Prevalence of childhood obesity in Spain and its relation with socioeconomic status and health behaviors: Population-based cross-sectional study

Objectives: This study aims to assess the current state of childhood overweight and obesity in Spain, and its relationship with socioeconomic status and health-related behaviors. Methods: Population-based cross-sectional observational study, based on the 2017 National Health Survey in minors in Spain. This study included all children surveyed who were aged 1–14 years. Childhood obesity was estimated from the z -score of the body mass index. Results: The study included 4882 children aged 1–14 years (mean 7.5). The prevalence of obesity was 18.6% (95% conﬁdence interval [CI] 18.5–18.7), while 13.5% (95% CI 13.4–13.6) were overweight. These ﬁgures represent over a million children in Spain who are obese and nearly 750,000 who are overweight. A north–south geographic gradient was apparent, with higher prevalence of unhealthy body weight in southern Spain. Factors associated with childhood obesity were low socioeconomic status, poor diet and sedentarism, among others. Conclusions: Childhood overweight in Spain is strongly associated with socioeconomic status and other factors such as diet and sedentarism. Multidisciplinary public health interventions are needed to reduce this serious health problem in children.


Introduction
In recent decades, the prevalence of obesity has increased significantly worldwide. 1In 2017, 2.4 million deaths and 70.7 million disability-adjusted life years in women, and 2.3 million deaths and 77.0 million disability-adjusted life years in men were attributed to high body mass index (BMI). 2In the USA, obesity has doubled in children and tripled in adolescents over the past three decades, becoming a serious public health problem. 3Trends across different countries are heterogeneous, with the growth in obesity stabilizing in high-income countries and accelerating in more disadvantaged areas such as some Asian countries. 4Estimates in Spain show that two thirds of the adult population and one third of children are overweight or obese. 5,6As in neighboring countries, the general rates of childhood obesity in the Spanish region of Catalonia indicate a very slight decrease in the last decade. 6he quality of child nutrition is strongly marked by socioeconomic status.Decades ago, low socioeconomic level was associated with poor growth and malnutrition, but this trend has now reversed, with low-income populations standing out as the most disadvantaged and the most affected by the global obesity epidemic. 7Reduced physical activity and sedentarism are known determinants of childhood obesity, 8 as are other classic factors such as diet 9 and the intake of sugary drinks. 10A meta-analysis of behavioral and nutritional factors in children 11 showed that having more than four sugary drinks a week increased the risk of obesity by 24%; watching more than two hours of television a day increased it by 42%, and inadequate sleep by 26%.In the opposite direction, this systematic review reported that having breakfast or performing regular physical activity reduced the risk by 34% and 30%, respectively.
Other factors associated with obesity have also been identified, such as a nationality different from the country of residence 12 and lower educational attainment in the parents.Using screens, watching television, sleep duration, and performing physical activity also have an influence on childhood obesity.
Although globally, childhood obesity continues to increase 1 studies from the past decade in a few of the more developed countries show that rates are stabilizing. 6However, it is difficult to predict the trends in the coming years, as the COVID 19 pandemic may constitute an unprecedented blow to efforts aimed at promoting a healthy body weight. 13Reducing childhood obesity would equate to an investment in population health, since half of these children will be obese in adolescence and 80% in adulthood, increasing the incidence of cardiovascular risk factors in adults. 14urrently, the largest study in carried out in Spain is that by De Bont et al., 6 while its analysis of trends from 2006 to 2010 is of great interest, its scope was limited to a single Spanish region (Catalonia).Other population-based studies at the national level have instead focused on the adult population. 15he aim of the present study was to estimate the prevalence of overweight and obesity in children in Spain as well as to study the possible sociodemographic and medical associations related to obesity.

