Current IssuesNo.50
May 2007
 
     

Obesity Wars
Tim Olds & Jo Stevenson

 

First published in Sport Health, Volume 24 – Issue 4, Summer 2006-07, Sports Medicine Australia page
6-10.
Perhaps the only thing increasing faster than the prevalence of childhood obesity is the rate at which we are talking about it. According to LexisNexis, the number of articles dealing with obesity which have appeared in major world dailies increased fivefold between 1999 and 2003. Concurrent with increases in the number of media reports of childhood obesity, there has been a tremendous upsurge of academic interest in the subject. A scan of the PubMed database shows that the number of articles appearing with the words "child(hood)" and "obesity" as search terms increased from an average of 389 between 1982 and 1993, to 1200 per annum between 1999 and 2003.
The volume of commentary on childhood obesity has generated a lot of background heat and noise, and has created a polarising and sometimes vituperative debate. Issues surrounding the causes of the “obesity epidemic”, who is “responsible”, and the best ways to deal with it have been vigorously debated in media, political and academic forums. Schisms have emerged between those who advocate individual responsibility, such as Australian Health Minster Tony Abbott, and those who argue that interventions must be at the level of the entire built and social environments. Dietitians blame an increased intake of energy-dense foods, while exercise scientists point to declines in physical activity and increased sedentary behaviours. This article aims to present available evidence on historical trends in children’s fatness, fitness, food intake and energy expenditure. The picture which emerges in not as clear-cut as many would like us to believe, nor are the solutions obvious.

1 Trends in the fatness of children
There is absolutely no doubt that children are getting fatter, no matter how fatness is measured. We have recently gathered data on the skinfold thicknesses in young people aged 0-18 years, covering a total of 116 studies and including data on 364,077 young people from 24 developed countries. Percentage body fat (%BF) was estimated using the Slaughter equations. There have been increases in triceps and subscapular skinfold thicknesses, at the rate of 6-8% per decade since the 1950s. Estimated percentage body fat has been increasing at the rate of 1% body fat per decade (equivalent to a relative increase of 6-7% per decade). Your children are therefore likely to be about 20% fatter than you were when you were their age. There are also distinctive patterns in the ways in which children are getting fatter.

1.1 Boys are getting fatter faster than girls
The ratio of the median percentage body fat of boys to that of girls has risen from 58% to over 75% between 1950 and 2005, at an average rate of over 2% per decade. At this rate, boys and girls will be equally fat in the year 2111.

1.2 The fatter are getting fatter faster
There has been an increasing positive skew in the distribution of subcutaneous fat thickness in children. We compared the distribution of triceps skinfold thicknesses of Australian children from the 1985 Australian Health and Fitness Survey, and data collected using identical methodologies between 1997 and 2002. The leanest children today are as lean as their counterparts 15-20 years ago. However, there are more fat children in the recent cohort, and these children are much fatter than their counterparts in 1985. The gap between the "fat-rich" and the "fat-poor" is increasing, and there are more "Rupert Murdochs of adipose tissue".

1.3 Body shape is changing
The distribution of fat on the body, as indexed by the triceps:subscapular (T:S) ratio, has become more centrally located since the 1950s. Fat is moving from the arms and legs to the trunk. Waist-hip ratio (WHR) is also increasing, even in children from different cohorts matched for fatness. Australian children aged 10-12 years tested in 1985 and 1997-2004 were matched for age, sex, height, weight, body mass index (BMI) and triceps skinfold thickness. Even in this group matched for overall fatness, waist girths and waist-hip ratios were significantly greater in the children from 1997-2004. Boys' WHRs had increased from 0.85 to 0.88, and girls' from 0.82 to 0.87 (Dollman & Olds, 2006). These trends describe very unfavourable changes in the body composition of young people, foreshadowing a potential increase in the incidence of cardiovascular and metabolic disease. They also suggest that BMI does not tell the whole story.

2 Tends in children’s aerobic fitness
At the same time as children’s fatness has been increasing, fitness performance has been declining, particularly on tests of aerobic fitness. We have collected data on the fitness performance of over 50 million children from more than 50 countries, reaching back almost 100 years. From at least 1960 until 1970, there was a clear improvement in aerobic performance, at which point it plateaued and started to fall. It is now declining at the rate of about 4-5% per decade. Your children are likely to be 10-15% less fit than you were when you were their age (Tomkinson, Léger, Olds & Cazorla, 2003). Furthermore, Australian children are nowhere near as fit as their northern European counterparts. The fittest children in the world are in Estonia and Iceland, and the least fit in Singapore and the United States (Olds, Tomkinson, Léger & Cazorla, 2006; Figure 1). Changes in other types of fitness performance — strength, speed and power — have been less marked, with trends towards recent declines.


