Key Findings from EarlyBird


EarlyBird is a cohort study, monitoring the causes and behaviour of insulin resistance in contemporary children. Insulin resistance, largely the result of obesity, is believed to underpin the high prevalence of diabetes and cardiovascular disease that characterises modern society.

EarlyBird is distinctive in combining objective measures of physical activity and body composition from the age of 5y with annual fasting blood samples, which reach beyond simple body composition (BMI and body fat) to metabolic health (glucose control, insulin sensitivity, blood fats, cholesterol, blood pressure).

The study is generating some novel and sometimes counter-intuitive findings, which have nevertheless withstood the rigours of peer review – over 60 papers are now in print. It has now accumulated eleven annual data sets, and retains over three-quarters of the original cohort. It is intended that the study continue until the youngsters reach 16 yrs of age. EarlyBird 2, a programme designed to repeat the measures when the metabolic impact of puberty and growth has ceased, is currently under discussion.


In Summary

Obese children – parents unaware and unconcerned. Today’s parents are oblivious of their children’s weight
Parents are key partners in the drive to halt obesity, but will have
little impact unless educated to recognise the problem

Children’s activity not determined by environmental opportunity Green spaces and sports centres do not influence the physical activity of children
Like most things biological, a child’s activity level seems to be ‘set’
by the brain, and therefore strongly defended against change

Social inequalities no longer a major factor in obesity. All children today are at risk, regardless of family income or postcode
Targeting by SES unlikely to be effective

Obesity leads to inactivity, rather than the other way round. Time-lag analysis from one year to the next suggests obesity comes first
Crucially important - calorie reduction, rather than physical
activity, appears to the key to weight reduction

Healthy weight for life? Start at birth. Most excess weight is gained before the child ever starts school
School initiatives are probably too late. Mums should be
encouraged not to overfeed low birth weight babies, who are
most at risk of later weight gain

Average pre-pubertal child no heavier than 25 years ago The rise in obesity is confined to a small group of children who are behaving differently from the majority who have not changed in a generation
Only at puberty does the whole childhood population get involved.

There may be no general childhood obesity epidemic before puberty

Taller children are fatter. Children who gain excess weight are taller and more insulin resistant than their shorter peers.

Taller does not always mean healthier.

Obese parents, in particular those of the same gender, key to childhood obesity. Daughters of obese mothers are 10 times more likely to be obese than the daughters of normal weight mothers, and the sons of obese fathers six times.
It may be more effective to target the obese parent than the
obese child

Girls at greater risk of type 2 diabetes than boys. Girls are intrinsically more insulin resistant than boys, which could explain why more girls get type 2 diabetes in childhood
Scarce resources might be better targeted at girls

Health risks of childhood obesity unrelated to parents. Although a child's height and weight are related to their parents', the health risks for obese children are much the same, whether or not the parent is obese.

The diabetic risk for obese children derives exclusively from their
own weight gain, and is thus potentially avoidable.

Type 1 and Type 2 diabetes essentially the same disorder of insulin resistance, differing only in tempo.
Keeping weight down should help prevent, or at least delay, the
onset of both type 1 and type 2 diabetes

In Detail

If obesity underpins insulin resistance, three questions emerge: Who, When and Why? Which children develop obesity, when does it start, and why do some, but not others?


There is no longer an association between low birth weight and later diabetes: Diabetes risk in contemporary children lies with excess weight gained in early childhood and yet parents are still encouraged to feed up low birth weight babies. This raises an important public health issue - the importance of keeping weight gain low in children of low birth weight (Wilkin TJ -Diabetes 2003).

Girls are at greater risk of type 2 diabetes than boys: Girls are intrinsically some 35% less sensitive to insulin than boys and childhood type 2 diabetes is much commoner in girls. Young overweight females may be more important to target than males (Murphy MJ - Pediatrics 2004).