Material and methods
This was a population-based cross-sectional observational study, conducted using data from the latest national health survey in minors in Spain (ENSME2017).Data were provided by the National Institute of Statistics (INE 2021). 16The survey used a three-stage sampling strategy, stratified by size of municipality.The population-based survey covered the whole territory of Spain and took place from October 2016 to October 2017 (INE  2017).
This study included all children surveyed who were aged 1 to 14 years.The exclusion criteria were not having data on dietary variables or other explanatory variables, weight or height, and extreme z-scores for the body mass index (BMI) (1st or 99th percentiles).Details of survey methodology are published online (https://www.ine.es/metodologia/t15/t153041917.pdf).
The BMI of the included children was calculated from their weight and height.The outcome variable was calculated from the BMI z-score, according to the Spanish pediatric population tables presented by Carrascosa et al. 17 Malnutrition was defined as a z-score of less than −2.00; normal weight, from −2.00 to 0.99; overweight, from 1.00 to 1.99; and obesity, 2.00 or higher.The analysis was performed by grouping 'obese' against 'non-obese'.
In addition, the analysis considered different explanatory variables from the survey.Sociodemographic characteristics were: age, sex, autonomous community of residence, size of the municipality of residence, children's country of birth, parents' occupational social class (Annex 1), relationship of the informant with the minor, educational level of the informant, and employment status of the informant.Behavioral variables comprised: levels of physical activity, exposure to tobacco smoke, daily hours of sleep, and tooth brushing habits.Health-related factors were: self-perceived health, chronic allergy, asthma, diabetes, epilepsy, behavioral disorder, mental disorders, injuries or accidents, autism spectrum disorder, and having limitations due to health problems in the previous six months.Health services data were also collected: time since the last visit to the pediatrician or specialist; visit to the psychologist, speech therapist or nurse in the last year; hospital admission, day hospital or visit to the emergency room in the last year; and use of radiography, CT, ultrasound, and MRI in the last year.Diet-related factors were: frequency of intake of fresh fruit, dairy products, vegetables, sweets, sugary soft drinks, fast food, snacks, and natural juices, as well as adherence to a particular diet.Household variables were: number of adults; number of minors; number of bedrooms; net monthly household income; odors; poor water quality; street cleanliness; pollution; green areas; and problems with external noise, animal disturbances, or crime or vandalism.

Statistical analysis
A descriptive analysis of all the data was performed by calculating frequencies for categorical variables, and minimum, maximum, mean and standard deviation values for the quantitative ones.
Factors associated with the presence of obesity were analyzed using contingency tables, applying Fisher's exact test for categorical variables, and the Mann-Whitney U test for comparing the means for quantitative variables.
The magnitude of associations between explanatory variables and the presence of obesity was estimated by fitting multivariable Poisson models with robust variance.Results were expressed as prevalence ratios (PR) and 95% confidence intervals (CIs).A stepwise variable selection procedure was performed based on the Akaike Information Criterion, taking into account the possible multicollinearity of the variables.Goodness-of-fit indicators were calculated, such as the likelihood ratio test (LRT) and a model calibration test. 18To obtain representative estimates of the Spanish population, we accounted for complex sampling features by using the elevation factor of the survey divided by its mean as a weighting factor, obtaining weights centered on its mean.Statistical analysis was performed using R software (v.4.0.2).

Ethics approval statement
This study is exempt from institutional review board review because ethical approval is not required in Spain for performing analysis with anonymized data provided by the National Statistics Institute.

