A cross-sectional survey was collected between May and June 2013 at three sites in Nanchang, Jiangxi, and China. Sites included two vaccination clinics run by the China Centers for Disease Control and a children’s hospital. Members of the investigating team had existing relationships with the three sites, which serve the majority of the Nanchang population who live in two of the nine major districts of the city. Because this was an exploratory study, we were unable to calculate power a priori. We aimed to collect at least 400 completed surveys to have adequate numbers of both genders and overweight and obese children. The Institutional Review Boards of the University of Hawaii at Manoa and Nanchang University approved the study.

Nanchang residents aged 18 years or older who were caregivers of a child between the ages of 2 and 10 years old were eligible. Interviewers underwent an extensive training session on survey data collection methodology to ensure rigorous adherence to interview protocols. Interviewers approached every possible participant who had a child appearing to be older than 2 years. Those who qualified were asked if they were willing to participate. If they consented, the face-to-face survey was administered.

A modified version of the Healthy Homes Survey was used as the main study instrument (21). This survey was developed in the United States and has good reliability and validity. The main modification to the survey was to substitute food items found in the American home with a list of the most common foods reported in Chinese households from the Shanghai Women’s Health Study (22). This list of food items has been previously used in China and has been shown to be valid and reliable (22). Items were translated and back translated by the research teams in China and the United States. Minor wording changes were included to make the questions clear and understandable in Mandarin. A 12-part questionnaire, which included 146 items that focused on home environment, comprised the survey. Each survey took on average 15–25 min to complete through face-to-face interviews. Surveys were extensively pre-tested with Nanchang caregivers prior to launching the study to ensure comprehension and face validity in the local population.

For this paper, data from four main sections were selected. These included: demographics, physical activity, sedentary behavior, and nutrition-related variables. Demographic variables included the respondent’s age, sex, height, weight, household income, food insecurity, and type of residence, along with the child’s general health status, age, sex, height, and weight. Food insecurity was assessed on a four-point scale by how often the family did not have enough to eat. Physical activity support was indicated as the sum of the presence of nine common sports related items in the home. A 6-item scale (α = 0.77) assessed participants’ encouragement of their child’s physical activity. The child’s participation in organized sports and the adult’s attendance at a gym or exercise classes was assessed. Variables related to sedentary behavior included having a television in the child’s room; child’s ownership of a handheld gaming device or video game system, time spent watching television or playing video games, and a 3-item scale on parental rules restricting the use of television and video games (α = 0.89). Fruit availability was operationalized as an index of how often eight common fruits were available in the household, and the vegetable availability consisted of how often sixteen common vegetables were available in the household. The availability of sugared sodas, diet sodas, other sugar-sweetened beverages, and fruit juice was also assessed. All availability quests were asked on a 5-point scale from 0 = never to 4 = always. Several eating habit questions were also included in this survey study, including how often the family eats dinner together (# of days per week, 0–7), how often the child eats breakfast, how often dinner includes freshly prepared foods, how often dinner included processed foods, and how often sodas or other sweetened beverages were served at meals and snacks (response options 0 = almost never to 3 = almost always). Parental food rules consisted of 8 items measuring the strength of household regulations and restrictions on the child’s eating behavior (α = 0.70). Parental behavior, including how often parents’ watch television, use computers at home, play video games, and play active video games, was also assessed (response options measured as time per day; 0–4 h a day).

The children’s height and weight were collected by self-report from the caregivers. A subset of participants’ height and weight were validated by objective measures collected at the CDC sites. Body mass index (BMI) was calculated using the height and weight reported on the survey. The CDC BMI growth chart was used to determine age- and sex-specific BMI percentiles (BMI%) used to categorize children as underweight (BMI% < 5th percentile), normal weight (5th ≤ BMI% < 85th), overweight (85th ≤ BMI% < 95th), or obese (BMI% ≥ 95th) (23).

