CHAPTER 6 SUMMARY AND CONCLUSIONS
This section summarises the main findings, discusses some of the limitations of the study and reflects on implications for policy and practice.
The study focused on the identification of potentially modifiable factors associated with children's overweight and/or obesity, low physical activity and high sedentary behaviour at age 6. The report also investigated mothers' awareness of, and concern about, their child's overweight or obesity.
Children's weight and height were measured by trained researchers at age 6 (average age 5 years 10 months). Overweight or obesity were classified using BMI cut-offs derived from national population growth charts. In the GUS sample of 2,992 children, 22% were classified as overweight or obese, and 9% were obese.
Physical activity and sedentary behaviour were measured using mothers' reports of their child's behaviour. Physical activity was based on active behaviour outside school including walking, organised sport and exercise and active play during the past week, using separate questions about weekdays and weekends, together with all active behaviour (including walking) at school. Children were divided into those meeting or exceeding UK Government recommended levels of 60 minutes daily, and those below this target level. In the GUS sample, 15% of children exercised for less than the target of 60 minutes daily while the majority (85%) appeared to meet the target. Sedentary behaviour was based on mothers' reports of typical weekday screen time (watching TV or playing electronic games). Children were divided into those watching up to 3 hours daily (69%) and those watching for three hours or more (31%). Children who watched screens for 3 hours of more were slightly less likely to meet recommended levels of physical activity compared to children using screens for shorter periods.
Children's overweight and obesity from BMI measurements taken at age 4 (average age 3 years 10 months) were associated with a much greater chance of being overweight or obese at age 6, when compared to children who had a healthy weight at 4 years old. Despite this "tracking" of overweight and obesity from age 4 to 6, there was considerable movement out of the overweight and obese groups. Around four in ten children who were overweight or obese at age 4 were classified as having a healthy weight two years later. Half the children who were obese at age 4 were no longer obese at age 6, although most were still overweight.
Children's physical activity and sedentary behaviour measured at earlier sweeps predicted their physical activity and sedentary behaviour respectively at age 6. This suggests that habitual physical activity or sedentariness may develop at an early age. However, children who spent long periods using screens at young ages were not more likely to show low physical activity, and conversely low physical activity did not predict high screen use.
Two main sets of possible influences on children's overweight/obesity and activity levels were considered: (1) parental factors, and (2) possible constraints on the child's healthy lifestyle related to family or neighbourhood factors.
Parental factors included mother's overweight and her modelling of active and sedentary behaviour; child health-related practices likely to be under the parent's control such as children's snacking on unhealthy foods or playing outdoors; and general patterns of parenting.
Family and neighbourhood constraints ranged from more "distal" or background factors (such as family income or area deprivation) to more "proximal" influences that might explain how some of the distal factors influenced either obesity or physical activity.
First, associations between separate individual risk factors and the outcome measures of children's overweight and/or obesity and children's activity levels were explored. Individual factors that had a statistically signficant association with the outcomes were included in multivariate modelling, where it was possible to control for a number of influences simultaneously. The multivariate models also allowed for various child and family characteristics. These were gender, ethnicity, age in months at age 6, family size, birth weight (overweight/obesity models only), child health (activity models only) and season (activity models only).
6.1 Factors associated with children's overweight and/or obesity
There were no differences in child overweight and/or obesity according to the child's gender, or between white and ethnic minority groups. However, low numbers in minority ethnic groups did not permit further subdivision. The Millennium Cohort Study has oversampled minority ethnic groups to enable more finegrained analysis, and has found ethnic differences in children's overweight/obesity (Brophy et al. 2009). Within GUS, children with a higher birth weight were more likely to be overweight and/or obese at 6 years, as were only children, compared with those from large families (four or more children).
In the final multivariate models of children's overweight and/or obesity, significant parental influences included mother's overweight or obesity, dietary practices (consumption of sweets and crisps when a toddler, skipping breakfast, not eating the main meal in a dining area of the home), and low parental supervision.
Mothers who were overweight or obese were more than twice as likely as mothers of healthy weight to have obese children. The strong association between mother's overweight or obesity and children's BMI classification has been well established elsewhere. It may reflect one or more influences including an inherited predisposition to put on weight, epigenetic effects, continuation of the same environmental influences that led to a mother being overweight, and shared health practices related to diet and exercise (Wu and Suzuki 2006).
