Abstract:

Habits and behaviors related to obesity risk are strongly associated with the family environment and are affected by socioeconomic factors. Structural equation modeling (SEM) allows us to hypothesize on how the relationships between these factors occur and measure their impact. This study aimed to explore the relationship between family socioeconomic indicators and childhood obesity, mediated by habits linked to energy balance, applying a SEM. A cross sectional study was performed on 861 Argentinian schoolchildren aged 6-12 years, from 2015 to 2016. The model included three latent variables: socioeconomic status, healthy habits, and obesity. Socioeconomic status indicators and healthy habits were surveyed by self-administered parental questionnaires, whereas obesity indicators were evaluated with anthropometry. The applied model showed an acceptable fit (NFI = 0.966; CFI = 0.979; RMSEA = 0.048). Socioeconomic status positively influenced parental education, health insurance, and car possession, while negatively influenced crowding (p < 0.001). Healthy habits significantly influenced physical activity, meals frequency, and sleep hours, while negatively influenced sedentary hours and mother’s nutritional status (p < 0.001). Obesity factor positively influenced body mass index, body fat, and waist-to-height ratio (p < 0.001). Finally, socioeconomic status positively influenced health habits, which in turn negatively influenced obesity factor. Healthy habits (especially physical activity and mother’s nutritional status) mediated the relationship between socioeconomic status and child obesity. Further research should include other indicators related to diet, eating habits, and physical activity like neighborhood characteristics.

Keywords:
Pediatric Obesity; Healthy Lifestyle; Structural Equation Modeling; Socioeconomic Factors

Resumen:

Los hábitos y comportamientos relacionados con el riesgo de obesidad están fuertemente asociados al entorno familiar y se ven afectados por factores socioeconómicos. El modelo de ecuaciones estructurales (MEE) permite plantear hipótesis sobre cómo se dan las relaciones entre estos factores y medir su impacto. El objetivo del presente estudio fue explorar la relación entre los indicadores socioeconómicos familiares y la obesidad infantil, mediada por hábitos relacionados con el balance energético, aplicando un MEE. Se realizó un estudio transversal con 861 escolares argentinos de 6 a 12 años entre 2015 y 2016. El modelo incluyó tres variables latentes: nivel socioeconómico, hábitos saludables y obesidad. Los indicadores nivel socioeconómico y hábitos saludables se midieron mediante cuestionarios autoadministrados a los padres, mientras que los indicadores de obesidad se obtuvieron mediante antropometría. El modelo aplicado presentó un ajuste aceptable (NFI = 0,966; CFI = 0,979; RMSEA = 0,048). El nivel socioeconómico influyó positivamente en la escolaridad de los padres, en el plan de salud y en el hecho de tener automóvil, mientras que influyó negativamente en el hacinamiento (p < 0,001). La variable hábitos saludables influyó significativamente en la actividad física, en la frecuencia de las comidas y en las horas de sueño, pero influyó negativamente en las horas sedentarias y en el estado nutricional materno (p < 0,001). El factor obesidad influyó positivamente en el índice de masa corporal, en la grasa corporal y en la razón cintura/talla (p < 0,001). Finalmente, el nivel socioeconómico influyó positivamente en los hábitos saludables, que, a su vez, influyeron negativamente en la obesidad. Los hábitos saludables, en especial la actividad física y el estado nutricional materno, moderan la relación entre el nivel socioeconómico y la obesidad infantil. Las nuevas investigaciones deberían incluir otros indicadores relacionados con la dieta, con los hábitos alimentarios y con la actividad física, así como con las características del vecindario.

Palabras-clave:
Obesidad Infantil; Estilo de Vida Saludable; Modelado de Ecuaciones Estruturales; Factores Socioeconómicos

Resumo:

