The present study was an attempt to jointly evaluate the relationships between three body fat measures with a set of covariates in Iranian mid-adolescents within different 4 districts, using a multivariate multilevel analysis. Given the multifactorial nature of childhood obesity which form a hierarchical structure, we analyzed the data through a multilevel model. One of the main finding of this study is the high positive correlations between TST, AST and SST at the child level, suggesting that children with higher TST tend to also have higher AST and SST after adjusting for a set of covariates at the child and district levels.
The prevalence of childhood obesity has sharply increased from 1990 to 2010 in low- and middle-income countries compared to the developed countries (32), which can have undesirable effects on physical, mental, and psychosocial health in adolescents (33-35). Studies reported that, the prevalence of overweight and obesity in adolescents varies in different parts of Iran (4, 19, 36). People, who were living in the same region with the same habits were similar in terms of growth, development, and body shape, which might be due to their lifestyle, dietary patterns, and socio-cultural factors (19, 20). The results of Table 2 revealed that, there were statistically significant differences between the anthropometric indices with respect to 4 districts. Therefore, the effect of individual level risk factors may vary according to the environment in which one lives.
To the best of our knowledge, limited studies have examined the association between individual factors and adiposity indices across children through multivariate multilevel analysis (20, 24). Results of multivariate multilevel approach showed that, some risk factors associated with the obesity in adolescents were consistent with those reported in previous researches in Iran (19, 20, 37). Results of multivariate multilevel analysis indicated a statistically significant association between the sex, family history of obesity, and SES with three adiposity indices. Sex was positively and highly associated with three outcomes, proving that girls had higher TST, AST, and SST than the boys. However, boys had better growth in terms of height, weight and subsequently in BMI than the girls. These results were in line with the previous studies which reported that, the percentage of subcutaneous adipose tissue was higher in females̓ bodies than that of males due to their sedentary lifestyle, less involvement in vigorous physical activities and less expenditure of energy (7, 16). Although, an agreement has been proved between BMI and TST in some studies (29, 38), BMI may not be a useful parameter in measuring the subcutaneous body fat of children, because changing the body shape occurs in childhood. Furthermore, it fails to differentiate the fat from the muscle mass and may classify children with large muscle into obese children group (18). Shriraam et al explained that, BMI is a crude measure, which does not provide a precise assessment of body density (10).
A positive association was found between family history of obesity and adiposity indices similar to other studies (20, 39). Khashayar et al. reported that, the odds of obesity in Iranian students with obese parents were about 2 times greater than the others (19). Environmental factors such as family lifestyle, eating habits and also becoming obese due to the genetic factors are considered as the subset of family history of obesity, and are the most important reasons influencing the persistence of obesity in adulthood (4, 40, 41). Therefore, modification of diet, having proper physical activities, and health care in the families could be an effective approach to decrease the risk of childhood and adulthood obesity.
In line with previous studies in Iran (19, 20), Our findings showed positive relationships between SES with three anthropometric measures, especially at high levels, which revealed that higher risk of overweight/obesity is related to the social environment. Bahreynian et al study reported that the prevalence of overweight was greater in areas with high SES, whereas underweight and short stature were more prevalent in areas with low SES (42). In the current study, students with higher anthropometric measures were living in families with higher SES, as confirmed in some other studies conducted in Iran and some other countries, in which positive significant associations were found between SES and adiposity among children and adolescents in developing countries (20, 24, 43). It is noteworthy that, the means of body fat, height, weight, and prevalence of overweight and obesity were lower in the students living in district 3 than other children (Table 2). Only 1.5% of families living in this district had a relatively high SES level and about 77% of them were classified as families with low income, educational and occupational levels. These findings highlight the need for planning to increase the level of awareness in the families in order to improve their lifestyle, nutrition and try to have more physical activities.
Several studies have reported time spent in watching TV or playing video games increased the risk of overweight/obesity in children (20, 24, 44). Moreover, the results obtained in some studies revealed a negative correlation between inactivity/sedentary behavior and physical activities in children and adolescents (25, 45). In our study, however, there was no statistically significant association between screen time and mild physical activities with adiposity indices. The results of Table 2 revealed that, the subjects living in districts one and four were more likely to be at risk of obesity with respect to body fat measures and BMI indices than other groups. Adolescents living in these two districts had more physical activities and also spent more time in watching TV or playing computer games compared to other two groups (Table 1). Watching TV and other sedentary behaviors increases the consumption of the most advertised goods, including sweetened cereals, sweets, salty snacks, and sweetened beverages leading to increased appetite, energy intake, thus affecting the body weight in children (46). Therefore, it seems that the presence of one behavior may be so strong that it cannot compensate for the presence of the other.
One of the strengths of the study was concerned with the results obtained in the random effects section in Table 3. The outcome variables were correlated at the districts and the subject levels, confirming the appropriateness of classifying the individual and district in the second and third hierarchical levels. The major portion of the total variance in TST (97.1%), AST (97.7%), and SST (97.5%) was found at the child level, meaning that children with higher TST tend to have high AST and SST. Results also highlighted the importance of clustering in assessing the relationships between demographic characteristics and anthropometric indices.
The cross-sectional nature of the study could be considered as a limitation in this study, because, it is not clear how response variables are influenced by the covariates. Further studies could take a prospective and time-based approach to obtain more accurate results. The lack of other predictor variables related to adolescent obesity such as eating habits, biological measures, as well as the selection of the district as the only variable in the third hierarchical level were also regarded as the second limitation of the study.