more reliable exposure and outcome assessment, and better consideration of potential confounding factors. Moderate-quality studies commonly showed limitations related to sample representativeness, comparability between groups, or incomplete adjustment for relevant confounders. Overall, the methodological quality of the included evidence was considered acceptable, although limitations in confounding control and participant selection were evident in several studies. Detailed NOS assessments are presented in Table 3.

Determinants of growth (stunting): the most frequently reported outcome was stunting, typically defined using WHO height-for-age z-scores below minus two standard deviations.

Maternal and prenatal factors: several studies identified maternal characteristics as key determinants. Short maternal height, low education level, and early age at first pregnancy were consistently associated with increased stunting risk. For example, Panatariono et al. [23] reported that maternal height, education, and early pregnancy were significantly linked to child stunting. Rahmawati et al. [24] and Niragire et al. [25] also found strong associations with maternal stature and antenatal care visits.

Environmental and household conditions: poor water, sanitation, and hygiene (WASH) conditions were associated with stunting. Heriana et al. [26] found that the use of unsafe water and poor toilet facilities substantially increased the risk of stunting. Oginawati et al. [27] reported that environmental exposure to arsenic and cadmium was significantly associated with stunting, alongside low parental education and income.

Nutrition and feeding practices: nutritional conditions during pregnancy and early childhood were significant predictors. Karbin et al. [28] found that prenatal food insecurity was linked to stunting at birth and in early infancy. Mbabazi et al. [29] observed that early introduction of cow's milk and low dietary diversity were associated with malnutrition indicators such as anemia. Modjadji et al. [30] reported that maternal alcohol and tobacco use increased the risk of underweight and thinness in infants.

Biological and child-related factors: perinatal characteristics such as low birth weight and preterm birth were also important. Utami et al. identified both factors as significant predictors of stunting. Niragire et al. [25] found that birth weight and sex of the child were consistently associated with stunting across national datasets.

Determinants of developmental outcomes: only two studies explicitly evaluated developmental outcomes. Utami et al. [31] reported that stunted children during their first two years of life showed lower cognitive scores at ages four to six years, although the association was not statistically significant. Psychosocial stimulation and parental education were also identified as important predictors of child development. Rahmawati et al. [24] found that maternal education was positively associated with developmental scores, while stunting had a negative impact, using the KPSP assessment tool.

Exposure timing and critical windows: most studies emphasized the first 1,000 days of life as a critical window for growth and development. Stunting was commonly assessed at birth, six months, or between 12 and 24 months of age. Developmental outcomes were typically measured between three and six years, reflecting the lag between early-life exposures and later developmental assessment.

Measurement tools: growth outcomes were assessed using WHO child growth standards, specifically height-for-age z-scores. Developmental outcomes were measured using tools such as the Kuesioner Pra Skrining Perkembangan (KPSP), a validated Indonesian screening tool adapted from the Denver Developmental Screening Test. Environmental exposures, such as heavy metals, were measured via biological samples in some studies.

Effect estimates and statistical associations: quantitative effect estimates varied considerably across the included studies. Heriana et al. reported that unhealthy household toilet facilities were associated with a higher risk of stunting (RR=5.41, 95% CI 2.78-10.50; p=0.001), as was water quality that did not meet recommended standards (RR=4.69, 95% CI 2.51-8.75; p=0.001). Karbin et al. found that household food insecurity during pregnancy was associated with stunting at birth (RR=2.01, 95% CI 1.17-3.46), four months (RR=3.03, 95% CI 1.21-7.61), and six months of age (RR=3.83, 95% CI 1.37-10.68). In Sari et al. low birth weight (OR=3.08, 95% CI 1.07-8.89; p=0.032), preterm birth (OR=3.71, 95% CI 1.41-9.77; p=0.006), and short birth length (OR=5.55, 95% CI 2.00-15.46; p=0.001) were associated with stunting. Shinsugi et al. similarly reported higher adjusted odds of stunting among children born small for gestational age, both in boys (aOR=5.78, 95% CI 3.48-9.60; p<0.001) and girls (aOR=4.04, 95% CI 2.43-6.73; p<0.001).

In Rwanda, Niragire et al. [25] demonstrated temporal variation in adjusted associations across three national survey rounds. Male sex remained associated with higher odds of stunting in 2010, 2015, and 2020 (POR=1.47, 1.61, and 1.53, respectively), whereas birth weight >2.5 kg was consistently associated with lower odds (POR=0.53, 0.50, and 0.41, respectively). Maternal height ≥145 cm was also consistently associated with lower odds of stunting (POR=0.24, 0.27, and 0.31, respectively). However, associations for maternal education, water source, and toilet facilities varied over time. Lolan et al. reported that lack of husband support was independently associated with stunting in the final multivariable model (OR=1.93, 95% CI 1.33-6.14; p=0.007) [32]. Oginawati et al. [27] found differences in drinking-water arsenic (p=0.008) and cadmium concentrations (p=0.024) between stunted and non-stunted groups, whereas lead concentrations did not differ significantly (p=0.207).