Hábitos e comportamentos relacionados ao risco de obesidade estão fortemente associados ao ambiente familiar e afetados por fatores socioeconômicos. A modelagem de equações estruturais (MEE) permite levantar hipóteses sobre como ocorrem as relações entre esses fatores e medir seu impacto. O objetivo do presente estudo foi explorar a relação entre os indicadores socioeconômicos familiares e a obesidade infantil, mediada por hábitos ligados ao balanço energético, aplicando uma MEE. Um estudo transversal foi realizado com 861 escolares argentinos de 6 a 12 anos entre 2015 e 2016. O modelo incluiu três variáveis latentes: nível socioeconômico, hábitos saudáveis e obesidade. Os indicadores do nível socioeconômico e de hábitos saudáveis foram medidos por meio de questionários autoaplicáveis aos pais, enquanto os indicadores de obesidade foram obtidos por meio de antropometria. O modelo aplicado apresentou um ajuste aceitável (NFI = 0,966; CFI = 0,979; RMSEA = 0,048). O nível socioeconômico influenciou positivamente a escolaridade dos pais, o plano de saúde e a posse de automóvel, enquanto influenciou negativamente a aglomeração (p < 0,001). A variável hábitso saudáveis influenciou significativamente a atividade física, a frequência das refeições e as horas de sono, mas influenciou negativamente as horas sedentárias e o estado nutricional materno (p < 0,001). O fator obesidade influenciou positivamente o índice de massa corporal, a gordura corporal e a relação cintura/estatura (p < 0,001). Finalmente, o nível socioeconômico influenciou positivamente os hábitso saudáveis, que por sua vez influenciou negativamente a obesidade. Hábitos saudáveis, especialmente atividade física e estado nutricional materno, mediam a relação entre nível socioeconômico e obesidade infantil. Novas pesquisas devem incluir outros indicadores relacionados à dieta, hábitos alimentares e atividade física, como características da vizinhança.

Palavras-chave:
Obesidade Infantil; Estilo de Vida Saudável; Modelagem de Equação Estrutural; Fatores Socioeconômicos

Introduction

Despite the efforts launched by international health organizations to reduce obesity in children, the rates are still increasing all over the world, exceeding 30% among school-aged children in developing countries ¹1. World Health Organization. Commission on Ending Childhood Obesity (ECHO): facts and figures on childhood obesity. http://www.who.int/end-childhood-obesity/facts (accessed on 20/Feb/2022).
http://www.who.int/end-childhood-obesity... . With few exceptions, the trend has been an increase in obesity rates worldwide. Argentina lacks systematic surveillance programs for children and adolescents that allows to quantify this phenomenon. However, our research team has verified this upward trend with consecutive surveys in the province of La Pampa (Northern Patagonia), since 1990 ²2. Orden AB, Bucci PJ, Petrone S. Trends in weight, height, BMI and obesity in schoolchildren from Santa Rosa (Argentina), 1990-2005/07. Ann Hum Biol 2013; 40:348-54.. Then, only 15% of schoolchildren were overweight and 2% were obese, which strongly contrasts with the current values of 20% and 12%, respectively ²2. Orden AB, Bucci PJ, Petrone S. Trends in weight, height, BMI and obesity in schoolchildren from Santa Rosa (Argentina), 1990-2005/07. Ann Hum Biol 2013; 40:348-54.^,33. Orden AB, Lamarque MS, Chan D, Mayer MA. Short sleep and low milk intake are associated with obesity in a community of school aged children from Argentina. Am J Hum Biol 2019; 31:e23224.^,44. Orden AB, Lamarque MS, Apezteguía MC. Trend in childhood obesity reflects socioeconomic status in Argentina. Ann Hum Biol 2019; 46:531-6.. This trend calls for primary and secondary interventions, which require the investigation of the responsible factors.

Several factors have been considered as relevant causes of obesity. At the basic level are the classic “big two”, that is, income and energy expenditure ⁵5. Chaput JP, Sjödin AM, Astrup A, Després JP, Bouchard C, Tremblay A. Risk factors for adult overweight and obesity: the importance of looking beyond the 'big two'. Obes Facts 2010; 3:320-7.^,66. Dhurandhar EJ, Keith SW. The aetiology of obesity beyond eating more and exercising less. Best Pract Res Clin Gastroenterol 2014; 28:533-44.. In turn, these components are influenced by multiple variables that make up networks rather than being a uni or bidirectional relationships. Notwithstanding this complexity, the determinants of overweight and obesity are traditionally analyzed using logistic or linear models ⁷7. Ells LJ, Hancock C, Copley VR, Mead E, Dinsdale H, Kinra S, et al. Prevalence of severe childhood obesity in England: 2006-2013. Arch Dis Child 2015; 100:631-6.^,88. White J, Rehkopf D, Mortensen LH. Trends in socioeconomic inequalities in body mass index, underweight and obesity among English children, 2007-2008 to 2011-2012. PLoS One 2016; 11:e0147614.. Although such models are widely used for this issue, they do not facilitate the estimation of causal and indirect effects in a single, integrated equation. Structural equation modeling (SEM), also known as analysis of the structure of covariance and analysis of the latent variable, is a technique that integrates confirmatory factor analysis and multiple regression analysis and considers multiple independent variables and modeling of interactions among variables ⁹9. Ainur AK, Sayang MD, Jannoo Z, Yap BW. Sample size and non-normality effects on goodness of fit measures in structural equation models. Pertanika Journal of Science & Technology 2017; 25:575-86..

