Study design and sampling
The children were recruited through random cluster design (all children from schools of the selected region Aalter) in the framework of the longitudinal Belgian ChiBS study (Children’s Body composition and Stress with measurements in 2010, 2011 and 2012) [22]. In this paper, only data from the cross-sectional survey conducted in 2012 was used, as we administered the CGPQ during this measurement period. Of the 455 children that participated in the ChiBS study in 2011, 331 participated again in 2012 (drop-out of 27.3%). Of the 331 participating children in 2012, only 288 children (87%) had complete data on the parenting constructs. Therefore, the study population for this paper includes 288 children. Based on the ChiBS data collected in 2011, we found that the 288 children participating in the analyses presented in this paper did not differ in age, gender, socio-economic status (SES) and BMI from the 455 children that participated in 2011 (p > 0.05).
The study was conducted according to the guidelines of the Declaration of Helsinki and was approved by the Ethical Committee of the Ghent University Hospital. All participants signed an informed consent.
Measurements
Fat mass determination by air displacement plethysmography
To determine the body fat percentage, Air Displacement Plethysmography (ADP) was used (BODPOD®, Software version 4.2.4, Life Measurement Inc, Cranlea and Co, Birmingham, United Kingdom). ADP is considered as a good reference technique for body composition measurements with a quick, comfortable, automated, non-invasive and safe measurement process, making it feasible in children [23]. Children had to refrain from physical activity and food consumption two hours before the measurement. The device was calibrated daily according to the manufacturer’s guidelines. Children were measured twice in tight-fitting bathing suit with swim cap to rule out air trapped in clothes and hair. Thoracic gas volume was predicted by the software with a validated child-specific equation and fat mass percentage was calculated using the equation reported by Wells [23]. If the first and second measurement of the body volume differed more than 150 ml, a third measurement was performed. To obtain reliable and valid body composition measurements, the ADP technique was conducted in all children by one trained study nurse.
Routine anthropometry
The routine anthropometric measurements were carried out in all children by one trained study nurse. Children’s weight was measured in underwear with an electronic scale linked to the BodPod ®(Tanita Corporation, Japan (Model BWB-627-A), modified by Life Measurement, Inc.) to the nearest 0.1 kg and height was measured with a stadiometer (SECA 225) to the nearest 0.1 cm. BMI (kg/m2) was calculated (weight (kg)/height2 (m2)). Age- and gender- specific BMI z-scores were calculated according to Cole’s method [24]. BMI categories for children were determined using the International Obesity Taskforce (IOTF) BMI categories [25]. Parental weight and height were self-reported. For the parents, BMI categories were determined as recommended by the World Health Organization (underweight BMI < 18.5 kg/m2, overweight BMI > 25.0 kg/m2, obesity BMI > 30.0 kg/m2) [26].
Questionnaires
Socio-demographic information
Parents reported their own educational level as well as that of their spouse and family characteristics (number of children, family structure, and number of older, younger or same-aged siblings than the participating child). Information on family structure was missing for 9 of the 288 children. Parental educational level was categorized according to the International Standard Classification of Education (ISCED): (level 0 ‘pre-primary education’, 1 ‘primary education’, 2 ‘lower secondary education’, 3 ‘upper secondary education’, 4 ‘post-secondary non-tertiary education’, 5 ‘first stage of tertiary education’, 6 ‘second stage of tertiary education’). The maximal ISCED level of both parents was used (this information was missing for 8 of the 288 children (2.8%) and based on information from only one parent for 13 of the 288 children (4.5%)). Most of the parents participating in this study had an educational level of 4 or more. Therefore, three groups were created: group 1 for the parents with educational level 0, 1, 2 or 3, group 2 for the parents with educational level 4 and group 3 for the parents with educational level 5 or 6. Children were described as youngest, oldest or middle child (when not being the youngest or the oldest).
Physical activity and sleep (parental-reported)
Parents were asked about the physical activity of their child: the hours of physical activity at sports clubs and outdoors per week was used. The number of screen time hours per week (i.e. television and computer time) was used as a measure of sedentary behavior. Parents also reported the typical hours of bedtime for weekdays and weekend days, from which the child’s average sleep duration per night was calculated as ‘(2*weekend + 5*week)/7’.
Comprehensive General Parenting Questionnaire (CGPQ) (parental-reported)
This 85-item questionnaire assesses five key parenting constructs that have been identified across multiple theoretical approaches of parenting: “nurturance”, “structure”, “behavioral control”, “coercive control”, and “overprotection” [21]. Each parenting constructs consists of different sub-constructs, each containing five items. It was designed by Sleddens et al. [21] and was validated in The Netherlands, Belgium and the United States. Parents were asked to indicate on a five-point Likert scale how much they agreed with each statement about parenting, ranging from 1 (strongly disagree) to 5 (strongly agree). The first construct ”nurturance” is the degree to which parents foster and recognize individuality and self-assertion by being supportive and responsive to their child’s needs, showing interest in child activities, spending time with their child, praising their child for good behavior, and expressing affection and care toward their child. It is composed by the parenting constructs ‘social rewarding’, ‘responsiveness’, ‘autonomy support’ and ‘involvement’. Parents scoring high on the second construct “overprotection” show higher scores on the two sub-constructs ‘excessive involvement’ (excessive nurturing) and ‘excessive monitoring’ (strict control). This negatively impacts child development through interfering with the development of children’s autonomy. The degree ‘excessive’ is applicable when parents show a level of involvement or monitoring that fits for a much younger child. The third construct “structure” indicates the degree to which parents organize their child’s environment, by helping their child when necessary to gradually achieve a certain goal, and consistently enforcing rules and boundaries. It consists of the sub-constructs ‘inconsistent discipline’, ‘consistency’, ‘organization’ and ‘scaffolding’. The fourth construct “behavioral control” can be regarded as parents’ supervision and management of their child’s activities, providing clear expectations for behavior and using disciplinary approaches in a non-intrusive manner. Parents scoring high on behavioral control provide adequate levels of control, they are not too strict or over-controlling, but rather allow their child to have enough space to develop independence and autonomy. It is composed of the parenting sub-constructs ‘monitoring’, ‘maturity demands’, ‘nonintrusive discipline’ and ‘considering child input’. The fifth construct “coercive control” is characterized by pressure, intrusion, domination, and discouragement of child independence and individuality. It consists of the parenting sub-constructs ‘psychological control’, ‘physical punishment’ and ‘authoritarian control’. The questionnaire was completed by the parent that accompanied the child to the survey center. In most cases, the mothers accompanied their children. However, we do not know exactly the number of questionnaires completed by mothers and/or fathers. Mean scores were calculated for each construct providing that at least 60% of the items for each subconstruct was completed. Cronbach’s alphas for each of the five higher order constructs were as follows: nurturance 0.74, overprotection 0.63, structure 0.53, behavioral control 0.33 and coercive control 0.63.