Obesity starts early: All children gain weight during growth, but EarlyBird is interested in the excess gained. There is much concern about school meals, PE time, after-school clubs, television viewing, computer games and green space in the cause of childhood obesity - all of them primary school issues. However, EarlyBird finds that over 90% of the excess weight in girls, and over 70% in boys, is gained before the child ever gets to school age. These findings support a need to re-direct public health initiatives towards an earlier period in childhood. (Gardner DS-L - Pediatrics, 2008).

There may not be just one childhood obesity epidemic, but two (or more) The widely-held belief that the entire childhood population is at risk of obesity may not be correct. The trends imply that, before puberty, only a subgroup of the population is at risk. With puberty, the skew in fatness that characterises early childhood gives way to widening of its variance, suggesting a new wave of obesity which this time involves most of the population. The changing pattern suggests that different factors at different ages are responsible for today's childhood obesity – some operating early in life, and others much later. More than one solution may be needed to prevent obesity in childhood (Perez-Pastor Int J obesity 2009, Mostazir in preparation)


Obese parents – in particular those of the same gender - have become a major contributory factor to childhood obesity: The average UK child is no heavier today than he/she was in 1990. The mean BMI has certainly risen, but the median hasn't changed. The "new fat" of the childhood obesity epidemic appears restricted to a subgroup of children - those whose same-sex parents are overweight or obese. Gender assortment of this kind suggests role-modelling. (Perez-Pastor E - International Journal of Obesity 2009)

Inactivity does not lead to obesity, rather obesity leads to inactivity Using time-lagged correlation to imply direction of causality, weight gain appears to lead to (precede) inactivity, rather than inactivity to weight gain. Crucially important – because it suggests that calorie reduction, rather than increased physical activity, may be the key to weight reduction (Metcalf BS Arch Dis Child 2010).

Taller children really are fatter Several studies suggest that taller children may be wrongly labelled as 'overweight' because body mass index (BMI) is not independent of height in childhood, and recommend adjustment to make it so. We used objective measures of %body fat and hormonal/metabolic markers of fatness to investigate whether BMI and the corresponding fat mass index (FMI) mislead in childhood, or whether taller children really are fatter. We have found that taller children really are fatter than their shorter peers, have higher leptin levels and are more insulin resistant. Attempts to render indices of body mass or fat mass independent of height in childhood seem inappropriate if the object of the index is to convey health risk. (Metcalf BS Int J Obesity 2011).

The secular trend in height among children may simply reflect their rising body fat Height in children correlates with BMI. Height is transmitted by both parents, and body fat largely by the same-sex parent, but the extra height associated with more fat in the child is unrelated to the height or weight of either parent. Excess fat is unhealthy, so the increasing height of contemporary children may not be healthy either. (Ajala O. Pediatric Obesity 2011).

The metabolic health of children is a function of their own weight gain, not that of their parents. There were large and significant differences in metabolic health between children categorized as normal weight or overweight/obese. At 13 years, for example, the metabolic risk score of the overweight/obese girls was 14-fold higher than that of the normal weight girls. However, there was no significant contribution from parental BMI or metabolic health.. while there were clear relationships between the metabolic risk score of the child and both his/her BMI. Diabetes risk appears to belong the child and is therefore potentially preventable (Ajala O Pediatric Obesity 2011 in press).

Age may be more important than stage in so-called 'pubertal' insulin resistance.It has long been believed, on the basis of cross-sectional studies, that the rise in insulin resistance that is thought to trigger the high rates of diabetes in early adolescence is due to puberty. EarlyBird's longitudinal data suggest that insulin resistance rises many years before puberty, and that body fat is more important than pubertal development. An important observation, because puberty is inevitable, but fatness is not. (Jeffery A Diabetes Care 2011 in press)

Parents unaware and unconcerned: Overweight is now perceived as the norm, and today's parents are no longer aware of their own or their children's weight. This report hit the world's media, because it revealed a serious issue for the campaign to reduce childhood obesity. Parents are essential partners in the struggle yet, crucially, they do not acknowledge the problem. (Jeffery AJ BMJ 2005).

