Introduction
Low length-for-age, or stunting, is an undernutrition condition marked by linear growth faltering due to inadequate nutrition in utero and early childhood, affecting more than 148 million children under 5 worldwide(1). It is associated with a greater risk of weight excess and non-communicable diseases later in life, as well as irreversible physical and brain damage with long-term impacts(Reference Martins, Toledo Florêncio and Grillo2). Stunted children may reach adulthood with short stature and may never achieve their full cognitive potential, resulting in learning difficulties in school and reduced lifetime income, a drawback that can even propagate to the next generation(Reference Victora, Adair and Fall3,Reference Wells, Sawaya and Wibaek4) .
This health condition can be seen as a synonym for social disadvantage, where children facing social, economic, political and emotional challenges are shorter than those in more favourable environments(Reference Scheffler, Hermanussen and Soegianto5). The Brazilian Conditional Cash Transfer Program, Bolsa Família, represents an important strategy to alleviate socio-economic disparities and improve the living conditions of families in situations of poverty and extreme poverty(6). In the State of Alagoas, the Brazilian state with the lowest Human Development Index(7) and where the programme plays a crucial role in supporting households, it is essential to explore the conditions influencing child health in this specific context.
The WHO framework on the Context, Causes and Consequences of Childhood Stunting, published in 2013(Reference Stewart, Iannotti and Dewey8), outlines a set of factors associated with stunting, examined at a global level. However, it is extremely useful to investigate how these factors manifest among young children living in poverty in one of the poorest Brazilian states. This investigation can provide crucial insights to support policymakers in decision-making within the realm of public policies for health and nutrition in the country.
Structural equation modelling (sEM) is a multivariate technique used to investigate the complex relationships among variables through the combination of elements of factor analysis and regression analysis, testing the direct, indirect and total effects on pre-assumed relationships(Reference Kang and Ahn9,Reference Schreiber, Nora and Stage10) .
Thus, this study was conducted to utilise sEM in exploring the direct and indirect effects of parental level of education and child individual factors on the length-for-age outcomes among children aged 6–24 months assisted by the Bolsa Família Program (BFP) in the State of Alagoas.
Methods
Design
This cross-sectional study is part of a project entitled ‘Evaluation of the Management and Operation of the National Program for Iron and Vitamin A Supplementation and their Relationship with the Nutritional Status of Children Aged 6–24 Months in Municipalities of Alagoas State’. This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects were approved by the Ethics Committee in Research of Alagoas Federal University (process no. 80416617.0.0000.5013). Written informed consent was obtained from the legal guardians of all subjects. The STrengthening the Reporting of OBservational studies in Epidemiology – Nutritional Epidemiology (STROBE-nut) checklist for cross-sectional studies was adopted for structuring the study(Reference von Elm, Altman and Egger11).
Setting and sample selection
Information on the study region and participant recruitment are described in detail elsewhere(Reference Mendes, Marçal and Fragoso12–Reference Normande, Marçal and Rinaldi14). Briefly, children aged 6–24 months, assisted by the BFP, and residents of six municipalities in the State of Alagoas, located in the northeast region of Brazil, were eligible for enrolment. The flow diagram for the selection of study participants is reported in Fig. 1.
Flow diagram for the selection of study participants. *As registered in the Consultation, Selection and Extraction of Information from the Unified Registry (CECAD). BFP, Bolsa Família Program.
Data collection
Data were collected between May and December 2018 by trained staff comprising four nutritionists and ten technical assistants. The children’s parents were interviewed using a structured questionnaire about their sociodemographic issues and the children’s nutritional status.
Outcome variable
The outcome variable was the length-for-age indicator (continuous). Subsequently, this indicator was classified as stunting when length-for-age < −2 Z-score(15), analysed using the WHO’s Anthro Survey Analyzer software, version 3.2.2(16). Length was measured in a recumbent position using the Alturexata® portable infantometer (Alturexata Ltd) placed on a flat and stable surface. The children were measured with no clothes, shoes, socks or hair ornaments, and their parents were requested to hold the children’s head while the researcher held the knees and took the measurement(17).