For developmental outcomes, Rahmawati et al. [24] reported that stunting was negatively associated with child developmental scores (b=-0.78, 95% CI -1.46 to -0.10; p=0.025), whereas higher maternal education was positively associated with development (b=1.08, 95% CI 0.41-1.75; p=0.001). In the same path model, greater birth length (b=-0.90, 95% CI -1.60 to -0.21; p=0.011) and maternal height ≥150 cm (b=-0.92, 95% CI -1.69 to -0.16; p=0.018) were associated with lower stunting scores. Utami et al. found that previous early-life stunting was associated with higher adjusted odds of low developmental outcomes, although the estimate was imprecise and not statistically significant (aOR=1.59, 95% CI 0.65-3.92; p=0.307). Current stunting likewise showed an elevated but imprecise estimate (aOR=3.86, 95% CI 0.85-17.45; p=0.079).

Not all included studies reported comparable inferential effect estimates. Ratnawati et al. provided descriptive prevalence data only, identifying stunting in 13 of 84 newborns (15.4%) [33]. Mbabazi et al. likewise primarily reported descriptive findings on anemia and socioeconomic characteristics among malnourished children and did not provide ORs, RRs, or confidence intervals for stunting determinants. Panatariono et al. reported regression coefficients for maternal height (B=0.178; p=0.005) and maternal age at first pregnancy (B=0.118; p=0.015); however, the reported p-value for maternal education was internally inconsistent between the regression table (p=0.11) and accompanying text (p=0.011), and this finding was therefore interpreted cautiously. Modjadji et al. did not demonstrate adjusted associations between current maternal tobacco use and stunting (aOR=0.67, 95% CI 0.23-1.93; p=0.455) or current alcohol use and stunting (aOR=0.72, 95% CI 0.35-1.46; p=0.362).

Discussion     

This systematic review identified a broad range of determinants associated with impaired growth and development among stunted toddlers. The findings reaffirm that stunting is a multifactorial condition influenced by maternal, nutritional, environmental, and socioeconomic variables, often interacting within critical developmental windows. The evidence consistently demonstrated that adverse conditions in early life, particularly during the prenatal period and first two years, were consistently associated with subsequent growth and developmental outcomes.

Maternal and prenatal factors such as short maternal height, low education, and early maternal age at pregnancy were repeatedly associated with a higher likelihood or risk of stunting. These findings are consistent with a recent systematic review and meta-analysis by Rezaeizadeh et al. which identified maternal education as an important factor associated with child growth and nutritional status during the first two years of life [34]. In the present review, Panatariono et al. [23] reported associations between stunting and maternal height, education level, and young maternal age at first pregnancy, while Rahmawati et al. [24] and Niragire et al. [25] similarly identified maternal stature and antenatal-related factors as relevant correlates of stunting. Taken together, these findings suggest that maternal education may operate alongside maternal biological characteristics, socioeconomic conditions, and access to antenatal care rather than as an isolated determinant of child growth.

Environmental exposures were also consistently associated with stunting. Heriana et al. reported higher stunting risk among children exposed to inadequate toilet facilities and unsafe water [26], while Oginawati et al. identified associations with arsenic and cadmium exposure [27]. These findings are consistent with the 2025 systematic review and meta-analysis by Hasyim et al. which identified unimproved sanitation, poor handwashing practices, unsafe or unimproved water sources, and household smoke exposure as recurrent environmental factors associated with childhood stunting in Indonesia [35]. The convergence of findings across the present retrospective synthesis and recent pooled evidence reinforces the importance of considering environmental conditions alongside nutritional and socioeconomic determinants [36,37]. However, substantial heterogeneity across environmental studies suggests that the magnitude of these associations may vary according to local infrastructure, exposure measurement, and socioeconomic context.

Nutritional deprivation, both prenatal and postnatal, was a consistent theme. Karbin et al. [28] showed that food insecurity during pregnancy substantially increased the risk of stunting in early infancy. Mbabazi et al. [29] and Modjadji et al. [30] emphasized poor feeding practices, early introduction of cow's milk, and maternal alcohol or tobacco use as contributing factors. These highlight the compounded risks stemming from inadequate maternal nutrition and harmful behaviors, which may be deeply rooted in poverty and limited access to health education.

Only two of the included studies directly evaluated developmental outcomes, indicating that developmental consequences remain substantially less studied than anthropometric outcomes in the retrospective literature. This limited evidence base may partly reflect the characteristics of retrospective data sources, which more commonly contain routinely recorded anthropometric indicators than standardized developmental assessments. Developmental outcomes also require age-appropriate instruments, trained assessment, and adequate temporal separation between early-life exposures and later developmental evaluation, making them less readily available in existing records or secondary datasets. This interpretation is consistent with Utami et al. who noted the limited availability of longitudinal evidence on stunting and child development, particularly in developing-country settings [31]. In the available studies, maternal education [24] and early-life stunting were associated with developmental outcomes [38,39], although the small number of studies and differences in assessment approaches limit the strength and generalizability of these findings. Further prospective longitudinal studies using standardized developmental instruments are therefore needed to clarify the relationship between early growth faltering and subsequent cognitive, motor, language, and socio-emotional development.