Social inequality is not associated with physical inactivity: Despite clear evidence of a socio-economic gradient in sports club attendance among the children, EarlyBird can find no evidence for corresponding differences in physical activity. The assumption that children of lower socio-economic status suffer from their lack of structured opportunity for physical activity is not reflected in the evidence. Indeed, analysis suggests that poorer boys may be marginally more, rather than less, active than those who are wealthier. (Voss LD - Child Care, Health and Development 2008).

Social inequality is not associated with metabolic risk: Lifestyle interventions to improve health in young children tend to target areas of relative deprivation, but the evidence for so doing is largely historical. We have re-examined the link between deprivation, obesity and metabolic risk in contemporary UK children. Our data do not support the assumption that obesity, metabolic disturbance and thus risk of type 2 diabetes are more prevalent among less affluent children. In today's increasingly obesogenic environment, youngsters from all backgrounds appear to be vulnerable, with population-wide implications for public health spending, and the prevention of diabetes in contemporary youth. (Voss LD epub ahead of print 2010).

Dietary habits appear to be established early in life and are retained throughout childhood. Notwithstanding, more children appear to experience a deterioration in their diet over childhood than an improvement. Dietary choice is moderately systematic, but quality diminishes from 5y to 13y in 30% of the children and improves in only a few. Early assessment of dietary habits seems important to disease prevention. (Frémeaux A. Brit J Nutr 2011).

Changes in metabolic rate in relation to insulin resistance and weight gain. The importance of REE lies in its potential to influence weight gain. Before puberty, children appear to experience an increase in resting energy expenditure (REE) which is greater than can be predicted from changes in body composition alone. The extra REE is in part explained by insulin resistance, but not by the amount of weight gained, suggesting that there may be an increase in energy requirement prior to the onset of puberty. Although the role of REE in future weight change remains controversial, we have been unable to support the hypothesis that it lies with adaptive thermogenesis. (Hosking J J. Clin Nutr, 2010).

Obesity leads to inactivity, rather than inactivity to obesity: Using time-lagged correlation to imply direction of causality, weight gain appears to lead to (precede) inactivity, rather than inactivity to weight gain. Crucially important – it suggests that calorie reduction, rather than increased physical activity, may be key to weight reduction (European Congress on Obesity Amsterdam May 2009).

Resting energy expenditure does not appear to provide an explanation for childhood obesity. Resting energy expenditure is the largest component of total daily energy expenditure, but its contribution to the development of obesity, especially in young children, is controversial and largely unknown. We evaluated the association between REE at 7y and subsequent change in weight and body fat over the following three years, and found it had little impact on either. (Hosking J. Nutr Res 2011)

Children who keep active gain no less weight, but they do become metabolically healthier: The UK and US Governments advise at least 60 minutes moderate physical activity every day. Only 42% of the EarlyBird boys and 11% of the girls met this guideline consistently over the three-year period from 5-8yr. Governments use BMI as their outcome measure, but there were no differences in the trend for BMI in either sex, while the more active children became metabolically healthier. The study questions the utility of BMI as the outcome measure of physical activity programmes in children, and whether the bar for girls should be lowered because girls systematically record less physical activity than boys (Metcalf BS Arch Dis Child 2008).

Children's activity is not determined by environmental provision: Those who do less in school do more out of school, and end up doing the same overall. Less than 1% of the four-fold variation in physical activity among children can be explained by the five-fold variation in PE provided at school. The original report was based on a single school term (Mallam KM BMJ 2003), but we have recently extended the analysis to four consecutive school terms. The result was replicated, and we question Government policy of linking physical activity to recreational facilities (Frémeaux AE, Int J Obesity 2011).