The shared home environment is probably the closer influence on childhood obesity, which is reflected by strong correlations between parents and children’s habits. Such influence characterizes lifestyles and behaviors transmitted from parents to children with socialization. In fact, the shared environment exposes parents and children to common obesogenic factors, such as unhealthy eating and sedentary habits ¹⁰10. Birch LL, Davison KK. Family environmental factors influencing the developing behavioral controls of food intake and childhood overweight. Pediatr Clin North Am 2001; 48:893-907.. The influence of socioeconomic status as a determinant or mediator of behavior adds further complexity to our understanding of obesity and its related behaviors ¹¹11. Hendrie GA, Coveney J, Cox DN. Defining the complexity of childhood obesity and related behaviours within the family environment using structural equation modelling. Public Health Nutr 2012; 15:48-57.. Eating habits - among others related to obesity - are probably influenced by family income. Although most people know about healthy eating, families with higher socioeconomic status may afford these kinds of foods when compared with lower income families ¹²12. Aguirre P. Ricos flacos y gordos pobres: la alimentación en crisis. Buenos Aires: Capital Intelectual; 2010.. Similarly, the influence of the family on energy expenditure and sedentary behaviors of children is related to socioeconomic status, both by access to sports and recreational activities ¹³13. Brockman R, Jago R, Fox KR, Thompson JL, Cartwright K, Page AS. "Get off the sofa and go and play": family and socioeconomic influences on the physical activity of 10-11 year old children. BMC Public Health 2009; 9:253., as well as several residential environment characteristics that encourage or restrict physical activity ¹⁴14. Murphy L, Demaio AR. Understanding and removing barriers to physical activity: one key in addressing child obesity. Br J Sports Med 2018; 52:1284-5..

The proposal of this study is to introduce a relational model among three latent variables that affect weight gain during childhood: socioeconomic status, healthy habits, and obesity. Our model hypothesizes that higher socioeconomic status is associated with healthier habits and lower childhood obesity. Although this statement is not new, the framework of the indicators may be specific to each community and therefore contribute to adapting local prevention or intervention strategies to reduce obesity.

Subjects and methods

Sampling and data collection

The study was conducted in Santa Rosa (La Pampa) between September 2015 and October 2016. We performed a cross-sectional study, based on a two stages cluster sampling. The first stage included neighborhood and school mapping. The second stage consisted of randomly selecting one school per neighborhood. At each school, one division from the first to the sixth grade was randomly selected as well. The sample size was estimated as n = z²p(1 - p)/e² with a 95% confidence level (z = 1.96), an overweight and obesity prevalence of 25%, and a 2.5% estimation error. The whole sample comprised 1,366 healthy children aged 6-12 years attending public and private schools.

The research protocol was performed following the guidelines and recommendations of the Ethics Committee of the Pediatric Research and Development Institute, Hospital Sor M. Ludovica, in accordance with national laws and the principles expressed in the Declaration of Helsinki. After being approved by the Ministry of Culture and Education of the Province of La Pampa an informed consent was requested from parents or legal guardians, who were previously informed about the objectives and methods of the research. They were also asked to complete a self-administered questionnaire that elicited information on the following domains:

(a) Socioeconomic status: among the variables related to socioeconomic status we consider the maximum educational level of the father and mother (i.e., primary, secondary, or tertiary/university studies); crowding or persons/room ratio; owning a car (categorized by model > 5 years or ≤ 5 years), while no car was scored as 0; and having (1) or not (0) health insurance.