Conceptualising frequent consumption of sweets and crisps as an indicator of unhealthy early dietary practices comes with the cautionary note that the prevalence of overweight and obesity was significantly lower only for the minority (around 20%) of toddlers who were given sweets and crisps once a week or less. The measure is therefore only useful as an indicator of children at lower risk of overweight and obesity, rather than predicting higher risk. Skipping breakfast was also found to be associated with 5-year-olds' overweight in the UK Millennium Cohort Study, and may reflect a more general association with poor dietary habits (Rampersaud et al. 2005).
Not eating the main meal in an area primarily designed for food consumption (either the kitchen, dining room or a combined living/dining room) could indicate eating more unhealthy foods, perhaps under less parental control. Low parental supervision could reflect lower maternal feelings of protectiveness as well as lower monitoring of a child's health-related behaviour. The finding appears to echo an effect of parental neglect on obesity found elsewhere (Lissau and Sorensen 1994; Whitaker et al. 2007; Knutson et al. 2010). Parental supervision was also found in an earlier GUS analysis to be important for other aspects of these children's health (Parkes and Wight 2011).
Many health-related practices suggested elsewhere as important for childhood overweight and obesity were not apparent in this study (Hawkins and Law 2006; Moreno and Rodriguez 2007; Monasta et al. 2010). These included maternal smoking, shorter duration or absence of breastfeeding, and early introduction of solids, although this report found that dietary practices later in childhood had a stronger effect. There was no effect of soft drink consumption or fruit and vegetable consumption as identified in some previous studies, nor of how often convenience or fast food was used for the child's main meal. There was also no effect of short sleep duration, low physical activity or high sedentary behaviour.
The analyses identified several constraints on children's healthy weight management, including poor maternal physical health and low "child-friendliness" of the neighbourhood. The effect of poor maternal physical health requires further investigation, and could be linked to longstanding overweight in mothers. Mother's BMI was only available at age 6, but other research has shown effects of pre-pregnancy overweight on children's obesity (Hawkins et al. 2009c; Wright et al. 2010). Low "child-friendliness" was based on mothers' asessments of the overall social and physical environment in which they were bringing up their child. It could reflect aspects such as neighbourhood safety, antisocial behaviour and the quality of local play areas and parks that individually had weaker associations with childhood obesity. The mechanism for any effect on obesity is unclear. It seems unlikely that child-friendliness is simply acting through features of the social and physical environment conducive to physical activity, since children's activity levels as measured in GUS did not have clear associations with obesity. Further research is needed using objective measures of neighbourhood quality, to see the extent to which mothers' perceptions are in line with these.
In preliminary modelling, small effects of low maternal education and area deprivation on increased risk of obesity were identified. However, these disappeared once infant feeding practices, eating crisps and sweets and neighbourhood child friendliness were included in the models. It appeared that the mechanism for the effect of low maternal education could be via poor dietary practices, while that for the effect of area deprivation could be low child-friendliness. There was no independent effect of maternal smoking in pregnancy on children's overweight and/or obesity, as found in some other research (Oken et al. 2008).
6.2 Factors associated with children's low activity levels (physical activity and sedentary behaviour)
Although there were no gender differences in meeting the recommended 60 minutes' daily physical activity in the 6 year olds studied, boys were more likely than girls to use screens for 3 or more hours daily. Children from ethnic minority groups were more likely to have low levels of physical activity, but there were no differences between white and ethnic minority groups' screen time. More work is required to explore the ethnic difference in physical activity, with a sample containing higher numbers from different ethnic minority groups. Children from larger families were more physically active than only children, which might suggest the importance of play with other children for maintaining good levels of physical activity in children. We do not have information in the GUS survey on who does physical activities with the child, but other research has shown the importance of sibling as well as parental activity (Sallis et al. 2000b). Physical activity (but not screen time) varied strongly with daylight hours, indicating the importance of outdoor activities and (possibly) when children are able to walk or cycle to school more safely.
Children were more physically active if their mother was herself more active, had a warm realtionship with the child and was more knowledgeable about desirable amount of physical activity for children. Lack of a nearby swimmming pool was associated with lower overall physical activity.
Children used screens more if they had a TV in their bedroom, if their mother watched TV or used a computer a lot, or if their mother did fewer activities with the child and had fewer rules for the child's behaviour. Unlike children's physical activity, screen time increased with the level of social deprivation, as indicated by low maternal educational level, mother smoking in pregnancy and poor quality local green spaces. The effect of mother smoking in pregnancy on increased screen time has been identified elsewhere (Oken et al. 2008). The mechanism is unclear, but may be linked to social deprivation. More work is required using an objective measure of green space quality alongside mother's reports. It remains unclear whether providing more attractive green spaces would encourage both more activity and less sedentary behaviour, since the analysis for this report suggests the two behaviours are not closely linked.