Figure 1. Performance in a putative race over 1600 m between children from different countries. The fittest children are from Northern Europe (Estonia and Iceland), and the least fit from the USA and Singapore. Australian children finish in the middle of the field.
It would not be surprising to find that decrements in aerobic test performance are largely due to increases in fatness — children are carrying much more ballast on those running tests. However, it turns out that only about 50% of the decline in performance is due to increased fatness. We tested this by again matching children from 1985 and 2000 for fatness (sex, age, BMI and skinfold thickness), and comparing their run performances. Even when matched for fatness, children in 2000 performed worse than their counterparts in 1985. So other factors appear to be involved in declining performance — lower activity levels, less familiarity with maximal exertion, perhaps a cohort effect when competing against generally less fit kids.

3 Trends in children’s food intake
Johnson: He eats too much, sir.
Boswell: I don’t know, sir. You will see one man fat, who eats moderately, another lean, who eats a good deal.
Johnson: Nay, sir, whatever may be the quantity a man eats, it is plain that if he is too fat, he has eaten more than he should have done. Boswell. The Life of Johnson (1791).

One thing which is certain is that people will only get fatter if there is an energy imbalance — too much energy in, too little out. There is strong prima facie evidence to suggest that children are eating more, and that what they eat is higher in energy, sugar and fat, and lower in fibre and micronutrient density. The number of fast-food restaurants has been increasing exponentially in Australia. In Adelaide, for example, the number of KFCs, Hungry Jacks, MacDonalds and Pizza Hut restaurants rose from just two in 1972 to 60 in 1997. Portion sizes have been increasing. Between 1987 and 2005 the number of kilojoules in standard portions of flavoured milk, cup cakes, scones, soft drinks and dinner rolls has on average doubled. In the last 50 years, the volume of Coca-Cola bottles has also increased dramatically: from 200 ml to 675 ml for the standard size bottle, and from 375 ml to 1 L for the king size bottle.
However, it is easy to be misled by anecdotal evidence. In an attempt to address the question of whether children really are eating less than they used to, we attempted to locate every study ever done on the energy intake of children. Thus far, we have located about 200 articles dating from the mid-1800s. They come from 24 countries classified by the World Bank as "high-income" or "upper-middle income". They are all countries of Europe or the developed Pacific Rim. The studies provide data on over 250,000 children and young people aged between 0 and 18, collated into 1800 age X sex X country slices. Surprisingly, these data clearly show declines in reported energy intake for both boys and girls across most age bands since 1955 (Figure 2).

Figure 2. Trends in reported energy intake of young people aged 2-18 in developed countries, and in Australia. Energy intake is expressed as a percentage of average intakes in 1990.

When analysed at the age X sex X country level, the data also show declines in reported energy intake over the period 1950-2000 in 75% of cases. Most declines are in the 3-6% per decade range. All countries show overall declines, ranging from 1.8% per decade for France to 7.9% per decade for Germany. The data can be combined across all countries using special statistical techniques. The overall average rate of decline is about 4% per annum. If these reported data reflect actual intakes, it is likely that your children will be consuming about 10% fewer kilojoules than you did when you were their age.

3.1 Under-reporting
These results are very surprising, and we should be systematically skeptical. What else could explain these changes? One possibility is under-reporting. Most people under-report energy intake, either because they forget what they have eaten, or because they don't want to appear to be gluttonous. We know this because there are objective ways of measuring actual food intake (for example, through observation or — most commonly — doubly-labelled water). A number of studies have compared actual and reported food intakes. We have reviewed 68 such studies, covering a total of 7350 subjects.
Under-report increases with age and percentage body fat. The latter is a concern when we are trying to decide whether children are eating more, because children have become fatter over time. This may mean that the apparent decline in children's energy intake is in fact merely a reflection of increasing under-reporting associated with increasing fatness. To correct for this, we would need to know
(1) the relationship between percentage body fat and under-report, and
(2) the historical trend in the percentage body fat of children.
Fortunately, we do have data on both of these relationships. In children, under-report increases by about 1.4% with every 1% increase in body fat. Percentage body fat has been increasing at the rate of about 1% per decade in children, so the nett effect of fatness-related under-reporting would be to reduce the calculated rate of decline in energy intakes from about 4% to about 2.5%.
We don’t know whether under-reporting really does increase with fatness over time (as opposed to within a cohort). There may be other factors which could explain the apparent decline in intake, but we don’t have good data. A common objection is that it is not valid to combine data from studies which use different collection methodologies, sampling frames and food databases. However, methodological differences will only be relevant if (a) the different methods result in systematically different estimates of energy intake, and (b) the methods show time-related variability (e.g. if one method were consistently used in the early years, and another more recently). If these conditions are not met, then methodological differences will only serve to increase the scatter or variability of the data, and will not bias trend estimates.

