Introduction

Obesity, an escalating global epidemic defined as a disease by the World Health Organization (WHO), affecting virtually both developed and developing countries of all socioeconomic status (SES) groups, including all age groups, is posing an alarming problem [1, 23]. China once with the leanest population of Asia is now rapidly catching up with the West in terms of prevalence of overweight and obesity which had occurred in a remarkably short time with an estimate of 215 million overweight and obese youngsters (2002 China health and nutrition survey) out of which 7% of those were aged 7–17 years [44]. In China, the prevalence of childhood obesity in different regions is determined by the SES, diet, and lifestyle. It varies from 3.9% to 17.6% in girls and 5.0% to 32.5% in boys of the western rural region and northern coastal cities of China, respectively [19]. A study from mainland China reported that one in eight children among urban Chinese was overweight in 1997. Based on recent trends, this figure is likely to be one in five urban children and one in 14 rural children being affected by 2010 [20, 42].

The rising prevalence of childhood obesity is also observed in Japan, Malaysia, and South Asia [5, 9, 40]. As a person’s body mass index (BMI) increases through the overweight and in to the obese category and beyond, the risk of developing a number of chronic noncommunicable diseases, metabolic disorders, and psycho-social problems starts to rise rapidly also in childhood [1].

A large number of studies have tried to identify the risk factors for the epidemic of overweight and obesity. It is generally agreed that overweight and obesity are a multi-factorial disease and its development is the result of multiple interactions between genetic and environmental factors [2, 26, 28, 45]. Factors contributing to the rising childhood obesity epidemic include higher SES [9, 31, 37, 41], urban residence [6, 46], dieting and place of service [5, 15, 24, 32], decreased level of physical activity [1, 5, 15, 27, 39], prolonged hours of watching TV, playing video games [3, 5, 19, 21, 25, 34, 45], high birth weight [18, 35, 47], and a mother’s history of gestational diabetes mellitus (GDM) [10, 33]. However, in contrast to the genetic determinants, finding relations to the lifestyle and other factors have shown to be inconclusive. Differences in methodology, as well as differences in cultural and social backgrounds of the study population, are likely to contribute to these inconsistencies. Better understanding of the factors associated with obesity would help policy makers, healthcare professionals, and the general public combat the epidemic.

To our knowledge, there is very little data on the prevalence of overweight, obesity, and associated risk factors among school children and adolescents in Tianjin. The aim of the present study was to assess the prevalence of overweight, obesity and associated risk factors among school children and adolescents in urban and sub-urban areas of Tianjin, a City in Northeast China.

Subjects and methods

Overall 3,205 students were screened for anthropometric measurements from May to August 2010. Initially, we selected schools using multistage random cluster sampling. Because of diversity in socio-demographic status, schools were stratified according to location and SES of their enrollment area. Students between 7 and 18 years from first to ninth grade participated in this study. Students who were under 6 and over 18 years of age, as well as those with overweight/obesity due to known metabolic and endocrine diseases were excluded. The research team visited the chosen schools to inform them about the survey and interviewed students to explain how to deal with the questionnaire. The questionnaire had two sections; the first section which included basic personal information in addition to anthropometric measurements, was filled out in the classroom. Students were asked to take the second section of the questionnaire, concerned the SES of family, birth weight, and parental information, to their home to be filled out by one of their parents or guardians. To gain information about those students’ lifestyle, a frequency questionnaire on their food and physical activity was used.

Body weight and height were measured using a standardized digital scale (to the nearest 0.5 kg) and wall-mounted stadiometer (to the nearest 0.5 cm) respectively. Student’s weight and height were taken while wearing school uniform, without shoes. Waist and hip circumferences of each student were also measured (to the nearest 0.1 cm). BMI—the weight in kilograms divided by the square of height in meters—was calculated. Overweight and obesity were defined using age- and sex-specific BMI cut-off points developed by the Working Group for Obesity in China (2004), using the 85th and 95th percentile to define overweight and obesity, respectively [14, 43]. Since overweight and obesity are the main criteria for identifying children at risk, the two categories were combined as overweight to compare with normal weight in this study.

Ethical consideration

This research was approved by the ethics committee of the Tianjin Medical University. Written informed consent was sought out from each participating institution and participant before data collection.

Statistical analysis

Data input was performed using Epi-data 3.1 software, and was exported to SPSS statistical package version 16 for windows (SPSS Inc., Chicago, IL, USA) for analysis. Quantitative indices are expressed as means ± standard deviations using t test.

The relationship between each factor of interest and the children’s weight status (overweight and normal weight) was explored by percent (%) or χ 2 tests (for categorical variables). Univariate logistic regression analysis was performed to evaluate the strength of the relationship between childhood overweight and each of those variables of interest. Only variables with significant associations (i.e. P value < 0.05) with children’s weight status in theχ 2 tests were considered in the logistic regressions. Odds ratios (OR) and 95% confidence intervals (95% CIs) were calculated for each factor. In the multivariate analysis, adjustment was made for parental obesity to control for possible familial influences. Two-tailed P value < 0.05 was considered statistically significant.