Exogenous and endogenous variables
The selection of parental and infant variables for the sEM was based on previous publications on early-life determinants of childhood stunting(Reference Islam, Zafar Ullah and Mainali18–Reference Mulyaningsih, Mohanty and Widyaningsih21). The following variables were investigated: the parental level of education, children’s characteristics (including age, sex and birth weight) and complementary feeding indicators (including ever breastfed, bottle feeding 0–23 months, sweet beverage consumption 6–23 months and minimum acceptable diet 6–23 months).
Children’s birth weight was collected from the Child Health Handbook and later classified as low weight (<2500 g), normal (≥2500 g and <4000 g) or high (≥4000 g)(Reference Puffer and Serrano22).
Children’s complementary feeding was assessed using the questionnaire proposed by Oliveira et al. (Reference Oliveira, Castro and Silva23), adapted for regional foods and based on the WHO indicators for assessing infant and young child feeding practices(24). The questionnaire was composed of pre-established questions about the child’s food consumption in the last 24 h before the interview. Noteworthy, posteriorly, the complementary feeding data were updated to the new revised young child feeding practice indicators published in 2021 by the WHO and UNICEF(25). The indicator of ever breastfed was calculated using the proportion of children born in the last 24 months who were ever breastfed. The bottle feeding indicator was obtained by the percentage of children 6–23 months of age who were fed from a bottle with a nipple during the previous day. Sweet beverage consumption was calculated through the proportion of children 6–23 months of age who consumed a sweet beverage during the previous day. Finally, the indicator of minimum acceptable diet was constructed based on specific criteria for breastfed and non-breastfed children. For breastfed children, a minimum acceptable diet was defined as meeting both minimum dietary diversity and minimum meal frequency requirements within the previous 24 h. Minimum dietary diversity refers to the consumption of foods and beverages from at least five out of eight defined food groups (breast milk; grains, roots, tubers and plantains; pulses, nuts and seeds; dairy products; flesh foods; eggs; vitamin A-rich fruits and vegetables; and other fruits and vegetables). Minimum meal frequency entails consuming solid, semi-solid or soft foods, along with milk feeds, at least the minimum number of times for their age during the previous day. Similarly, for non-breastfed children, a minimum acceptable diet was defined based on meeting minimum dietary diversity and minimum meal frequency requirements within the previous day, along with at least two milk feeds(25).
Statistical analysis
Data were tabulated by double-independent typing and analysed using the R software (version 4.3.2; R Core Team, R Foundation for Statistical Computing). The variables were described in absolute and relative frequencies. Initially, the exogenous and endogenous variables were compared according to the presence of stunting using either Pearson’s χ2 test or Fisher’s exact test, as appropriate.
For the sEM, the lavaan 0.6–16 package was used. For this analysis, cases with missing data cannot be included (Fig. 1). The conceptual path diagram used to address the hypothesis that poor level of parental education and child factors have effects on length-for-age indicator was generated from a set of variables based on previous research(Reference Islam, Zafar Ullah and Mainali18–Reference Mulyaningsih, Mohanty and Widyaningsih21), using causal assumptions. The model applied in sEM was built using a directed acyclic graph (DAG) through DAGitty 3.1 (available at www.dagitty.net)(Reference Textor, Benito van der and Gilthorpe26). The model parameters were estimated using the maximum likelihood technique, given that the variables are normally distributed, the sample size is large and the endogenous variable is continuous. Nevertheless, we integrated modifications that align with theoretical plausibility and avoided introducing unnecessary complexity to the model. Standardised direct effect (SDE) and standardised indirect effect (SIE) were presented as standardised coefficients from simultaneous linear regressions. The model fit was analysed using the following indices: χ2 test, χ2/degrees of freedom ratio, root mean square error of approximation, comparative fit index and Tucker–Lewis index. Concerning these indices, it is expected that there will be no difference between the implied and observed matrices, as indicated by absolute fit criteria, and foresees a strong correlation among the examined variables, as suggested by relative fit criteria(Reference Schreiber, Nora and Stage10).
