Background
Primary prevention of childhood overweight is an international priority [
1]. In Australia in 2007, 18-21% of 2-8 year olds were already overweight, including 4-6% who were obese [
2]. Corresponding figures for New Zealand children aged 2-14 years show 21% overweight, including 8% who were obese [
3]. Childhood overweight has an immediate negative impact on physical and psychological health and well-being [
4,
5]. In addition, overweight children have a substantially increased risk of becoming overweight adults, with an attendant increased risk of morbidity and mortality [
6,
7].
Childhood obesity prevention studies to date have largely targeted older children in school or community settings, and have had limited effectiveness, partly due to design and methodological issues [
8‐
10]. The 2005 Cochrane Review of interventions for preventing childhood obesity included 22 quality trials, only three of which included children less than five years of age [
10]. A more recent systematic review of the effect of interventions on weight status of children aged from birth to five years identified seven of 22 studies that included children less than two years of age [
9]. However, only two of these seven studies provided growth data and hence the majority of studies in preschool aged children provide impact evaluation rather than outcome evaluation of the direct effect of interventions on weight status. Given that so few studies have focused on very young children, it is likely that existing interventions have started after feeding, eating, sleeping, television (TV) viewing and activity patterns have been established and are more difficult to modify.
There are numerous reasons why starting interventions to prevent childhood obesity very early may be effective. Rapid early weight gain before two years of age is associated with increased risk of overweight in later childhood [
11‐
13] and most excess weight gained before puberty is gained by the age of five years (91% girls, 70% boys) [
14]. A range of potentially modifiable factors operating early in life are also likely to be linked with later obesity or obesity-promoting behaviours. These are summarised below.
Dietary patterns and eating habits
A meta-analysis has shown that children who were breastfed have significantly lower levels of obesity than those who were formula fed [
15] although this association remains controversial [
16]. The types and texture of foods offered to infants as they transition from milk to family foods are key determinants of early food preferences, intake patterns and dietary quality [
17‐
19] which then track to older ages and are associated with later obesity risk [
20].
Parental early feeding practices
Parental feeding practices strongly influence children's eating behaviours, which are firmly established by five years of age and lay the foundation of adult eating habits [
21,
22]. Parental feeding practices determine infant exposure to food (type, amount, frequency), and include responses (e.g. coercion) to infant feeding behaviour (e.g. food refusal). The degree of parental control in child feeding (including restriction, monitoring or pressure) is associated with later child feeding behaviour (preferences and intake) and weight status [
23].
Physical activity
Most studies show an inverse relationship between physical activity levels and adiposity in children. High levels of physical activity reduce the likelihood of weight gain over time [
24,
25]. A systematic review of strategies to reduce obesity in children has shown that changes in weight occur most frequently in interventions that also demonstrate change in physical activity levels [
26]. However, the systematic review noted that only 3% of studies (n = 5) were conducted in preschool settings.
TV viewing practices and sedentary behaviours
Children who watch TV for more than two hours per day are more likely to be obese, have unhealthy dietary patterns and low levels of physical activity [
27]. Many young children exceed this viewing threshold and such patterns of sedentary behaviour track throughout childhood [
2,
28].
Sleep patterns
Numerous cross-sectional and prospective studies have linked shorter sleep duration with obesity [
29‐
31] and increased cardiovascular risk [
32]. Studies are urgently required to determine the effects of altering sleep patterns on body weight during childhood.
Parenting style
Emerging evidence has linked parenting style (e.g. authoritative, authoritarian, libertarian) to early feeding practices, child eating behaviour and weight status. A recent review concluded that the majority of the evidence is cross-sectional, with only seven of 67 studies providing longitudinal follow-up, and none including children aged less than five years [
23]. Only two studies (both in preschools) examined whether parenting feeding practices can be modified and most studies failed to evaluate covariates, particularly maternal weight status and family socioeconomic status.
