Participants
The participants of this study were preschool children aged 2-6 in Shenyang, China. Multi-stage sampling was used to select the sample. From the seven districts in Shenyang, four were randomly selected. From each of the four districts, two kindergartens were randomly selected. Convenience sampling was employed to select one class from each grade in each kindergarten, yielding a total of 635 children. Physical examinations were performed on these children, and their parents were asked to complete the questionnaires. Children were excluded from the study if: (1) Their parents’ ethnicities were not Chinese; (2) Their age was not between 2-6; (3) They had developmental diseases; (4) Their parents refused to give consent; (5) Their parents could not complete the questionnaires. Finally, 605 children and their parents participated in this study.
This study was approved by the Ethics Committee of Institute of Health Science (Ethics Approval No. [2017] 055), China Medical University.
Measurements
Dependent variable
The dependent variable of the first aim was BP. The physical examinations were conducted by trained and licensed doctors during school time. The children rested for at least 5 minutes before their blood pressures were taken. Blood pressures were measured by using an auscultation mercury sphygmomanometer with an appropriate cuff size. The systolic pressure was recorded as the pressure when the sphygmomanometer made the first Korotkoff sound. The diastolic pressure was taken when the sphygmomanometer made the fifth Korotkoff sound. BP was measured twice for each child and the average was taken. Based on the gender- and age-specific distribution of Chinese children’s blood pressures [28], elevated BP was defined as having either systolic or diastolic pressure above the 90th percentile, and normal BP was defined as both systolic and diastolic pressures below the 90th percentile. A discrete variable was created with ‘normal BP’ being the reference group (0 = normal BP and 1 = elevated BP).
The dependent variables of the second aim were obesity indexes. Heights and weights were measured with the children standing up, barefoot and wearing under garments. BMI was calculated as weight (kg) divided by the square of height (m). According to the gender- and age-specific BMI cutoff points provided by the Obesity Working Group, China (OWGC) [29], subjects were classified into one of three categories: healthy weight, overweight or obesity. We created a discrete variable “BMI Category” (0= healthy weights, 1= overweight or obesity).
WC was measured by wrapping a nonelastic flexible measuring tape around children’s waists, 1 cm above the navel. The measurements were taken at normal expiration. Children whose WC was above the gender- and age-specific 80th percentile provided by OWGC were determined to be abdominally obese [30]. We created a discrete variable “WC Category” (0= not abdominally obese, 1= abdominally obese).
WHtR was calculated as waist circumference (cm) divided by height (cm). Children whose WHtR was greater or equal to 0.5 were determined to be abdominally obese. The variable “WHtR Category” was defined as a discrete variable, which was 0 if WHtR was less than 0.5, and 1 if WHtR was greater or equal to 0.5.
Because males and females have different distributions of BMI, WC and WHtR values, we standardized these three obesity indices. Based on the gender- and age-specific averages and standard deviations of BMI, WC and WHtR provided by OWGC, BMI, WC and WHtR values were standardized into standardized z-score values ZBMI, ZWC, and ZWHtR, respectively [29,30]. Standardized z-score values for BMI, WC and WHtR were all calculated as:
Questionnaires and independent variables
A self-designed questionnaire was used, and the parents were asked to complete the questionnaire. The questionnaire asked about children’s and parents’ demographic information, children’s food preferences, children’s eating behaviours, parental control of children’s diets, children’s sleep duration and children’s levels of physical activity.
Demographic information included: the child’s age and gender, parents’ heights, weights, levels of education and familial income. Paternal BMI and maternal BMI were calculated. Familial income level was recorded as < 3000 yuan, 3000 to 5000, 5000 to 8000, >8000. Parents’ education was categorized into three levels: secondary school or less, college or bachelor’s degree and graduate degree. The child’s age, paternal BMI and maternal BMI were used as continuous variables, and familial income level and parents’ education were used as dummy variables in the regression models.
The familial dietary information included the child’s food preferences, eating behaviours and parental control and guidance on the children’s diets. The child’s food preferences: parents were asked whether their child normally eat wheat and wheat-based products as their staple food (0=no, 1=yes). They also answer how many times a week their child eats deep-fried food, potatoes and other root vegetables, milk products, desserts and sweets, nuts and puffed food. The child’s eating behaviours included: eating speed (1=slow, 2=average, 3=fast), whether they watch TV while eating (0=yes, 1=no), whether they snack while watching TV (0=yes, 1=no) and appetite (1=good, 2=average, 3=poor). Parental control and guidance on the children’s diets included: whether they limit the amounts of snacks their child eats (0=yes, 1=no), whether they use food as rewards (0=yes, 1=no), whether they encourage their child to eat a diverse assortment of foods (0=yes, 1=no).
Sleep duration is the amount of sleep the child gets during the night combined with the amount of sleep the child gets during the day (in hours). The amount of physical activity the child gets is the total amount of physical activity in a day (in hours).
Statistical analysis
Numerical variables in normal distribution were expressed by mean and standard deviation, and numerical data of skew distribution was expressed by percentile. Categorical variable was showed by frequency and percentage. The difference of ZWC between elevated and normal BP group was analyzed by two samples t test, and the difference of ZBMI and ZWHtR between these two groups was analyzed by Wilcoxon Mann-Whitney test. The difference of prevalence of overweight and abdominal obesity between elevated and normal BP group was analyzed by chi-squared test. Logistic regression was used to analyze the association between obesity indices (ZBMI, ZWC, ZWHtR, BMI Category, WC Category, WHtR Category) and elevated BP in preschool children. Elevated BP was the outcome variable in the logistic regression, and five models were constructed. Each model included two different obesity indices which were both numerical or categotical to compare the sensitivity of the indices to elevated BP. All models were adjusted by age and amount of physical activity.
In order to investigate the relationship between familial factors and the obesity indices, ZWC and WC category were used as dependent variables in the linear regression and logistic regression, respectively. First, each independent variable was entered in the univariate regression and variables that yielded a P value less than 0.2 were remained. Second, a multivariate stepwise linear regression and a multivariate stepwise logistic regression containing independent variables remained from the previous step were then performed to find the independent factors for ZWC and WC Category, respectively.
Excel 17.0 was used to record the collected data. SPSS 22.0 was used for statistical analysis.
Table 1 Basic characteristics of participating children.
*significant at P<0.05.
SD, standard deviation; BMI, body-mass index; WC, waist circumference; WHtR, wasit-to-height ratio; DPB, diastolic blood pressure; SBP, systolic blood pressure; P, percentile; BP, blood pressure; ZBMI, standardized z-score values of BMI; ZWC, standardized z-score values of WC; ZWHtR, standardized z-score values of WHtR.