(b) Nutrition and energy expenditure were assessed by daily meal frequency, nocturnal sleep duration, do or not do extra-school physical activity, and daily sedentary hours, i.e., activities without energy expenditure, especially the use of screens.

(c) Parental anthropometry: parents were asked their weight and height to calculate the body mass index (BMI). Then, father and mother’s nutritional status was categorized according to the International Obesity Task Force (IOTF) cut-offs (thinness: < 18.5kg/m²; normal weight: 18.5-24.9kg/m²; overweight: 25-29.9kg/m²; and obesity: ≥ 30kg/m²) ¹⁵15. Cole TJ, Lobstein T. Extended international (IOTF) body mass index cut-offs for thinness, overweight and obesity. Pediatr Obes 2012; 7:284-94.. Only a few subjects were classified with thinness, so they were grouped with normal weight.

Anthropometric measurements of children were made at the schools following standardized procedures by a trained observer (A.B.O.). Weight was measured to the nearest 0.1kg with a digital scale Tanita BF-350 (Tanita Corp.; https://www.tanita.com/en/) and height was measured to the nearest 0.1cm with a stadiometer SECA S-213 (SECA; https://www.seca.com/en_us.html), with the head of the child in the Frankfurt plane position. BMI was calculated as weight (kg) divided by squared height (m²) and transformed into standard deviation scores (Z_BMI) based on the IOTF reference ¹⁵15. Cole TJ, Lobstein T. Extended international (IOTF) body mass index cut-offs for thinness, overweight and obesity. Pediatr Obes 2012; 7:284-94.. Waist circumference was measured as the smallest circumference between the lower end of the sternum (xiphoid process) and the umbilicus with an inelastic fiberglass tape at the end of normal expiration. Then we estimate the waist-to-height ratio (W/H). Skinfold thicknesses were measured with a Lange caliper at four sites: tricipital (SK₁), at the mid-point on the posterior line of the upper arm; bicipital (SK₂) was measured at the same point on the anterior line of the arm; subscapular skinfold (SK₃) was taken under the inferior angle of the left scapula; and suprailiac skinfold (SK₄) was measured by lifting a horizontal skinfold above the iliac bone surface. Then, body fat percentage was calculated using the equations of Brook ¹⁶16. Brook CGD. Determination of body composition of children from skinfold measurements. Arch Dis Child 1971; 46:182-4. based on density (D = 1.1690 - 0.0788*log[Σ⁴ _i=1 SK_i ] for boys, and D = 1.2063 - 0.0999*log[Σ⁴ _i=1 SK_i ] for girls, and according to Siri ¹⁷17. Siri WE. Body composition from fluid spaces and density: analysis of methods. In: Brozek J, Henzchel A, editors. Techniques for measuring body composition. Washington DC: National Academy of Sciences; 1961. p. 224-44. (BF(%) = [(4.95/D) - 4.5]*100).

Structural equation modeling

SEM allows us to simultaneously evaluate the relationship between indicators (observable or measurable variables) and one or more latent factors (non-observable variables, but indirectly evaluated with indicators), as well as the relationship between latent factors ¹⁸18. Ortiz MS, Fernández-Pera M. Modelo de Ecuaciones Estructurales: una guía para ciencias médicas y ciencias de la salud. Ter Psicol 2018; 36:47-53.. These multiple relationships between observable and latent variables are represented by path diagrams, which constitute a hypothesis about the proposed relationships. Thus, SEM is a confirmatory method of that hypothesis. In SEM, a model is said to fit the observed data to the extent that the model-implied covariance matrix is equivalent to the empirical covariance matrix ¹⁹19. Schermelleh-Engel K, Moosbrugger H, Müller H. Evaluating the fit of structural equation models: tests of significance and descriptive goodness-of-fit measures. MPR-online 2003; 8:23-74.. Also, a good-fitting measurement model is required before interpreting the causal paths of the structural model.

Our model proposes the existence of relationships between three latent variables: socioeconomic status and healthy habits, both influencing a third called obesity. In the case of socioeconomic status, its indicators are maximum education level of the father, maximum education level of the mother, crowding or persons/room ratio, health insurance, owning a car. Healthy habits is measured by physical activity, nocturnal sleep duration, sedentary hours, meal frequency, mother’s nutritional status, and father’s nutritional status. And obesity is measured by BMI, W/H, and body fat.