Other potential effect modifiers
A range of other factors also influence the development of excess weight gain in childhood and may, in turn, modify responses to obesity prevention interventions. These include birth weight (both low
and high birth weight) [
33]; lower socioeconomic status [
34]; parental, and especially maternal, body mass index (BMI) [
35]; and maternal smoking during pregnancy [
35].
Overall, there is a strong argument for interventions that start early in childhood to be effective in preventing childhood obesity. However, there are as yet no published trials that provide quality evidence to guide the design, content and implementation of effective interventions that target infants. There are currently four randomised controlled trials underway in Australia and New Zealand (Healthy Beginnings [
36], Nourish [
37], The Melbourne InFANT Program [
38] and POI.NZ [
39]) that are investigating the effects of early intervention in preventing childhood obesity. The trials are similar with respect to design, participants, intervention, comparators and outcome measures. Table
1 gives the details of each trial.
Table 1
Characteristics of studies included in EPOCH Collaboration
Registration number | ACTRN12607000168459 | ACTRN12608000056392 | ISRCTN81847050 | NCT00892983 |
Funding source | NHMRC | NHMRC | NHMRC | HRC New Zealand |
Number randomised | N = 667 first time mothers | N = 698 first time mothers | N = 559 first time mothers | N = 400 first time mothers |
Baseline data | Antenatal | Infant aged 4-6 months | Infant aged 3 months | Antenatal |
Primary outcome | Child weight and height at 24 months | Child weight and height at 24 months | Child weight and height at 18 months | Child weight and height at 24 months |
Usual care | Wk1 child health nurse home visit Self access to child health clinics | Self directed access to child health clinics | Access to 10 scheduled child health clinic visits | First 4 wks midwife home visits; Well Child (Plunkett) nurse: 8 visits in 5 yrs |
Intervention mode | Home visits; 8 visits over 2 years (antenatal, 1, 3, 5, 9, 12, 18, 24 months); maternal advice | Two education peer support modules (6 fortnightly sessions each) commencing when infant 4-7 months and again at 13-16 months at community health venues | Six 2 hour sessions delivered at 3 monthly intervals within pre-existing mothers groups, commencing at 3 months (3, 6, 9, 12, 15 and 18 months) | 3 groups: sleep (home visits at 3 wks, 4 months); healthy eating and activity (mix of 7 home visits and group based sessions at 1 wk + 3, 4, 7, 9, 12, 18 months) or both |
Intervention content (in addition to usual care) | Sustaining breastfeeding; Timely solid food introduction; Responsive to child cues of hunger & satiety; Healthy child food intake; Reduced TV viewing; Promote active play; The "how" of child feeding (e.g. managing food fussiness) | Neutral repeated exposure to a variety of foods; Responsive to child cues of hunger & satiety; Healthy child food intake; Reduced TV viewing; Promote authoritative parenting style; The "how" of child feeding (e.g. managing food fussiness) | Responsive to child cues of hunger & satiety; Healthy child food intake; Reduced TV viewing; Promote active play; The "how" of child feeding (e.g. managing food fussiness) | Sustaining breastfeeding; Timely solid food introduction; Responsive to child cues of hunger & satiety; Healthy child food intake; Reduced TV viewing; Promote active play; The "how" of child feeding (e.g. managing food fussiness); Developing good sleep habits |
Control group | Usual care plus written home safety/tobacco prevention information at f/up sessions plus three mail outs | Usual care plus quarterly newsletter on general child health messages excluding sleep, food and activity | Usual care plus quarterly newsletter on general child health messages excluding sleep, food and activity | Usual care |
The Early Prevention of Obesity in CHildren (EPOCH) collaboration was formed in 2009 with the objective of conducting an IPD PMA of these trials to provide the necessary evidence regarding the efficacy of interventions to reduce obesity-related risk factors early in infancy. The key feature of this prospective collaboration is to define and clearly specify the objectives, research questions, specific aims, hypotheses, subject eligibility criteria, subgroups, predictors, outcomes (primary and secondary) and the general analysis plans of eligible studies in advance of knowing or publishing individual trial results [
40]. IPD PMA provides more reliable estimates of treatment effects through prospectively planned combined analyses of randomised controlled trials.