Results
There were 6297 eligible households (with at least one minor); of these, 6016 minors were included (response rate of 97%).The reasons for no inclusion were: 133 due to refusal to participate, 56 due to absence and 2 due to disability. 16Of the 6106 children included in the ENSE2017 for minors, the following were excluded from the analysis: 207 children under 1 year of age, for not presenting values in the dietary variables; 733 children for missing values in weight or height; 180 children for missing data on some of the explanatory variables; and 104 children for presenting extreme values for the z-score (less than the 1st percentile, z-score < −4.3 or greater than the 99th percentile, z-score > 12.2).The final analyzed sample was 4882 children.
The mean age of the sample was 7.5 years (SD 4.1); 51.2% were male, 94.3% were born in Spain, and 7.3% lived in households with a net income of less than EUR 800 per month.Regarding health status, 6.1% reported fair, poor, or very poor self-perceived health, 12.2% had chronic allergy, 6.3% asthma, and 5.5 some physical limitation in the previous six months.In terms of the use of health services, 29.6% had visited the pediatrician in the previous month, and 8.3% had seen a specialist.In the year prior to the survey, 5.9% had visited a psychologist, 3.5% had been admitted to hospital, and 37.3% had visited the emergency room.Behavioral data showed sedentarism in 20.5% of respondents, while 6.3% never brushed their teeth or did so occasionally, and 6.4% had some daily exposure to tobacco smoke.Regarding diet, 5.7% consumed less than one serving of fresh fruit a week, 7.6% ate vegetables less than once a week, and 1.7% had dairy less than once a week.On the other hand, 44.2% ate sweets at least once a day, while 5.8% had sugary soft drinks, 2.8% fast food, and 1.8% salty snacks on a daily basis.About 30.1% never consumed natural juices, and 2.8% followed some diet.As far as household characteristics, 14.9% lived in dwellings with two bedrooms or less, 29.1% reported problems with external noise, and 28.8% poor access to green space (Table S1).
In 2017, 32.1% of children aged 1-14 years were above the norms for a healthy weight; prevalence of obesity was approximately 18.6% (95% CI 18.5-18.7)and overweight 13.5% (95% CI 13.4-13.6).These figures represent a total of 1,027,466 children with obesity and 744,940 children with overweight in Spain in 2017.
Table 1 shows the most noteworthy associations between explanatory variables and childhood obesity; results of the complete analysis are in Table S1.A higher prevalence of obesity was observed in younger versus older children (20.5% at 1-2 years compared to 11.3% at 11-14 years), in more disadvantaged versus more advantaged social classes (25.0% in class VI compared to 13.3% in class I), and in children from households with less versus more income (31.9%from households with less than EUR 800 net monthly income compared to 7.4% from households with more than EUR 3600 monthly income).Obesity was also more prevalent in children who had visited their pediatrician more recently (21.7% in the last month compared to 11.9% in more than a year) and in those who reported less frequent visits to a specialist (25.1% never compared to 17.9% in the last month) or a psychologist (19.1% with no visit in the previous year compared to 11.7% with a visit).Other factors associated with obesity were less physical activity (24.1% sedentary versus 15.4% with physical activity several times a week), more exposure to tobacco smoke (30.6% more than 1 h/day compared to 18.1% never or almost never), lower consumption of vegetables and greens (21.8% 1-2 servings/week compared to 18.0% 1 or more serving/day), greater consumption of sugary soft drinks (25.3% 1 or more a day vs. 15.1% 1-2 a week), lower consumption of natural juices (21.4% less than once a week vs. 14.5% more than once a day), and adherence to some diet (31.6% vs. 18.3%).
Fig. 1 shows a map of Spain along with the estimated prevalence of childhood obesity in each autonomous region.A north-south gradient is apparent, with higher prevalence in southern regions.Table 2 shows the childhood obesity prevalence ratios estimated using the multivariable Poisson model.A significantly higher adjusted prevalence was observed in the 3-5-year-olds (PR 2.05) compared to the 11-14 age group; in the Canary Islands (PR 3.81), Andalusia (PR 3.54), Murcia (PR 3.49) and Catalonia (PR 3.22) compared to Cantabria; and in sedentary children (PR 1.32) compared to those who took part in physical activity several times a week.Compared to children who never consumed sweets, the prevalence of obesity was about twice as high in children who had them regularly (1-2 times/week, PR 1.87; 3 times/week, PR 2.16; 4-6 times/week, PR 2.14; and ≥1 times/day, PR 2.03).Children with monthly household income of less than EUR 800 were also at higher risk compared to those with a monthly income over EUR 3600 (PR 3.63).Finally, obesity was associated with a lower frequency of visits to the pediatrician (PR 1.54), not having an X-ray in the last year (PR 1.45), not having chronic allergy (PR 1.49), and having asthma (PR 1.62).