Being driven to school may not be eco-friendly, but it does not reduce a child's overall activity The activity cost at the age of 7y of being driven to and from school during the hours 8am-4pm is 16%, but is nil (<0.1%) over an entire 24h. As in the schools study, those who lack the opportunity for physical activity at one period of the day appear to compensate for it at another (Metcalf BS BMJ 2004)

The Activitystat hypothesis The evidence we are accumulating suggests the activity of children is "programmed" - either genetically or as a result of very early experience. There is little evidence from EarlyBird studies that the physical activity of free-living children is linked to recreation or environmental opportunity. (Wilkin TJ Int. J Obesity 2006).

Can we modulate the physical activity of children? Contribution to a debate with Prof JJ Reilly (Glasgow) on the ability (or not) of interventions to increase the physical activity of children. The article explains the workings a control system, and distinguishes modulation (change in setting) from modification (change in output, but only for as long as the intervention lasts). It points to evidence of systematic variation from child to child, from place to place and from time to time, and argues that systematic variation of this kind implies internal control. Finally, it summarises wheel-running studies in rodents which clearly indicate heritable physical activity behaviour (Wilkin TJ Int J Obesity, 2010).

Girls are born lighter than boys: the fetal insulin hypothesis Birth weights are universally lower in girls. Birth weight is mostly related to growth during the third trimester of gestation and this, in turn, to levels of foetal insulin. The hypothesis proposes that the greater insulin resistance which renders females more susceptible to diabetes may also be responsible for their lower birth weight. Finding the sex-linked gene responsible could unlock a mechanism that controls insulin resistance, the basis for type 2 diabetes. (Wilkin TJ - Int J Obesity 2006).

• The Accelerator Hypothesis: Type 1 and type 2 diabetes are the same disorder of insulin resistance and differ only by tempo, largely determined by different genetic backgrounds

Weight gain is the missing link between Type 1 and Type 2 diabetes. The incidence of type 1 diabetes is rising as fast as type 2, and both are rising in parallel with obesity. Obesity leads to increasing insulin resistance which underlies both type 1 and type 2 diabetes. (Wilkin TJ Diabetologia 2001).

Support for the principal prediction of the Accelerator Hypothesis: Among children who develop Type 1 diabetes, those of higher body mass index (BMI) develop it younger - true acceleration. (Kibirige M Diabetes Care 2003; Editorial by Arlan Rosenbloom, same issue).

Further independent support for the Accelerator Hypothesis: This includes evidence that waist circumference is greater in children who develop type 1 diabetes. It satisfies another prediction that the age-adjusted BMI at presentation will rise as the environmental accelerator (body mass) increases over time. (Betts P Diabetic Medicine 2005; editorial by Dennis Daneman, same issue).

A closely reasoned plea, with a group of international opinion leaders, to consider lifestyle change or insulin sensitising drugs (as proof of principle) in the prevention/management of type 1 diabetes: Weight gain appears to be an important environmental accelerator contributing to the increase in type 1, as well as type 2 diabetes. Lifestyle intervention, clearly effective in type 2 diabetes, should now be trialled in type 1 before embarking on further trials using more aggressive forms of treatment such as immunotherapy or islet cell transplant. Insulin sensitisation (eg metformin) might first be considered as proof of principle. (Wilkin TJ Diabetes Care 2004).

A recent update reviewing progress with the hypothesis, expanding on the concept of tempo and addressing the critiques that have challenged it: There are now over 100 articles in the literature referring to or testing the Accelerator Hypothesis. The focus of the hypothesis remains where it started - with the concept of tempo as it relates to the loss of the insulin-secreting beta cells. (Wilkin TJ IntJ of Obesity 2009).

Type 1 and type 2 diabetes are the same disorder of insulin resistance set against different genetic backgrounds. Contribution to a debate with Prof Marian Rewers (Denver, USA) on the merits of the accelerator hypothesis. The clinical convergence in children of the two syndromes once known as type 1 and type 2 diabetes has caused considerable puzzlement to diabetologists and epidemiologists alike. The accelerator hypothesis points to the insulin resistance of childhood obesity as the load which accelerates the loss of beta cells and the onset of diabetes. Children with reactive immune response (HLA) genes respond fastest (Wilkin TJ Pediatric Diabetes).