3.2 Trends in the consumption of specific types of food
It is often suggested that increases in fatness can result from increases in the consumption of particular types of food, such as fatty foods, high glycaemic index (GI) foods, soft drinks or confectionery. It is not at all clear that this can happen independently of increases in overall energy intake, but it is interesting to look at trends in intakes of some macronutrients and specific foods.

3.2.1 Dietary fat
Data were collated from 1128 reports (at the age X sex X country level) of fat intake of children, expressed as a percentage of total energy intake. The data come from 180,472 children from 23 high-income countries, and were gathered between 1920 and 2003 using self- or proxy-report. The pattern was quite clear: fat intake peaked about 1965, at close to 40% of total energy intake, and fell thereafter, to about 35% today (Figure 3). If self-reports are to be believed, children's diets today contain less fat than they did 40 years ago. Similar patterns have been found for adult Americans. This may seem incomprehensible in the light of media reports of over-consumption of fat-rich fast food. However, it is easy to forget the increased consumption of rice, pasta and other “multicultural” foods; increasing availability of low-fat milk, cheese and other dairy products; and year-round availability of fresh fruit and vegetables.

Figure 3. Trends in reported fat consumption in young people aged 2-18, expressed as a percentage of total kilojoule intake. Each point represents a report at the age X sex X country level.

3.2.2 Soft drinks
Recently, much media attention has turned to soft drinks as the culprit in increasing pediatric obesity. Various mechanisms have been proposed, for example the low GI associated with corn syrup sweeteners. In the US, there has been a rapid increase in soft drink consumption amongst adolescents. However, data on the relationship between intake of soft drinks and overweight status are far from unequivocal. In Australia, apparent per capita soft drink consumption across the whole population rose rapidly from the 1970s until 1993, but has been relatively stable (at about 110-115 L per annum) for the last 12 years. The largest study ever done of the eating and activity patterns of children is the Heath Behaviours in School Children (HBSC) study, a multi-national study conducted in 34 (mainly European) countries. It asked a total of 137,953 children aged 11, 13 and 15 about their dietary, physical activity and sedentary behaviour patterns, and their height and weight (Jannssen, et al., 2005). The authors attempted to find out which characteristics were best correlated with overweight and obesity. In 30 of the 34 countries, there was no relationship between soft drink consumption and overweight. In two countries (Belgium and Israel), there was a negative association (fatter kids consumed less soft drink). In only two countries (the Ukraine and Wales) was there a positive association between soft drink consumption and overweight.

3.2.3 Confectionery
The HBSC study also looked at the relationship between confectionery (sweets and chocolate) intake and overweight. Here the results were even more surprising: in 31 of the 34 countries, there was a significant negative correlation between intake of sweets and overweight. There are several possible ways of looking at these data. First, kids with a high intake of sugar may have a low fat intake. Second, overweight kids may deliberately restrict their intake of sweets to control their weight. It is possible that overweight kids will under-report their intake of "bad" foods. It is also possible that overweight use have larger portion sizes, as only the number of portions was recorded. Finally, it is possible that more active kids eat more confectionery, because they are more active, and need more energy. Once again, the data are not unequivocal, and support many possible interpretations.

4 Trends in children’s activity patterns
“Physical exercise, Epigenes, is of considerable importance for health. Its predominance over food was established in the past by the best philosophers and doctors.”
Galen. The exercise with the small ball. c AD170.
Unfortunately, there are no good serial data on how children use their time. A wide variety of incommensurable instruments have been used over the last 100 years to capture children’s activity patterns.
We do have good data on what Australian children do today. Using a computerised use-of-time instrument, the Multimedia Activity Recall for Children and Adolescents (MARCA), we have gathered over 15,000 24 h recall profiles. These data allow us to map the breakdown of moderate-to-vigorous physical activity (MVPA) in peripubertal children.
MVPA can be divided into four main categories:
(1) locomotion (38% of all MVPA time)
(2) sport (35%)
(3) play (22%)
(4) other (11%)
Locomotion involves mainly walking (88% of time), with cycling contributing 9% and new locomotions 3%. The new locomotions are skateboarding (60%), scooter and rollerblading (about 20% each). The most popular sports (in order) are soccer, basketball, Aussie Rules, dancing, cricket, swimming and netball. Nine tenths of play is outdoor play. Outdoor play divides fairly evenly into playground games and just "mucking around". MVPA which is neither locomotion, sport nor play involves either schoolwork or chores, in roughly equal proportions. Most chores (a distressingly small component of children’s overall time budgets) involve tidying the bedroom, with a smattering of garden work.