6.3 Mother's recognition and concerns for their child's overweight
Mothers of an overweight or obese child were much more likely to describe the child as "normal" rather than overweight. Only 3% of mothers of overweight but not obese children described them as overweight, although recognition was higher when children were obese (30%). Mothers were also more likely to recognise overweight in a daughter than in a son.
Some previous work (Layte and McCrory 2011) has suggested that mothers who are overweight themselves are less likely to describe their child as overweight, once the child's objectively measured BMI has been allowed for. There was no evidence for this among mothers in the whole GUS sample, in agreement with another study (West et al. 2008). In the subsample containing overweight or obese children, mothers who were overweight or obese themselves were more, rather than less, likely to recognise overweight in their own child.
Mothers' recognition of overweight was associated with greater concern for the child's weight. Mothers' concern was greater if their overweight child was a girl rather than a boy, but did not appear to be linked to the mother's own weight.
These findings suggest that more could be done to increase mothers' understanding of what constitutes a healthy weight in young children. However, mothers' recognition and concerns for overweight at age 4 were not associated with child weight loss in the two-year period to age 6 BMI measurement.
6.3.1 Strengths and limitations of the study
This study was able to examine a wide range of potential risk factors for overweight and/or obesity and low activity in a large sample of Scottish aged just under 6. The use of statistical techniques to counteract the effects of attrition in the GUS sample over the six sweeps of data collection helped to ensure that the findings are likely to be representative of the six-year-olds in the Scottish population.
Most, but not all, potential risk factors were measured in previous sweeps. This longitudinal element makes reverse causation less likely, but it is important to note that the associations found do not necessarily show causal relationships. Associations may be due to other confounding factors that have not been studied, including influences that predate the birth of the child. The study cannot distinguish between environmental and genetic effects that may be inherent in, for example, the association found between mother and child overweight.
The analysis presented in this report is limited by the type and detail of the questions included in GUS. Although children's and mother's BMI were based on researchers' measurements of weight and height, most other measures rely on mothers' reports. Mothers' reports may be subject to various biases, including social acceptability and positive or negative affect. Parent-reported activity measures may be unreliable, and overestimate children's habitual physical activity measured objectively using accelerometers (Basterfield et al. 2008). Screen time was used as a proxy measure for sedentary time. Self-report of activity rather than objective measurement might explain the lack of association between activity and children's BMI in this study.
The study did not include detailed dietary information on portion sizes and frequency, which would require the use of food diaries. Other relevant information was also not available. In the family, there was no information on partner's BMI classification (found elsewhere to be important in addition to maternal overweight, Whitaker et al. 2010), mother's BMI at earlier ages and mother's diet. In the wider community, missing information includes objective measures of neighbourhood facilities and safety issues, and school level factors. Recent studies looking at the spatial distribution of resources and facilities have found that people living in more deprived areas in Scotland have more resources such as outdoor play areas, sports centres and swimming pools close by than those living in affluent areas (Macintyre et al. 2008; Ogilvie et al. 2011).
6.4 Policy implications
While more research could be done to supply more objective measures of some of the outcomes and risk factors, and to investigate causal mechanisms, this study has suggested a range of risk factors for young children's overweight and low activity, at both family and neighbourhood levels. These factors are potentially modifiable, or would at least allow risk groups to be identified. The "tracking" of child overweight and activity from earlier ages to age 6 suggests that early intervention may be most valuable.
The strong association between mother's own overweight and that of her child (well established elsewhere) and the importance of other parental factors suggest that interventions to reduce child overweight and obesity may be most successful if they treat the family as a unit, rather than focusing exclusively on the child's weight problem. Improving a mother's recognition and concern for her child's overweight is likely to be only a first step in tackling the problem, as greater awareness may not translate into better weight management without addressing other aspects of parental behaviour. For children's weight control, these appear best targeted through improving dietary practices and overall parental supervision.
For physical activity, there are also signs that interventions should be at the family rather than the child level, encouraging parents and children to share a similar active lifestyle with a close parent-child relationship and structured behaviour management. Increasing mothers' awareness of desirable levels of physical activity for their children may also have a positive effect on parental support for the child.
Better access to places where children can be physically active, including attractive green spaces, may increase activity levels and reduce sedentary behaviour. The "child-friendly" aspects of the neighbourhood associated with a lower level of child obesity require further investigation in order to understand what is most important here.