Data analysis

Quantitative variables were examined for normality using the Kolmogorov-Smirnov test: z-score of BMI and W/H had an approximately normal distribution and were reported as mean ± standard deviation (SD), while crowding or persons/room ratio, sedentary hours, and body fat had a log-normal distribution and were reported as geometric mean (GM) with its respective 95% confidence interval (95%CI).

The structural equation modeling was executed using the lavaan 0.6-12 ²⁰20. Rosseel Y. lavaan: an R package for Structural Equation Modeling. J Stat Softw 2012; 48:1-36. and lavaanPlot (https://CRAN.R-project.org/package=lavaanPlot) packages of the R software, version 4.2.1 (http://www.r-project.org). For the parameter estimates, we used weighted least squares method with mean and variance adjusted (WLSMV); the diagonally weighted least squares (DWLS) method was used to estimate model parameters and the full weight matrix to compute robust standard errors, and a mean- and variance-adjusted test statistics. Factor loadings of physical activity, maximum educational level of the father, and BMI were fixed to 1 as an identification restriction. All categorical variables included in the model were binary or ordinal and were declared as such in the model function. When dealing with categorical variables polychoric or polyserial correlations were computed. SEM procedure relies on several statistical tests to determine the adequacy of the model fit to the data. Therefore, once the model was estimated, we verified the compatibility of the model adjustment as well as the obtained estimates. Seven fit indices were reported. The goodness of fit was analyzed with the most commonly used indices: comparative fit index (CFI), Tucker-Lewis index (TLI), relative fit index (RFI), normed fit index (NFI), and incremental fit index (IFI). Indices greater than 0.90 indicate a good fit, and those greater than 0.95 a very good fit ¹⁹19. Schermelleh-Engel K, Moosbrugger H, Müller H. Evaluating the fit of structural equation models: tests of significance and descriptive goodness-of-fit measures. MPR-online 2003; 8:23-74.. In addition, two measures related to the unexplained variability of the model were calculated: the root mean square error of approximation (RMSEA) which is an index of the degree to which a confirmatory structure approximates the data being modeled using the variance/covariance matrix, and the standardized root mean squared residual (SRMR). RMSEA values less than 0.06 and SRMR less than 0.08 are considered acceptable. The level of significance was set at p < 0.05.

Results

The whole sample included 653 boys and 713 girls aged 9.2 ± 1.8 years. After excluding the questionnaires with missing data, the remaining subsample included 861 children (407 boys and 454 girls), aged 9.1 ± 1.8 years. Table 1 shows descriptive statistics regarding the subsample. Table 2 shows the differences with the missing sample. Almost 31% of children were either overweight (20%) or obese (10.6%), as estimated for the total sample ³3. Orden AB, Lamarque MS, Chan D, Mayer MA. Short sleep and low milk intake are associated with obesity in a community of school aged children from Argentina. Am J Hum Biol 2019; 31:e23224.. On average, children had normal W/H values and BMI (z-score > 0) and body fat (> 25%) above the reference values. The education level was variable among the parents, with 25~35.7% of them having a tertiary or university degree. The remaining socioeconomic status data indicated that around 30% of the families had higher levels of socioeconomic status indicators while 25~30% had lower socioeconomic status.

Table 3 shows the results of model fitting based on the SEM. Values of all measures of goodness of fit were within the acceptable range, indicating that the proposed model fitted our data at the 5% significance level. Finally, the measures related to the percentage of variance not explained by the model were RMSEA = 0.048 and SRMR = 0.054. By evaluating all these indicators together, it can be concluded that the overall adjustment of the model was good.

Figure 1 and Table 4 show the results of the structural model. Values shown in the figure and table are standardized estimates. It can be observed that socioeconomic status had a positive impact on parental education, health insurance, and owning a car; but negative on crowding or persons/room ratio (p < 0.001). Regarding healthy habits, it was observed to directly and strongly influence physical activity. Likewise, this factor also had a significant positive impact on meal frequency and nocturnal sleep duration, with a negative influence on sedentary hours and mother’s nutritional status (p < 0.001). Regarding the third latent variable, it was observed that obesity had a significant positive impact on the three indicators analyzed (p < 0.001). Finally, socioeconomic status had a significant positive impact on healthy habits, which in turn negatively influenced obesity. In other words, socioeconomic status indirectly impacted obesity through healthy habits.