Results

From the total 3,205 students, 65 questionnaires were cleared; 32 were above or below the age range and 33 did not respond properly. 3,140 fully responded, i.e., a response rate of 99% (Table 1). Approximately half (49.7%) of the participants were male (Table 1). Tests have shown a statistically significant association between BMI and gender (Table 2). There are more overweight males compared to females as shown in (Table 3). Univariate and multivariate logistic regression showed male gender was associated with overweight (Table 4). The association remained significant (OR = 1.89, 95% CI = 1.61–2.21) and (OR = 1.84, 95% CI = 1.49–2.27), respectively, when adjusted for parental obesity.

Table 1 Distribution of male and female according to BMI
Full size table
Table 2 Means and standard deviation of anthropometric measurements for males and females
Full size table
Table 3 Relationship between overweight and independent factors
Full size table
Table 4 Adjusted odds ratios (95%CIs) of being overweight by parent’s BMI and selected factors
Full size table

Prevalence of overweight was statistically significantly different among urban and rural residents. Children from an urban area showed a significantly higher proportion of being overweight (p < 0.001) (Table 3). OR of multivariate logistic regression analysis adjusted with parental obesity showed as double as univariate 1.31 (1.11–1.54) and 2.68 (2.16–3.32; Table 4), with rural residents as reference.

There was a significant positive association between child’s birth weight, mother’s history of GDM and their weight status among 7–18-year-olds (p < 0.01; Table 3). OR of univariate and multivariate logistic regression analysis showed that high birth weight (>4 kg) was 2.25 (1.46–3.46) and 2.49 (1.58–3.90) times more likely to be overweight compared with children birth weight of (2.5–4.0 kg) as a reference (Table 4).

The relation between BMI and physical activity was investigated based on the parameters; the kind of transportation to school and the length of outside play. Of the factors related to the lifestyles, decreased physical activity (<2 h/day) had a significant association with overweight children. Univariate and multivariate regression analysis showed that children who did physical activity (<2 h/day) were 25% more likely to be overweight than their counterpart (Table 4). Similarly, those who use motorized transportation to and from school were more overweight than those traveling on foot or by bike. The difference remained significant (p < 0.001; Table 3). Logistic regression analysis showed those children who travel on foot or by bike were 54% and 62% less likely to be overweight than those who travel by motorized transport (Table 4). Children’s weight status is not associated with the time spent on watching TV/ playing video games for (>2 h/day; Table 3).

Children who participate in an outdoor activity are significantly more protected from being overweight. The logistic regression showed those who participated in an outdoor activity were 63% less likely to be overweight in comparison to children who did not take part in an outdoor activity (Tables 3 and 4).

Table 3 shows a significant association between parental educational level and BMI, (p < 0.001); a direct proportion was detected in the OR of mother’s education level and children’s weight status. Maternal educational level of secondary or junior (OR = 1.67, CI = 1.23–2.26, and OR = 2.79, CI = 1.96–3.96) and tertiary and above (OR = 1.66, CI = 1.20–2.29 and OR = 2.97, CI = 2.03–4.36) in univariate and multivariate logistic regression analysis respectively. But only paternal educational level of tertiary and above has significant difference (OR = 2.27, CI = 1.53–3.35 and OR = 2.48, CI = 1.63–3.78) within the two weight groups compared with the primary or below parental educational level. No significant association was detected between household income and BMI of children (Table 3).

Subjects consumed homemade lunch (48%); 45.1% and 6.9% consumed lunch from school dining and fast-food/restaurants, respectively. Places of service as well as the kind of food had a significant correlation with BMI. Students who ate sweet (desserts) foods more frequently have higher BMI compared with those who used to eat sweet food on a less frequent basis (result not shown).

Discussion

Using the definitions for overweight and obesity by the Working Group for Obesity in China [14], prevalence of overweight and obesity among school children and adolescents was found to be 12.5% and 15.7% respectively. Compared to a recent study from Xi’an (China), prevalence of obesity was higher in Tianjin [45]. Parental obesity was independently associated with childhood overweight. Previous studies have given a strong evidence for parent–child correlation of weight status [18, 45]. Obesity often tracks in families. Having obese relatives increases one’s risk for obesity, even if family members do not live together or shares the same patterns of exercise and food intake [2, 26, 40, 44]. SES usually presented as a composite index combining income, parental educational level, occupation and in some developing countries—place of residence (urban/rural) has mixed findings similar to another study [36]. The association reflects both genetic and environmental influences for the development of overweight. As it has been said that a child’s genetic make-up “loads the gun” while their environment “pulls the trigger” [4], several studies concluded that besides the high genetic similarity among family members [26], parents play an important role in the development of children’s physical activity patterns [29], eating behavior, and attitudes [8] .