Results
A total of 1448 children were analysed by the sEM technique. Among them, 9·3 % were stunted, with 60·7 % of these being male, and 48·9 % of their parents had less than 9 years of education. The sample has a homogeneous distribution of sex and age. Regarding food intake, 93 % of the children were ever breastfed, and nearly 68 % had a minimum acceptable diet. In contrast, almost 33 % of the children had consumed sweet beverages (Table 1).
Parental level of education and child characteristics according to the stunting status of socially vulnerable children aged 6–24 months, Alagoas, Brazil, 2018
* Stunted = length-for-age < −2 Z-score.
† Non-stunted = length-for-age ≥ −2 Z-score.
‡ Values are presented as total number (n) and frequency (%) for categorical variables.
The path diagram of the structural equation model is shown in Fig. 2. The values shown over the solid arrows indicate the statistically significant SDE and SIE. The model had a good fit according to the recommended indexes and its acceptable cut-offs (Table 2). When analysing the SDE, we found a negative effect of the children’s younger age (sdE: −0·06; 95 % CI: −0·025, −0·002) and the use of bottle feeding (sdE: −0·11; 95 % CI: −0·407, −0·157) and do not have a minimum acceptable diet (sDE: −0·09; 95 % CI: −0·350, −0·102) on the length-for-age indicator. On the other hand, being female (SDE: 0·08; 95 % CI: −0·075, 0·300), a higher birth weight (SDE: 0·33; 95 % CI:0·618, 0·826), being ever breastfed (sdE: 0·07; 95 % CI: 0·105, 0·552) and a higher level of parental education (SDE: 0·09; 95 % CI: 0·015, 0·047) showed a positive effect on the child’s length-for-age (Fig. 2, Table 3).
Structural equation model exploring the effects of parental level of education and child factors on length-for-age indicator in socially vulnerable children aged 6–24 months, Alagoas, Brazil, 2018.
Study model fit analyses and acceptable cut-offs fit indexes according to Schreiber et al., 2006
df, degrees of freedom ratio; RMSEA, root mean square error of approximation; CFI, comparative fit index; TLI, Tucker–Lewis index.
Direct and indirect effects on length-for-age indicator in socially vulnerable children aged 6–24 months, Alagoas, Brazil, 2018
SDE, standardised direct effect; SIE, standardised indirect effect.
* The differences are statistically different from zero at the 95 %CI.
Furthermore, among the SIE, the model also demonstrated a negative effect of sweet beverage consumption (SIE: −0·08; 95 % CI: −0·980, −0·196) and a positive effect of being ever breastfed (SIE: 0·06; 95 % CI: 0·155, 1·564) on the child’s length-for-age through parental level of education as a mediator (Fig. 2, Table 3).
Discussion
This study used sEM analysis to understand the effects of parental level of education and children’s individual factors on the length-for-age indicator in socially vulnerable children under 2 years of age from a Brazilian state. Concerning the theoretical model proposed by this study, all indicators demonstrated to have either an SDE, SIE or both on the children’s length-for-age indicator.
In view of our findings, we highlight the effects of adequate feeding practices on increasing the child’s length-for-age. Being ever breastfed was the only factor that exhibited both positive SDE and SIE (mediated by parental level of education) on the child’s length-for-age indicator. Coinciding with this discovery, a prior investigation conducted by Campos et al. (Reference Campos, Vilar-Compte and Hawkins27) with children aged 6–35 months revealed evidence supporting a protective impact of any breast-feeding against stunting in comparison with children who were never breastfed. Victora et al. (Reference Victora, Bahl and Barros28) state that breast-feeding offers substantial protection against diarrhoea and infections in children, thereby reducing the risk of undernutrition during early life.