Discussion

Studies concerning nutrition and health behaviors among children are somewhat scarce in Argentina. Most of the studies on risk factors of obesity use bivariate models. Orden et al. ³3. Orden AB, Lamarque MS, Chan D, Mayer MA. Short sleep and low milk intake are associated with obesity in a community of school aged children from Argentina. Am J Hum Biol 2019; 31:e23224. combined this method with geometrical techniques such as multiple correspondence analysis. Under this approach, they identified profiles or clusters of individuals with similar BMI status and found that physical activity and calcium intake, as well as sleep, are protective factors of obesity. In contrast, parental obesity, shorter sleep, and lower socioeconomic status have a negative effect on weight gain in childhood. Although the influence of socioeconomic status is undeniable, its pattern of social distribution differs between countries. The global trend is consistent: obesity has increased worldwide as the increase in BMI has shifted from high to low-income sectors ²¹21. Popkin BM, Slining MM. New dynamics in global obesity facing low- and middle-income countries. Obes Rev 2013; 14:11-20.^,2222. Ulijaszek SJ, Pentecost M, Marcus C, Karpe F, Fruhbeck G, Nowicka P. Inequality and childhood overweight and obesity: a commentary. Pediatr Obes 2017; 12:195-202.. Studies in our population have verified such a trend during the last decade, in which obesity is still growing only in low socioeconomic status children ⁴4. Orden AB, Lamarque MS, Apezteguía MC. Trend in childhood obesity reflects socioeconomic status in Argentina. Ann Hum Biol 2019; 46:531-6..

The fit model using SEM allowed the evaluation of the influence of socioeconomic status on the health behaviors of children and childhood obesity. Screen time, physical activity, and sleep have been previously identified as relevant factors for children’s obesity risk ²³23. Saunders TJ, Gray CE, Poitras VJ, Chaput J-P, Janssen I, Katzmarzyk PT, et al. Combinations of physical activity, sedentary behaviour and sleep: relationships with health indicators in school-aged children and youth. Appl Physiol Nutr Metab 2016; 41(6 Suppl 3):S283-93.^,2424. Felso R, Lohner S, Hollódy K, Erhardt E, Molnár D. Relationship between sleep duration and childhood obesity: Systematic review including the potential underlying mechanisms. Nutr Metab Cardiovasc Dis 2017; 27:751-61.^,2525. Fang K, Mu M, Liu K, He Y. Screen time and childhood overweight/obesity: a systematic review and meta-analysis. Child Care Health Dev 2019; 45:744-53.. We showed that these behaviors are influenced by latent factor “healthy habits”, which acts as a mediator between socioeconomic status and obesity in children. Also, the relationship between higher socioeconomic status and healthy habits could explain better food access, which is associated with better diet quality ²⁶26. Darmon N, Drewnowski A. Does social class predict diet quality? Am J Clin Nutr 2008; 87:1107-17.^,2727. Elsenburg LK, Corpeleijn E, van Sluijs EM, Atkin AJ. Clustering and correlates of multiple health behaviours in 9-10 year old children. PLoS One 2014; 9:e99498.. Contrary to our findings, Géa-Horta et al. ²⁸28. Géa-Horta T, Silva RCR, Fiaccone RL, Barreto ML, Velásquez-Meléndez G. Factors associated with body mass index in Brazilian children: structural equation model. Nutr Hosp 2017; 34:308-14. used SEM to test a model that predicts a positive relationship between child BMI and socioeconomic status. This relationship was mediated by more frequent intakes of obesogenic foods, according to data from a 2006-2007 Brazilian National Health Survey. These findings could be explained by rapid changes in the food environment, which, in the past decade, has experienced an increase of ultra-processed, high-energy, and less nutrient-dense foods ²⁹29. Pan American Health Organization. Ultra-processed food and drink products in Latin America: sales, sources, nutrient profiles, and policy implications. Washington DC: Pan American Health Organization; 2019.. In Argentina, data from the first national nutrition survey collected in 2005 revealed that higher quintiles of income showed a higher proportion of ultra-processed food consumption ³⁰30. Drake I, Abeyá Gilardon E, Mangialavori G, Biglieri A. Description of nutrient consumption based on the level of industrial food processing. National Survey on Nutrition and Health of 2005. Arch Argent Pediatr 2018; 116:345-52.. While recent reports from the second national nutrition survey showed that lower-income quintiles had higher consumption of sugar-sweetened beverages and snacks and less consumption of fruits, dairy, and fish ³¹31. Ministerio de Salud y Desarrollo Social de la Nación. 2ª Encuesta Nacional de Nutrición y Salud: resumen ejecutivo. Buenos Aires: Ministerio de Salud y Desarrollo Social de la Nación; 2019.. A food price analysis suggested that, compared to an unhealthy food basket, a healthy one costs around 80% more ³²32. Viego V, Temporelli KL, Cattáneo CA, Merino C, Miotto C. Costs of healthy and unhealthy food in Argentina: 2002-2015. Applied Food Science Journal 2017;1: 4-6..