High birth weight and a mother’s history of GDM were significantly associated with an increase in BMI levels of children in this study (Tables 3 and 4). Similar studies done in, China Delhi India, and Hong Kong, reported that the percentage of overweight/obese children increases with birth weight [18, 35, 47]. Having been born from a mother with GDM was also associated with increased childhood and adolescent overweight [10, 33]. Targeting and preventing excessive weight gain in children positive for parental obesity, high birth weight, maternal history of GDM, and women of childbearing age as well as antenatal and postnatal care would therefore appear to be a more practical and acceptable approach to explore. In order to succeed, it needs whole family intervention. In this study, we found male gender and children aged 7–11 years have significantly higher prevalence of being overweight compared to female and adolescents aged 12–18 years (Tables 3 and 4), which is similar to the previous studies [6, 20, 45].

The relation between the prevalence of obesity and SES has been reviewed previously in China and Brazil where Brazilian adolescents in a high-income group were two to three times more liable to be obese than their lower-income counterparts [31]. Studies from developed countries conclude that the prevalence of overweight/obesity is negatively correlated with SES [37]. For example, in the United States (US) and Norway, overweight was higher in rural areas, in poor families, and in children with mothers having a low educational level [11, 13, 22, 37, 38, 41]. In contrast, a study in developing countries like China, unlike to a recent study from Xi’an [45], showed overweight was more common in children of mothers with high educational achievement [30]. This is similar to our findings. Parent’s educational level may be associated with more sedentary transportation, less housework, and less active work, which in turn is associated with higher risk for developing childhood obesity.

Food intake and physical activity are the two primary factors which determine body weight. Based on the present cross-sectional observation, overweight children consumed significantly more sweet foods and often use food services outside of the home, either fast-foods, restaurants or school cafeterias rather than their counterparts with normal weight (Table 3), a result which is consistent with previous studies in China [26]. Evidence indicates a positive association of outside home food service and high portion of food consumption at each time [27, 32]. Parents and children should develop the habit of cooking and serving food at home with their preferred taste and ingredients. Previous studies have focused on the impact of decreased consumption of fresh fruits and vegetables and skipping breakfast on weight of children [1, 15, 24]. However, an association of these dietary components was not found in this overall study, possibly as a result of decreased quantification on the children and parents reports. In the present study, urban residents in Tianjin are consuming increasingly higher levels of fat, pork, and sweet products which are detrimental to their health, similar to another study [6]. The possible explanation is that those living urban area are more likely to be “westernized” and therefore may experiences a higher risk of being overweight.

Lifestyles have changed in China during the last two and a half decades in response to rapid economic development, improved food supplies, expansion of television, computerization, increased car ownership, and improved public transport. Physical activity in children has declined; decreased physical activity is one of unwary for its short- and long-term health impacts. Several studies have concluded a positive association between increased BMI and a decreased physical activity [1, 3, 7, 15, 21, 24, 34, 46], which is similar with our findings, in which there is a significant association between BMI and <2 h of physical activity per day, but not associated with ≥2 h of watching TV/playing video games per day, (Tables 3 and 4) different from previous studies [3, 21, 34]. Children who participated in outdoor activities either for a part time job or family support were 63% less likely to be overweight in this study (Table 4).

Commuting mode to school is likely to be influenced by a number of factors including; distance to school, cost, SES of the family, and availability of motorized transport (either household owned or public). Walking has been shown to be beneficial to health and weight control [12, 17] while motorized vehicle use seems to be associated with overweight and other disorders. Among young adults in the US the proportion of individuals using active transportation is higher among non-overweight compared to overweight ones (to work, 9.2% vs. 6.8%; to school, 29.7% vs. 22.6%) [12]. Among Chinese adults, going to and from work by walking or by bicycle seems to reduce the risk of being overweight by half compared to going by bus [16]. In this study, we found a significant association between overweight and motorized vehicle transport. The odds of overweight among children who transport by walking and bicycling are less likely by 54% and 62%, respectively (Table 4). Easy access to transportation which has led to a reduction in the energy expenditure along with an increase in energy intake due to an increase of purchasing power and availability of high fat; energy-dense foods are affecting the physical activity practice of school children and adolescents in Tianjin. In the advanced metropolises of China such as Tianjin, Beijing, and Shanghai, children’s lifestyle in these municipalities are more like the developed countries [19, 20]. This high prevalence of obesity in Tianjin is probably due to lifestyle differences from other cities like Xi’an [45]. Public health interventions at the individual and policy-making levels need to be instigated at the earliest to tackle this problem in the country. The utility and practicality of such an approach should be carefully evaluated.

Limitations

There may be reporting bias in some values of the self reported questionnaire, additionally; BMI fails to distinguish between fat and fat-free mass (muscle and bone). Regarding the ethnicity, we could not examine the ethnic subculture variation in this area, as only a very small proportion of the subjects are from minority ethnic groups, the sample size did not allow us to conduct meaningful comparisons.