The prevalence of obesity was calculated, stratified by age, and sex. Descriptive statistics were calculated for the children’s demographic characteristics, physical activity, sedentary behaviors, and eating habits. Bivariate analyses (t-tests and chi-squares) were used to assess the relationship between obesity and potential risk factors. Logistic regression models with obesity as the dependent variable (obese = 1, not obese = 0) were then run with all variables that had a p-Value less than.10 in the bivariate analyses. Significance was set at p < 0.05 for all analyses, which were conducted using SPSS 20.0 (IBM, New York, NY, USA).

In total, 480 surveys were completed (response rate = 87.9%), and 470 surveys with complete height and weight data were retained for analysis. More than half (61.1%) of the respondents were the child’s mothers. Slightly more than half of the children (54.7%) were boys. About half (58.7%) of caregivers believed their child’s health was excellent or very good. A minority (38.7%) of families’ household income was higher than 60,000 RMB (approximately USD $10,000) per year. Most caregivers (64.0%) worked full time outside the home, and most families (87.4%) lived in an apartment or condominium.

About one in five (21.7%) of the children had BMI scores in the obese range. Obesity was more prevalent in boys (24.9%) than in girls (17.8%), but this difference was not statistically significant (p = 0.064). Obesity prevalence did not differ by either the caretaker’s age or the child’s age. Children from low-income households (26.2%; less than ¥30,000 annually) were more likely to be obese than middle (18.0% for ¥30,000–60,000) and high-income households (19.7% for more than ¥60,000). Obese children were also more likely to come from households with higher levels of food insecurity. General health status was not related to obesity. Demographics by children’s obesity status are presented in Table 1.

Several supportive and risk factors were examined to see if they were related to obesity in the children. Non-obese children (M = 4.7) had significantly more physical activity supports (e.g., bicycle and jump rope) than obese children (M = 4.1, p < 0.05). There was no significant difference in obesity prevalence between children participating in organized sports and children who did not. Participants’ encouragement of their child’s physical activity differed significantly by weight status, with caregivers of obese children providing less encouragement (M = 20.8, SD = 3.8) than parents with non-obese children (M = 21.6, SD = 3.5, p < 0.05). However, caretakers of non-obese children did not report going to the gym or exercise classes more often than parents of obese children. Non-obese children (M = 4.05, SD = 1.87) had more parental restrictions on use of television and video games than obese children (M = 3.51, SD = 1.84, p < 0.05).

Analyses showed no significant differences in the prevalence of obesity by presence of a television in the child’s room or ownership of a handheld gaming device. Non-obese children (42.1%) were more likely to have a video game station or computer than obese children (28.4%, p < 0.05). Table 2 displays physical activity supports and sedentary risk factors by obesity status.

Nutritional variables were then examined for their possible association with obesity status. Some variables showed little/no variability and were removed prior to data analyses. This included the child eating breakfast (97.4% often/always), families eating fruits and vegetables with their main meal (87.9% often/always), and the family freshly preparing food for the main meal (100% often/always). No significant differences by obesity status were seen for availability of fruits/vegetables, sugary soft drinks, diet sodas, or fruit juice in the home. Rules around food consumption did not differ by child’s obesity status. Table 3 displays the nutrition variables by obesity status.

Parents with obese children watched more television (M = 4.09 SD = 1.42 vs. M = 3.70, SD = 1.5, p < 0.05) but spent less time working on the computer (M = 2.82, SD = 1.75 vs. 3.38 SD = 1.71, p < 0.05) than parents of non-obese children. The time parents spent playing video games or active video games was not significantly associated with their child’s weight status.

The eight variables in the bivariate analysis that had p-Values <0.10 were entered into a logistic regression with the child’s obesity status as the dependent variable. The overall regression was significant [χ²(8) = 32.7, p < 0.001], and three variables were found to be significant (p < 0.05). Protective factors included the number of days the family eats meals together (odds ratio = 0.84, 95% CI 0.73–0.96) and parental home computer use time (odds ratio = 0.83, 95% CI 0.72–0.96). Parental television time (odds ratio = 1.25 95% CI 1.07–1.47) was related to higher levels of obesity. Table 4 displays the results of the logistic regression.