Many studies examining the relationship between the nutritional status of parents and children have found a strong association between parents and children with obesity. Systematic reviews have observed that both fathers and mothers with obesity have influence on the obesity of their children ³³33. Wang Y, Min J, Khuri J, Li M. A Systematic examination of the association between parental and child obesity across countries. Adv Nutr 2017; 8:436-48.^,3434. Lee JS, Jin MG, Lee HJ. Global relationship between parent and child obesity: a systematic review and meta-analysis. Clin Exp Pediatr 2022; 65:35-46.. Similar results were previously found using logistic and multivariate methods ³3. Orden AB, Lamarque MS, Chan D, Mayer MA. Short sleep and low milk intake are associated with obesity in a community of school aged children from Argentina. Am J Hum Biol 2019; 31:e23224.. SEM model allowed us to observe intermediate relationships regarding maternal nutritional status and healthy habits. This may be related to women’s role in childcare and household feeding. Factors that mediate the relationship between the father’s nutritional status and children obesity will need further exploration.

Previous research has examined the relationship between various behavioral factors associated with eating habits (breakfast, fast food, junk food, and sugary drinks), physical activity, and sleep. Researchers have attempted to identify clusters in which certain behavioral profiles are associated with overweight/obesity. However, there does not appear to be a linear cause-effect process, leading to more complex patterns (for instance children with high levels of physical activity and sedentary behavior), and inconsistencies between studies ³⁵35. Leech RM, McNaughton SA, Timperio A. The clustering of diet, physical activity and sedentary behavior in children and adolescents: a review. Int J Behav Nutr Phys Act 2014; 11:4.. These “mixed patterns” are frequently found in the literature ³⁶36. Pereira S, Katzmarzyk PT, Gomes TN, Borges A, Santos D, Souza M, et al. Profiling physical activity, diet, screen and sleep habits in Portuguese children. Nutrients 2015; 7:4345-62.^,3737. Schmiege SJ, Gance-Cleveland B, Gilbert L, Aldrich H, Gilbert KC, Barton A. Identifying patterns of obesity risk behavior to improve pediatric primary care. J Spec Pediatr Nurs 2016; 21:18-28.^,3838. Kaar JL, Schmiege SJ, Vadiveloo M, Simon SL, Tovar A. Sleep duration mediates the relationship between health behavior patterns and obesity. Sleep Health 2018; 4:442-7., and they may explain the negative correlation between sleep and physical activity found in our study.

The typification of clusters seems complex, even more so if other variables such as age, sex, and socioeconomic status are considered. It has been proposed that relationships between physical activity, weight status, and screen-based sedentary behaviors change over time and that a critical window exists between 6 and 10 years. During this time, sedentary patterns such as screen time may predict BMI more than physical activity ³⁹39. Hands BP, Chivers PT, Parker HE, Beilin L, Kendall G, Larkin D. The associations between physical activity, screen time and weight from 6 to 14 yrs: the Raine Study. J Sci Med Sport 2011; 14:397-403., highlighting the importance of prospective studies in the evolution of obesogenic behaviors.

