CHAPTER 2 METHODS
The data used were taken from the sixth sweep/year of GUS which was conducted over a 14-month period starting in April 2010, when survey children were approaching their sixth birthdays (average age 5 years 10 months). In all, interviews were conducted with 3657 families (representing 70% of the 5217 families interviewed in the first sweep of data collection). At sweep 1, the survey was designed to be representative of the Scottish population. To compensate for biases introduced through survey attrition, longitudinal survey weights were used in all analyses (see 2.6).
Among the sweep 6 families, we excluded children from multiple births ( n=67) and cases where the survey respondent was not the child's natural mother (a further 89 cases), leaving 3501 cases (96% of families contacted at this sweep). To avoid inconsistencies arising from differences in the identity of the main carer interviewed at earlier sweeps, we then further restricted the eligible sample to cases where the natural mother had provided information at all previous sweeps ( n=3160). This restricted sample was similar to the full sweep 6 sample, while containing slightly lower proportions of children from ethnic minority groups (2.9% compared to 3.6% in the full sample), mothers without formal educational qualifications (8% compared to 9.4% in the full sample), families with absent fathers (22.2% compared to 23.4%) and families living in the most deprived areas (20.5% living in the most deprived SIMD quintile, compared to 21.9% in the full sample).
At each sweep/year of fieldwork, interviews took place around six weeks before the child's next birthday, therefore in the first year of the study, children were 10 months old, in the second year they were 22 months old and so on. For the purposes of this report, beyond the first interview, the child's age will be referred to in years. It is worth bearing in mind however, that a 4-year-old child at sweep 4 for example, is actually 46 months old or just under 4, and a 6-year-old child at sweep 6 is actually 70 months old or just under 6.
2.2 Defining overweight and obesity
Body Mass Index ( BMI) scores were calculated using children's height and weight measured by trained researchers at the sweep 4 and 6 interviews. Classification of overweight and obesity in children using BMI is complicated by variation in patterns of growth, weight gain and changes in body composition. To define overweight and obesity, UK thresholds appropriate for the child's gender and age using the National BMI percentiles classification (based on UK 1990 reference growth curves) were applied to children's BMI scores, following a method developed by ISD Scotland also used for the Scottish Health Survey (Gray and Leyland 2011). Most children (those below the 85th percentile) are classified as of healthy weight, with a small number at or below the 5th percentile who are underweight. Children at or above the 85th percentile but below the 95th percentile are considered to be overweight. Children at or above the 95th percentile are classed as obese, including those at or above the 98th percentile who are classed as morbidly obese.
This study examined two main groups of children: those considered either overweight or obese (all those at or above the 85th percentile), and those considered obese (all those at or above the 95th percentile).
2.3 Defining physical activity and sedentary behaviour
Two measures were created to show whether children were meeting two of the recommended UK guidelines (Department of Health 2011) for physical activity. Intensity of children's physical activity is difficult to measure with survey questions. There was no measure of vigorous activity on at least three days a week, so - like the Scottish Health Survey - this report focuses on the first and third guidelines.
The two measures were:
1. Physical activity for 60 minutes or more, seven days a week
At sweep 6, mothers were asked about their child's participation in four types of physical activity (for at least 5 minutes) in the past week. For activities outside school, these included walking, organised sports or exercise activities such as swimming lessons or a gymnastics class, and other active behaviour such as running about, riding a bicycle or kicking a ball around. Mothers were also asked whether their child had taken part in walking, sports, exercise or other active things when the child was at school.
For each of these different types of physical activity, mothers were asked for the number of weekdays and the number of weekend days on which their child had engaged in the activity, and for the duration of each type of activity on both a weekday and a weekend day in the past week. Average duration was recorded using a 10-point scale, running from (1) at least 5 minutes but under 15 minutes to (10) 4 hours or more. Following procedures developed for the Scottish Health Survey, this information was used to estimate typical total physical activity time per week and to classify children into three groups:
- those active for 60 minutes or more on all seven days
- those active for between 30 and 59 minutes on all seven days
- those with lower levels of activity
This report compared the first group (those meeting the guideline for recommended physical activity) with the second and third groups of children who had lower than recommended activity levels.
2. Sedentary behaviour
Screen time was used as a measure of sedentary behaviour. Mothers were asked how long their children spent watching television (including DVDs and videos), and how long they spent using a computer or games console, on a typical weekday and a typical Saturday. Duration was recorded in hours, or fractions of hours.
This report compared children who used screens for under 3 hours on a typical term time weekday with those who used screens for 3 hours or more. (See section 4.7 for more information on why this cut off was chosen).
2.4 Conceptual framework to explore factors associated with children's overweight and obesity at age 6
We explored associations between possible risk factors and children's overweight and obesity at age 6. There were two main sets of risk factors: parental factors, and family and neighbourhood constraints. Detailed information on derived measures used is provided in the Appendix, section 8.1.
2.4.1 Parental factors
There were three main types of parenting factors (Table 2.1 shows the sweep(s) at which information was available):
- Parental example. Mother's overweight or obesity measured at age 6, and mother's physical activity and screen time were measured at age 5. Note that there are no measures of maternal diet in the GUS survey.
- Health-related practices regarding children's nutrition, physical activity and sleep. Here, measures of children's behaviour such as eating sweets or watching TV have been included under "parenting", on the assumption that these behaviours are to a large extent modifiable by parents.
- General patterns of parenting. This covered "connection" (loving and caring for the child), control (management of the child's behaviour) and conflict. These measures were previously used in a Growing Up in Scotland report on Parenting and Children's Health (Parkes and Wight 2011). Connection was measured using questions about the warmth of the parent-child relationship, and the number of joint parent-child activities (such as reading and playing). Control was measured using questions about rules for the child's behaviour, parental supervision of the child and "household chaos" or disorganisation. Conflict was measured using questions about the degree to which parent and child struggle with one another, or how easily the child becomes angry with the parent.