It is well known the health benefits of breast-feeding to the infant. Victora et al. (Reference Victora, Bahl and Barros28) suggested that breast milk acts as personalised medicine for babies. Along with this, the authors observed, through The Lives Saved Tool(Reference Walker, Tam and Friberg29), elevating breast-feeding rates to nearly universal levels could potentially save 13·8 % of the lives of children under 2 years. More recent findings emphasise the crucial role played by the combined nutritive and non-nutritive bioactive components of breast milk, such as hormones, immunological factors, oligosaccharides and live microbes. These elements, in conjunction with the biological, social and psychological factors of the mother–child relationship, are fundamental for ensuring proper child growth and development. This unique combination sets breast milk apart, making it unparalleled compared with commercial milk formula(Reference Bode, Raman and Murch30–Reference Pérez-Escamilla, Tomori and Hernández-Cordero32). Therefore, the WHO(33) recommends exclusive breast-feeding during the first 6 months of life and continuing breast-feeding alongside the introduction of complementary foods until the child reaches at least 2 years of age.
In contrast, the bottle feeding practice was revealed to have a negative SDE on the child’s length-for-age. Similarly, evidence from the literature has found bottle feeding to be a significant predictor of stunting. Jeyakumar et al. (Reference Jeyakumar, Babar and Menon34) and Fekadu et al. (Reference Fekadu, Mesfin and Haile35) identified that children under 2 years of age who were bottle-fed had an increased risk of stunting by 1·5 and 3·8 times, respectively, compared with their counterparts. The use of bottle feeding can exert enduring influences on a child’s feeding patterns. Opting for convenience feeding via bottles might hinder the consumption of age-appropriate semi-solid and solid foods. This practice also curtails the intake of diversified foods, impacting dietary diversity and contributing to undernutrition. Furthermore, insufficient knowledge about the nutritional adequacy of bottle content and the proper dilution of formulas may result in nutrient deficits that hinder optimal growth(Reference Jeyakumar, Babar and Menon34).
Additionally, we found that not reaching a minimum acceptable diet has a negative SDE on length-for-age. This aligns with previous research indicating that achieving a minimum acceptable diet is associated with a significant and adverse impact on length-for-age, which suggests that children meeting the minimum acceptable diet indicator are less susceptible to experiencing stunting(Reference Binamungu, Kimera and Mkojera36,Reference Umwali, Kunyanga and Kaindi37) . Achieving a minimum acceptable diet implies that the child has been provided with meals at appropriate frequencies and diversity, ensuring that energy and nutrient requirements are met, which is crucial for proper growth and development. Insufficient meal diversity and frequency place children at risk of malnutrition, particularly stunting and micronutrient deficiencies, and increase susceptibility to morbidity and mortality(25). According to UNICEF(38), globally, 478 million children under the age of 5 years experience food poverty, meaning they lack access to and consumption of a minimum dietary diversity in early childhood. Among them, 202 million children (one in three) endure severe food poverty, indicating that they are being fed with only 0–2 out of the recommended eight food groups per d.
In line with Wells et al. (Reference Wells, Sawaya and Wibaek4), the correlation between inadequate energy intake, micronutrient deficiencies and the intergenerational cycle of undernutrition hampers growth and diminishes the metabolic capacity for homoeostasis. Consequently, it leads to a spectrum of adverse health outcomes, including infection, inflammation, gut dysbiosis, cardiometabolic disease and feto-pelvic disproportion. Furthermore, specific diseases increase susceptibility to various forms of malnutrition. Encountering malnutrition in the critical phase of the early childhood can lead to the depletion of human potential, perpetuating the intergenerational cycle of poverty and undernutrition(Reference Victora, Adair and Fall3,Reference Wells, Sawaya and Wibaek4) .