Children’s eating habits questionnaire (CEHQ)- food frequency questionnaire (FFQ) (parental-reported)
The CEHQ-FFQ is a screening instrument designed to investigate food consumption frequency and eating behaviors of children. It consists of 43 questions and was developed and validated within the EU FP7 IDEFICS project [27],[28]. Parents were asked to report on the frequency of their child’s consumption of selected food items in a typical week during the preceding 4 weeks using the following response options: ‘never/less than once a week’, ‘1-3 times a week’, ‘4-6 times a week’, ‘1 time per day’, ‘2 times per day’, ‘3 times per day’, ‘4 or more times per day’ or ‘I have no idea’. There were no questions about portion sizes. In this study we used the following categories: snacks (nuts, seeds, chips, popcorn, savory pastries, chocolate, candy, cookies, ice cream), sweet food (sweet drinks, sweet sandwich filling such as jam and chocolate spread, sweet breakfast cereals, sweeten diary and sweet snacks), soft drinks (light and non-light), fatty food (fried potatoes, high fat sandwich filling such as butter and chocolate spread, high fat dairy, sauces, cheese, fat meat preparations and high fat snacks), and finally the healthy group of fruit and vegetables. When parents did not fill in a question that was needed to calculate the consumption frequency of a certain food group, this lead to a missing value for that food group and that child.
Dutch eating behavior questionnaire (DEBQ) (child-reported)
The DEBQ is a 33-item questionnaire that assesses three types of eating behavior in children: eating in response to negative emotions (emotional eating, 13 items, e.g. Do you have a desire to eat when you are emotionally upset?), eating in response to the sight or smell of food (external eating, 10 items, e.g. If you have something delicious to eat, do you eat it straight away?) and eating less than desired to lose or maintain body weight (restraint eating, 10 items, e.g. Do you try to eat less at mealtimes than you would like to eat?). In all three types of eating behavior, the appropriate self-regulating mechanism of food intake is diminished or lost. Children could answer the questions with ‘never’, ‘almost never’, ‘sometimes’, ‘often’ or ‘very often’ as response alternatives [29]. Answers were than recoded in digits (never = 1, almost never = 2, sometimes = 3, often = 4 and very often = 5). When children did not fill in a question that was needed to calculate the score for one of the three eating behavior types, this lead to a missing value for the score of that eating behavior type. The following Cronbach's alpha values were found: emotional eating 0.91, external eating 0.77 and restrained eating 0.88.
Statistical methods
The analyses were conducted using SPSS, version 21.0. The cut-off for significance was chosen at p < 0.05. Variables with a non-normal distribution were log-transformed. Descriptive statistics were performed to determine means and standard deviations for continuous variables and frequencies for categorical variables. Pearson’s correlations were used to test associations among continuous key variables. Based on the work of Cohen [30] we considered correlation coefficients smaller than 0.3 as small, those between 0.3 and 0.5 as moderate and those larger than 0.5 as large. ANOVA analyses and t-tests were used to test differences among key variables between groups. When a significant association was found between a parenting construct and a variable, the associations between the corresponding parenting sub-constructs and that variable were also tested for significance (e.g. when a significant association was found between “overprotection” and maternal BMI, than we also tested the relation between the maternal BMI and the sub-constructs of “overprotection” – being “excessive monitoring” and “excessive involvement” – were investigated. Multiple linear regressions were performed to determine the relationship between the parenting constructs and the key variables. First, the association of family characteristics (number of children, educational attainment of the parents, BMI of the parents) and child characteristics (age, gender, BMI of the child and child birth order in the family) with parenting behavior was investigated. Second, it was investigated whether the parenting behavior shows an association with several types of the children’s health related behavior (food consumption frequency, eating behavior, physical activity, sedentary behavior and sleep duration) controlling for child and parental characteristics. All models were corrected for the child's age and gender, parental ISCED-level, BMI father, BMI mother, child birth order and the other parenting constructs assessed by the CGPQ. The confounders were chosen on the basis of existing knowledge [6],[31] and univariate analyses done in this study, i.e. parameters that showed some evidence of association (p < 0.05). In model 1, the mutual relation between the different parenting constructs was tested. In five other models, the association between the parenting style and health related behaviors of the children were tested: in model 2 with the food consumption of the children, in model 3 with eating behavior, in model 4 with physical activity and sedentary behavior, in model 5 with sleep duration and in model 6 with children’s body composition (i.e., BMI z-scores and fat mass percentage). In model 6, we also adjusted for sleep duration, physical activity and consumption frequency of different food items: sweet foods, soft drinks, fruits and vegetables. This was done because these parameters are known to influence children’s body composition.