While the individual trials in the EPOCH Collaboration will provide important data in their own right, combining such data via an IPD PMA will yield even more powerful information. This is particularly important in this area as any one trial, even with many hundreds of children, may alone not have sufficient statistical power to detect an overall reduction in population-relevant outcomes. For example, to show a reduction in the prevalence of overweight/obesity at age two years from 20% to 15%, 1800 participants would be required. At the current recruitment and retention rates, the combined data from the four EPOCH trials will yield a final sample size of 1810 children by age two years. The additional power conferred by the PMA, combined with the availability of the IPD, will enable a range of subgroup analyses to be performed, which will explore other factors that may be associated with obesity in children and may modify intervention effects. The information obtained from the EPOCH Collaboration will be used to guide decision making regarding investment in child health interventions which are effective in reducing the prevalence of childhood obesity and associated negative short- and long-term health outcomes.
Funding
Initial funding for the EPOCH Collaboration has been received from the Meat and Livestock Association of Australia and further funding will continue to be sought from relevant funding agencies. Each individual trial has received funding from their own respective funding bodies, including the National Health and Medical Research Council (Australia), Health Research Council (New Zealand) and the Sydney South West Area Health Service. Funding bodies must provide support on condition that they will not have any input into the protocol design, data collection, data analysis or in any decisions to publish the results of the EPOCH Collaboration or any of the individual trials.
Summary
We have described the protocol for the EPOCH Collaboration which is the first individual patient data prospective meta-analysis of early childhood obesity prevention trials in the world. It will help to answer a key public health question: do interventions implemented in the first year of life prevent obesity, and influence weight status and a range of lifestyle-relevant behavioural outcomes at 18-24 months of age? We anticipate that the results will be available for publication in 2013 (see Table
2). The proposed research pushes knowledge boundaries in terms of the individual trials and the use of the innovative IPD PMA methodology. The information obtained from the EPOCH Collaboration is needed to justify investment in child health services to provide universal access to best-practice programs that are the most effective in reducing the prevalence of childhood obesity and associated negative short- and long-term health outcomes.
Table 2
Timeline of studies included in the EPOCH Collaboration
Healthy Beginnings
36
| Australia (NSW) | 667 | June 2007 | Dec 2008 | 2 years | June 2011 |
Nourish
37
| Australia (QLD, SA) | 698 | Feb 2008 | April 2009 | 2 years | June 2011 |
The Melbourne InFANT Program
38
| Australia (VIC) | 559 | April 2008 | March 2009 | 1.5 years | Nov 2010 |
POI NZ39
| New Zealand | 400 | March 2009 | Dec 2010 | 2 years | Dec 2012 |
Acknowledgements
Others who have made significant contributions to the protocol, funding applications and the coordination of the EPOCH Collaboration include: Lyra Butler, Kylie Hunter and Andrew Martin.
Competing interests
Several authors on this manuscript (see below) are Chief Investigators of the individual trials included in the EPOCH Collaboration. These include Healthy Beginnings [
36] (LB, CR, LW), Nourish [
37] (LD, AM), The Melbourne InFANT Program [
38] (KC, KH) and POI.NZ [
39] (BT, RT) trials. SMs postdoctoral fellowship is funded by HJ Heinz.
Authors' contributions
This protocol document was developed by the authors in consultation with members of the EPOCH Collaboration as listed below. All members of this group were sent draft versions and invited to comment and contribute changes. All authors have read and approved the final manuscript.
Members of the EPOCH Collaboration who contributed directly to this manuscript are: LA, LB, KC, LD, KH, AM, SM, CR, JS, BT, RT, MV and LMW.