Although the sweet beverage consumption did not demonstrate an SDE on the child’s length-for-age, the observed SIE, mediated by parental level of education, could be associated with increased sugar intake in the diet, leading to disruptions in dietary patterns and nutrient intakes, as the consumption of sugary beverages could displace other foods needed to meet nutrient requirements for optimal growth and nutrition(Reference Fidler Mis, Braegger and Bronsky39). Jaime et al. (Reference Jaime, Prado and Malta40) drew attention to the elevated prevalence of sugary beverage consumption among Brazilian children under the age of 2 years. The authors demonstrated that sociodemographic characteristics, such as low level of parental education, and family habits influence this inappropriate feeding practice. Besides that, numerous studies have shown a robust association between the consumption of sugar-sweetened beverages and weight gain, which can contribute to childhood malnutrition and has potential metabolic impacts extending into adulthood(Reference Calcaterra, Cena and Magenes41,Reference Vinke, Blijleven and Luitjens42) .
The above-mentioned negative impact of the sweet beverage consumption on child growth could be more pronounced in vulnerable populations, given their already limited access to diverse and nutritious diets, as well as reduced access to healthcare(38). While the rise in financial support facilitated by the BFP’s income transfer provides beneficiaries with better access to food, it does not ensure its nutritional quality(Reference Mendes, Marçal and Rinaldi13,Reference Almeida, Mesquita and Silva43) . An unprecedented study published by UNICEF Brazil(44) investigated the knowledge, attitudes and practices related to nutrition among beneficiaries of the BFP. The findings revealed a concerning prevalence of unhealthy feeding practices, with 72 % of children under 2 years old consuming at least one type of ultra-processed food and 32 % consuming sugar-sweetened beverages. The study linked the high consumption of sugary drinks to inadequate breast-feeding practices and early weaning. Additionally, it is crucial to emphasise that sugar-sweetened beverages, as well as foods with added sugar, should not be consumed by children under 2 years old(45).
We also identified that the parental level of education not only mediates the SIE effect of the sweet beverage consumption but also has an SDE on the child’s length-for-age. In line with this, a study conducted in Ethiopia found that maternal education was strongly associated with appropriate complementary feeding practices among children aged 6–23 months, demonstrating that mothers with higher education were almost 3x more likely to practice suitable feeding practices than those with no formal education(Reference Kassa, Meshesha and Haji46). In the same vein, Nkurunziza et al. (Reference Nkurunziza, Meessen and Van geertruyden47) observed that children from a non-educated household head were at risk of stunting and severe stunting 1·9 and 2·1 times higher, respectively, than those from households headed by an individual with secondary school education and above. Enhanced maternal educational attainment correlates with improved childcare practices in the context of health and nutrition. It is also associated with a decreased likelihood of stunting and an increased capacity to access and derive benefits from health interventions(Reference Black, Victora and Walker48,Reference Vaivada, Akseer and Akseer49) .
Furthermore, being female and a higher birth weight were linked to a positive SDE, while the child’s younger age was to a negative SDE on length-for-age. This aligns with findings in other studies. However, our results deviate from the common pattern observed in most studies, particularly regarding younger age. Unlike the widespread association noted between older age and stunting in the literature(Reference Islam, Zafar Ullah and Mainali18,Reference Das, Chanani and Shah More50–Reference Benjamin-Chung, Mertens and Colford53) , our study reveals a different trend, supported by new evidence. A recent pooled analysis of longitudinal studies in low- and middle-income countries encompassing children aged 0–24 months demonstrated that the most significant onset of stunting occurred between birth and 3 months of age. This pattern was notably prevalent in countries with lower healthcare spending, higher under-5 mortality rates and increased levels of poverty(Reference Benjamin-Chung, Mertens and Colford53). In the face of these new insights, Victora et al. (Reference Victora, Christian and Vidaletti54) underscore the importance of directing attention to the initial 1000 d from conception to age 2, providing fresh perspectives on the pivotal significance of maternal nutrition.