According to our results, the impact of healthy habits on obesity is significant and similar to other studies ¹¹11. Hendrie GA, Coveney J, Cox DN. Defining the complexity of childhood obesity and related behaviours within the family environment using structural equation modelling. Public Health Nutr 2012; 15:48-57.^,2828. Géa-Horta T, Silva RCR, Fiaccone RL, Barreto ML, Velásquez-Meléndez G. Factors associated with body mass index in Brazilian children: structural equation model. Nutr Hosp 2017; 34:308-14.. While the impact may seem “small”, this is expected given that childhood obesity is determined by a myriad of biological, behavioral, environmental, and physiological factors that impact at different levels (individual, household, community, society) ⁴⁰40. Lehmann F, Varnaccia G, Zeiher J, Lange C, Jordan S. Influencing factors of obesity in school-age children and adolescents - a systematic review of the literature in the context of obesity monitoring. J Health Monit 2020; 5 Suppl 2:2-23.. In low and middle-income regions like Latin America, obesity prevention strategies should acknowledge different levels of interventions that address households with social disadvantages ³3. Orden AB, Lamarque MS, Chan D, Mayer MA. Short sleep and low milk intake are associated with obesity in a community of school aged children from Argentina. Am J Hum Biol 2019; 31:e23224.^,44. Orden AB, Lamarque MS, Apezteguía MC. Trend in childhood obesity reflects socioeconomic status in Argentina. Ann Hum Biol 2019; 46:531-6.^,4141. Ayala-Marín AM, Iguacel I, Miguel-Etayo P, Moreno LA. Consideration of social disadvantages for understanding and preventing obesity in children. Front Public Health 2020; 8:423.. It is important to understand obesity determinants at a population level; but different population groups may have different associations between the model’s factors, which is reflected in the resulting strengths of the path coefficients. This information may be useful to health professionals to tailor interventions to specific population groups.

Among the strengths of our work is the fact that childhood obesity was derived as a latent variable that explained several measured variables (W/H, body fat, and age-standardized BMI). Other studies have mainly used child BMI as a proxy for child obesity, and while BMI is a practical tool to classify individuals with high adiposity, it has been pointed out that BMI alone might have low sensitivity ⁴²42. Javed A, Jumean M, Murad MH, Okorodudu D, Kumar S, Somers VK, et al. Diagnostic performance of body mass index to identify obesity as defined by body adiposity in children and adolescents: a systematic review and meta-analysis. Pediatr Obes 2015; 10:234-44.. Similarly, socioeconomic status showed a strong effect on several observed variables, which indicates a high construct validity to represent households living conditions. In addition, the use of SEM supports our model, which can form the basis for generating hypotheses to be tested in future studies. On the other hand, our findings must be interpreted with caution. The major limitation of the study is not considering the dependency structure due to the lack of information from the sampling process (survey parameters), which could change the obtained estimates. Another limitation is the absence of food intake variables (diet diversity, consumption of fruits vegetables, or ultra-processed foods), as well as factors of the physical environment (green spaces) and family (access, support) related to energy expenditure. However, there are few studies in this field that use structural equation modeling, and to our knowledge, this is the first in Argentina. We believe that this new approach appears to be appropriate to deal with the complexity of obesity, and future research should advance this type of analysis to include even more factors in larger population studies.

Another important limitation is the loss of sample data. Indeed, children with missing data showed higher values of BMI, attended mostly public schools, and had parents with lower educational level. The relationships between public school, low parental education, and childhood obesity, previously described in this population ³3. Orden AB, Lamarque MS, Chan D, Mayer MA. Short sleep and low milk intake are associated with obesity in a community of school aged children from Argentina. Am J Hum Biol 2019; 31:e23224.^,44. Orden AB, Lamarque MS, Apezteguía MC. Trend in childhood obesity reflects socioeconomic status in Argentina. Ann Hum Biol 2019; 46:531-6., suggest that the associations observed using SEM could be stronger if there were no missing data.

In conclusion, healthy habits, especially physical activity, and mother’s nutritional status mediate the relationship between socioeconomic status and childhood obesity. The low correlation between healthy habits and obesity suggests adding other indicators such as diet, eating habits as well as neighborhood characteristics related to leisure time and physical activity.

Acknowledgments

The authors are grateful to the authorities of the Ministry of Culture and Education of the Province of La Pampa, school authorities, teachers, parents, and children for their valuable participation.

References

Publication Dates

History