Potential pathways linking early growth faltering with developmental outcomes include inadequate nutrient availability during periods of rapid brain development, reduced opportunities for active exploration and learning, recurrent illness, and lower psychosocial stimulation. These mechanisms may interact with socioeconomic disadvantage and parental education, suggesting that developmental outcomes cannot be attributed to nutritional status alone.

The limited developmental evidence identified in the present review should also be interpreted in light of recent broader evidence. A 2025 meta-analysis by Sideropoulos et al. found no clear association between childhood stunting and overall cognitive development in the primary pooled model, although domain-specific analyses identified poorer outcomes in intelligence, executive function, visuospatial cognition, and selected socio-emotional domains. The authors further highlighted the importance of demographic, socioeconomic, parental, and health-related contextual factors in shaping developmental outcomes. These findings support the cautious interpretation of the present review, suggesting that the relationship between early growth faltering and later development is heterogeneous and cannot be attributed to nutritional status alone [40]. Similarly, Upadhyay et al. reported in a 2024 systematic review and meta-analysis that improvements in linear growth beyond two years of age were associated with only small improvements in cognitive and language outcomes, with no clear association with motor or socio-emotional development. This further suggests that growth recovery and developmental recovery may not occur in parallel and that developmental outcomes are likely shaped by multiple contextual and caregiving factors in addition to physical growth [41].

The reviewed studies also highlighted the importance of timing. Several authors referred to the "first 1,000 days" from conception to age two as the most sensitive period for intervention. Stunting onset during this phase was found to be particularly predictive of poor developmental outcomes in later childhood. Moreover, child-level variables such as low birth weight, preterm birth, and high birth order were also identified as consistent predictors of stunting, as evidenced in the work of Sari et al. [42] and Shinsugi et al. [43], reinforcing that biological vulnerability further compounds risk in socially disadvantaged environments. The importance of early-life timing is also supported by recent intervention-focused evidence. Azwar et al. in a systematic review of psychosocial stimulation interventions for stunted toddlers during the first 1,000 days, highlighted the importance of integrating psychosocial stimulation with family empowerment, nutrition-related support, hygiene, and health education [44]. This evidence suggests that interventions addressing growth alone may be insufficient to optimize developmental outcomes and supports a multidimensional approach during early childhood.

Despite the consistent identification of several risk factors, variability in findings was observed across regions and study populations. For example, while maternal education was almost universally associated with improved outcomes, its strength and significance varied. Similarly, socioeconomic indicators such as household income or parental employment had differing impacts depending on local context, healthcare access, and underlying infrastructure.

Limitation: this review has several limitations. First, the retrospective observational nature of the included studies limits causal inference because residual and unmeasured confounding cannot be excluded. Reliance on previously collected records, routine datasets, or participant-reported information may also introduce incomplete exposure ascertainment and recall bias, depending on the original data source. Selection bias may have occurred in studies using health-facility records, local databases, or non-random sampling, potentially limiting the representativeness of study populations. In addition, publication bias cannot be excluded because the review focused on published peer-reviewed studies, and studies reporting null or unfavorable findings may have been less likely to be published or identified. Publication bias was not formally assessed because the substantial clinical, methodological, and statistical heterogeneity precluded meta-analysis and did not provide a sufficiently comparable set of effect estimates for funnel-plot or related assessments. Most included studies were conducted in Indonesia, which may further limit generalizability to other settings. Developmental outcomes were evaluated in only a small number of studies and with different instruments and assessment time points, limiting direct comparison. Finally, substantial heterogeneity in study designs, exposures, outcomes, and statistical reporting prevented meaningful quantitative pooling.

Conclusion     

This systematic review identified multiple maternal, nutritional, environmental, biological, and socioeconomic factors that were associated with growth outcomes among children under five, particularly during pregnancy and the first two years of life. Developmental outcomes were examined in only a small number of studies, and the available evidence suggests an association between early-life stunting and poorer developmental outcomes, including cognitive development. Given the retrospective observational nature of the included evidence, these associations should not be interpreted as causal, particularly because residual confounding and differences in study design and measurement cannot be excluded. The findings support integrated approaches addressing maternal and child health, nutrition, sanitation, environmental conditions, and psychosocial stimulation. Future research should prioritize prospective designs and standardized developmental assessments to clarify temporal relationships and strengthen causal inference.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

All the authors read and approved the final version of this manuscript.

Tables and figure     

Table 1: search strings in databases

Table 2: characteristics of included retrospective studies examining determinants of growth and developmental outcomes among stunted children under five years across multiple countries (n=5 studies)

Table 2.1: characteristics of included retrospective studies examining determinants of growth and developmental outcomes among stunted children under five years across multiple countries (n=3 studies)

Table 3: quality assessment of included studies using the Newcastle-Ottawa Scale (NOS), evaluating selection, comparability, and outcome/exposure domains with total scores ranging from 0 to 9

Figure 1: PRISMA flow diagram of the study selection process for the systematic review of determinants of growth and developmental outcomes in stunted toddlers, including database identification, screening, eligibility assessment, and final inclusion of 13 retrospective studies

References