Concerning the other aspects, Ali et al. (Reference Ali, Saaka and Adams55) and Thurstans et al. (Reference Thurstans, Opondo and Seal56) have reported higher odds of stunting among male children than female children (OR 1·99 and OR 1·29, respectively). Although this association is already well established in the literature, the explanations currently offered for these variations are diverse and frequently conjectural, lacking substantial support from direct evidence. Regarding birth weight, Jana et al. (Reference Jana, Dey and Ghosh57) reported high odds of stunting in children born with low birth weight (OR 1·46), and Aboagye et al. (Reference Aboagye, Ahinkorah and Seidu58) found a similar association with an OR of 1·68. This can be explained by inadequate prenatal nutrition and restricted fetal growth leading to lower birth weights, which in turn impact postnatal growth. Insufficient nutrient reserves acquired during gestation, combined with increased vulnerability to infections and compromised immunity, hinder a child’s optimal growth, contributing to early childhood stunting(Reference Victora, Adair and Fall3,Reference Hack, Klein and Taylor59) .
Moreover, the stunting prevalence found in this study (9·3 %) was more than twice (4·6 %) that reported previously by Ferreira et al. (Reference Ferreira, Albuquerque and Santos60). They used data from 2015, focusing on children under 24 months from all socio-economic statuses in the state of Alagoas. It is important to emphasise that our study sample consists entirely of socially vulnerable children assisted by the BFP, and the data collection occurred before the COVID-19 pandemic. With the documented growth of health inequities and economic disparities, resulting in an increase in food insecurity, our findings raise concerns about the potential impact of the pandemic on long-term stunting prevalence in Alagoas and globally(1,38) .
It is important to consider that the present study has some limitations. The conceptual path diagram used in this study does not include sociodemographic characteristics that are often associated with child stunting, such as household income, family size, housing location (urban/rural), drinking water, sanitation and food insecurity(Reference Stewart, Iannotti and Dewey8). We made an effort to include these factors in the sEM model, however, their inclusion did not result in an improvement in the fit indexes, suggesting that these variables did not enhance the overall model fit. This can probably be attributed to the homogeneity of the sample, which comprises individuals with high social vulnerability assisted by the BFP. Another limitation was the non-probabilistic sampling approach, which may cause selection bias, including only those who have more access to health services and possibly greater healthcare. However, we reinforce that all eligible children were invited to participate in the study, and the data collection was performed in all Community Health Centre in the municipalities, both in urban and rural areas. Despite these limitations, this study provides up-to-date data using the sEM technique regarding the effects of parental and children’s factors on length-for-age among socially vulnerable children under 2 years of age from a poor Brazilian state. Considering the potential similarities between the investigated population and those in various regions nationally and globally, the results of this research may prove useful to other regions.
In conclusion, the findings of the sEM analysis underscore the continued significance of addressing stunting among socially vulnerable Brazilian children under 2 years old. Notably, our research emphasises the pivotal role of proper feeding practices, including promoting breast-feeding initiation and duration, avoiding bottle feeding, ensuring access to a diverse and nutrient-rich diet at a proper frequency to achieve a minimum acceptable diet, and limiting sweet beverage consumption, in promoting increased length-for-age in this vulnerable population. Additionally, urgent regulatory measures are needed to control the dissemination of unhealthy foods, particularly those targeted at children. These insights contribute valuable knowledge for crafting targeted interventions, policies and programmes, which, in conjunction with the BFP, aimed at enhancing the nutritional well-being and promoting adequate linear growth and development of young children in similar contexts.
Acknowledgements
The research in the municipalities had the consent and partnership of the Municipal and State Health Departments.
This work is part of a project entitled: Evaluation of the management and operationalisation of the National Program for Iron and Vitamin A Supplementation and their relationship with the nutritional status of children aged 6–24 months in municipalities of Alagoas state, carried out with funds from Conselho Nacional de Desenvolvimento Científico e Tecnológico - CNPq (process number 408445/2017-3). This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brasil (CAPES) – Finance Code 001. The funder had no role in the design, analysis or writing of this article.
M. M. M. N. participated in designing the study, acquiring and entering data, analysing and interpreting the results, and drafting the article. L. C. S., R. C. E. M. and T. M. M. F. were involved in data analysis, interpretation and writing. A. P. G. C. obtained financial support, contributed to project conception, coordinated all implementation steps and performed the final article review. All authors approved the final version for submission.
The author(s) declare none.