Table 2.1 Parental factors examined in relation to children's overweight and obesity at age 6
|Building blocks of model||Measures||Age measured/Sweep(s) providing information|
|1) Parental example||Mother's overweight/obesity||6|
|Mother's physical activity||5|
|Mother's screen time||5|
|2) Health-related practices i) Nutrition||Infant feeding (duration of breastfeeding, timing of introduction of solids)||1|
|Pattern of eating (snacks and/or meals, whether has regular mealtimes, whether skips breakfast), room where child eats main meal||2,5|
|Snacks (what child mainly snacks on: crisps, fruit, etc), frequency of sweets, crisps and soft drinks, mother's perceived ease of regulating sweet consumption)||2,5|
|Fruit and vegetable consumption (variety)||2,5|
|Consumption of fast food and convenience food||3,5,6|
|ii) Physical activity and sedentary behaviour||Use of local facilities (swimming pool, play park)||2|
|Active behaviours (walking, playing outdoors, etc)||2,5,6|
|TV in bedroom||4,5|
|Time watching TV/playing electronic games||3,4,5,6|
|iii) Sleep||Duration of sleep||1,3,6|
|3) General patterns of parenting||Warmth of relationship with child||5|
|Joint activities with child||2,3,4|
|Conflict in relationship with child||5|
2.4.2 Family and neighbourhood constraints on healthy weight management
Family and neighbourhood constraints are factors that might limit a family's ability to follow a healthy diet and engage in recommended levels of physical activity. Information on the measures available in GUS is shown in Table 2.2.
Family constraints may be related to lack of time, low income, cultural norms, low awareness of healthy weight and healthy eating and exercise practices, and stress on family functioning due to poor maternal physical and mental health. Mother's smoking during pregnancy has been included as a potential health-related constraint. There is evidence that maternal smoking during pregnancy is associated with child overweight even after allowing for sociodemographic and behavioural differences between smokers and non-smokers (Oken et al. 2008), although the mechanisms for this effect are not clear.
Neighbourhood constraints might reflect availability, access, quality and safety in relation to various aspects of the physical and social neighbourhood environment.
Within each of the two sets of family and neighbourhood constraints, we can distinguish factors that are distal or more proximal. Proximal factors might explain how more distal factors affect obesity. For example, within the set of family factors a mother's long working hours is here considered as a distal factor, possibly influencing obesity through the more proximal factor of time constraints on food preparation (perhaps leading to greater use of convenience foods?). Within the set of neighbourhood factors a more deprived neighbourhood (distal factor) might influence obesity via more proximal factors, such as low neighbourhood safety and poorer quality play areas for children.
Table 2.2 Information on family and neighbourhood constraints associated with children's overweight and obesity at age 6
|Building blocks of model||Measures||Age measured/Sweep(s) providing information|
|Mother's time at home||Mother's employment (hours)||5|
|Income, culture and knowledge- related||Mother's age at birth of first child||1|
|Biological father in household||1 to 6|
|Household income||1 to 6|
|Mother's social class||1|
|Health-related||Mother's mental health||1,3,5|
|Mother's physical health||1,3,5|
|Mother's smoking during pregnancy||1|
|Time||Mealtime quality (a rush, time to talk, enjoyable)||2,5|
|Mother's food preparation affected by time||2,5|
|Income, culture and knowledge- related||Mother's food preparation affected by cost, knowledge of healthy eating||2,5|
|Mother's recognition and concern for child's weight||4,6|
|Mother's views on desirable levels of children's physical activity||6|
|Distal||Area deprivation||1 to 6|
|Urban-rural classification||1 to 6|
|Access to places to play/exercise/eat||Access to garden||1|
|Access to play park||1|
|Proximity to green space||6|
|Swimming pool in area||2|
|Take away in area||6|
|Quality of places to play/exercise||Rating of green spaces and play areas||6|
|Safety/trust in local area||Safety by day and night, antisocial behaviour problems, safe play areas, "child friendly" neighbourhood||3,6|
2.5 Conceptual framework to explore factors associated with children's physical activity and sedentary behaviour at age 6
Exploration of factors associated with children's physical activity and sedentary behaviour followed a similar pattern to the outline given in the previous section. Under parental factors, parental example, health-related practices related to children's physical activity and general patterns of parenting were included. Health-related nutritional practices were excluded. Under family and neighbourhood constraints, there was the same list of distal factors. Proximal family and neighbourhood constraints included those related to exercise ( e.g. mothers' views on desirable levels of children's physical activity and quality of local play areas). Other measures related exclusively to nutrition, such as mother-reported time or money constraints on food preparation, were omitted.
All analysis allowed for the stratified and clustered survey design and weighted the data to compensate for attrition of the survey sample between sweeps 1 and 6. Tables display both weighted (Wtd) and unweighted (Uwtd) bases. Bivariate exploration was carried out using SPSS complex surveys syntax. Multivariate modelling was carried out using STATA after multiple imputation of missing data. Multivariate modelling controlled for child characteristics (gender, birthweight and white/minority ethnic group 10 ) and family characteristics (number of siblings). Birth weight was standardised for gender, gestational age and parity using tables recently published for Scottish infants (Bonellie et al. 2008). Readers interested in the details of the analyses should consult the Technical Appendix published alongside this report.
This section describes the prevalence of overweight and obesity amongst Scottish children at age 6. It examines whether age 4 overweight and obesity were associated with being overweight or obese two years later. It then goes on to explore factors associated with children's overweight or obesity at age 6.