Discussion
This work shows a high prevalence of overweight (13.5%) and obesity (18.6%) in children in Spain in 2017, representing nearly 750,000 children aged 1-14 years who were overweight, and more than 1 million who were obese.The variable with the largest association with childhood obesity was socioeconomic status.Prevalence of childhood obesity was 3.6 times higher in households with monthly incomes of less than EUR 800 compared to households with incomes more than EUR 3600.
International data on the prevalence of childhood obesity show a stabilization worldwide and even a decrease in countries with more  economic resources. 5Our data are in line with previous national health surveys, which have shown a modest decline in prevalence since the peak about 10 years ago. 19he international (International Obesity Task Force; IOTF) BMI cut-offs are widely used to estimate the prevalence of child obesity because they make it easier the comparison with other approaches. 20Previous studies about the prevalence of child obesity in Spanish regions used the BMI-Z score to define the obesity 6,21 since, due to ease of acquisition, the BMI is the most widely used clinical outcome variable.Our results are consistent with those reporting on the pediatric population in Catalonia, where the prevalence of obesity was 17.5% in 2016 and the combined prevalence of overweight and obesity, 39%. 6The authors of a recent study analyzed data from children aged 2 to 17 years in eight autonomous regions in Spain and found that the prevalence of overweight and obesity decreased from 2005 to 2017 in most regions. 21Significant geographical differences were observed, with a higher prevalence in southern regions.Further research is needed to know the reasons of these differences.A similar north-south gradient was already observed in a European study, where the southern countries had a higher prevalence, 22 although those authors pointed to the loss of the Mediterranean diet in favor of a westernized one along with a more sedentary lifestyle and lack of physical activity.Notably, southern European countries also generally have a lower economic level compared to northern ones.
The strong association between socioeconomic status and obesity is well known and is consistent with the results published elsewhere 7,23 Occupational social class is closely related to socioeconomic status, so likewise a greater prevalence of obesity was observed in lower occupational social classes, 24 although the magnitude of the association was greater for income level.We were not able to detect an association between parents' country of birth and childhood obesity, as in neighboring European countries. 12e consider that the migratory flows in Spain may be further in the past than elsewhere, allowing greater social integration.In our population, poor diet, increased consumption of sweets, and sedentary behaviors were associated with childhood obesity.Despite the present study reports a weak significant association between physical activity and obesity, probably due to the way of measuring physical activity, these factors are being tackled in multiple diet and exercise intervention programs that have been shown to be effective when used in combination. 24Notably, even moderate intake of sweets once or twice a week was associated with nearly double the risk of obesity, compared to never consuming sweets.These data confirm the findings of previous studies, 10 highlighting the importance of diet in overweight or obese children.More obesity was observed in children who used health services less frequently and who did not perform other health-promoting behaviors, such as tooth brushing.The association between cavities and childhood obesity has been described in other studies. 25The national health survey does not collect analytical data on vitamin D or hours of sleep, but in other studies obesity has been related to vitamin D deficiency, 26 just as healthy sleeping habits are linked to less obesity in children 27 As in other studies, a higher prevalence of obesity was observed in children with asthma. 28As in the adult population, cardiovascular risk factors tend to cluster in the same individuals, heightening a single person's risk of developing cardiovascular disease. 29Along the same lines, a large study in concluded that economic subsidies in more deprived areas, taxes on unhealthy foods, and policies promoting physical activity would be the most effective measures to reduce childhood obesity. 30mong the limitations of our study, its cross-sectional design precludes any conclusions about the causal relationship between obesity and associated factors.In addition, other factors associated with obesity may not have been measured in the survey.The selfreported nature of the survey responses could also introduce a risk of recall bias.On the other hand, the definition of obesity using the BMI-Z score may have limitations in children under 2 years of age and may not reflect current conventions.In contrast, the large sample size, and population-based estimates, considering the elevation factor of the survey in a complex sample design, are the major strengths.
This study provides updated, representative data on the pediatric population in Spain in 2017 and describes its most relevant associations.According to these data, as in neighboring countries, in 2017 the prevalence of obesity showed signs of beginning to stabilize, but not evenly throughout the national territory.A National Health Survey in Spain is currently underway and will be published in 2025.This data will allow for analysis and comparison with our current research.
We conclude that the combined prevalence of overweight and obesity in 2017 exceeded 32% of the pediatric population in Spain, representing more than 1.7 million children.Low socioeconomic status and other factors such as poor diet or sedentarism were the main factors associated with childhood obesity.Although the cross-sectional design of this study prevents establishing causality, the study findings allows set very cautious hypotheses about the observed associations to be explored in future studies.Prevalence estimates show geographical disparities, with the southern regions standing out as the most affected.Further research with more specific study objectives is needed to clarify aspects such as the socioeconomic factor or the differences of prevalence between regions.

Strengths and limitations of this study
• The major strengths are large sample size and population-based estimates, taking into account the elevation factor of the survey in a complex sample design.
• The cross-sectional design precludes any conclusions about the causal relationship between obesity and associated factors.
• Other factors associated with obesity may not have been measured in the survey.
• The self-reported nature of the survey responses could also introduce a risk of recall bias.
Francisco Sanchez, Ana Pilar Nso and Jose Antonio Quesada drafted the manuscript.Vicente Bertomeu, Juan Miguel Ruiz and Adriana Lopez revised this manuscript critically.All authors approved the final version of this manuscript.

Ethical considerations
This study is exempt from institutional review board review because ethical approval is not required in Spain for performing analysis with anonymized data provided by the National Statistics Institute.

Fig. 1 .
Fig. 1.Prevalence of obesity in children aged 1-14 years in Spain in 2017, by autonomous region.

Table 1
Prevalence of obesity according to explanatory variables.Occupational social class: I. managers of ≥10 employees/II.Managers < 10 employees/III.Intermediate occupations and self-employed/IV.Skilled technical worker/V.Skilled worker in the primary sector/VI.Unskilled worker/NA: not available.
a Non-obese group: underweight, normal weight, overweight.b

Table 2
Prevalence ratios of childhood obesity, multivariable Poisson model with robust variance.