4.1 A window on the past
Occasionally we have data from very old studies on how children used their time in days gone by. In 1919 a young women, Miss E Bedale, was completing postgraduate studies in work physiology in England. As part of her work, she spent three years at a school in rural England, by happenstance called Bedales School. There she meticulously monitored the food intake and energy expenditure of the children, measuring energy costs using Douglas bags, and calculating the energy value of foods using a bomb calorimeter. The activity levels of the children are truly surprising. The average daily physical activity level (PAL) of these children was generally over 2 (i.e. they required on average more than twice their basal metabolic rate). This compares to average values of about 1.6 for children of a similar age today. If the values Miss Bedale measured were typical of children of those times, then we have seen a reduction of 25% in overall energy expenditure, and a reduction of 40% in energy expenditure over and above resting values.

4.2 Active transport
One area in which good serial data exist is the use of active transport (walking and cycling) to school. Here studies from the US, Canada, the UK, New Zealand and Australia have shown systematic declines, at the rate of about 2% per annum. Each year, another 2% of children decide to get a lift with Mum or Dad rather than walking or riding (Harten & Olds, 2004). The main factor determining whether a child (or adult) will use active transport is the distance to destination. The probability of using active transport decreases with the log of distance. Almost every child will walk up to 200 m; almost no child will walk or ride 4.4 km (Figure 4). Trends towards declines in active transport are likely to be exacerbated in the future, as the distance between home and school, and home and the shops, increases. In the UK, for example, the average distance from home to school increased from just over 2 km to over 3 km between 1985 and 1998. At 2 km, about 25% of children will choose to walk or ride to school; at 3 km, fewer than 10% will.

Figure 4. The relationship between the probability of using active transport (Y-axis) and the distance to destination (X-axis) in 136 10-13 year old South Australian children. At a distance of 1 km, about 46% of children will walk or ride. All children will walk 200 m, but no children will actively travel 4.5 km.

4.3 Levine's solution
Perhaps we are focussing too much on moderate-to-vigorous physical activity in trying to understand how energy expenditure has changed. American researcher James Levine argues that non-exercise activity thermogenesis (NEAT) — the energy we expend by standing, fidgeting, shifting about — plays an important role in energy balance. He measured the time spent lying down, sitting and standing (including moving about) in 10 lean and 10 obese self-declared "couch potatoes". The obese volunteers spent 164 more minutes sitting each day, and 152 minutes less standing/walking each day, than the lean volunteers. This amounted to an energy difference of 1470 kJ each day. There was a strong inverse linear relationship (explaining 52% of the variance) between the amount of movement (determined by accelerometry) and percentage body fat determined by DXA.
To determine whether obesity led to differences in posture allocation, or vice versa, Levine then had his volunteers gain or lose weight (8 kg and 4 kg respectively). Weight gain and weight loss made no difference to posture allocation, suggesting that there are biological mechanisms driving these behaviours. In an attempt to apply these findings to the childhood obesity problem, Levine has devised a "classroom of the future" where there are no chairs. Children stand for several hours each day at benches containing laptops. There are cushions on the floor where they can sit occasionally. They may be following such "role models" as Donald Rumsfeld, who claims to stand for 10 h each day.

5 Trends in sleep
There is mounting evidence that sleep deficits (low quantity, poor quality) are associated with obesity. This has been shown in studies on French and German 5-6 year olds, and American 11-16 year olds. The risk of obesity increases as sleep time decreases, rising to a fivefold increase in risk among those groups sleeping the least. In the German study, obesity increased in a dose-response fashion as sleep time decreased. In a recent analysis of the 1985 Australian Health and Fitness Survey data, similar trends were found in boys but not girls. There are a number of possible mechanisms, including disturbances to appetite-controlling (leptin and ghrelin) and glucoregulatory (cortisol) hormones. Alternatively, it may be that obesity leads to poorer sleep due to disorders such as sleep apnea, or that some third factor (perhaps television viewing) predisposes both to obesity and to poorer sleep.
We recently compared self-reported sleep time in 10-15 year old South Australian children in 1985 to their counterparts in 2004. Among girls, there was a reduction in total sleep time of approximately 28 minutes, while boys slept for 33 minutes less in the latter survey. Almost all of the reduction in sleep time is attributable to later bed-time. The reduction in sleep time was more pronounced in low SES boys (44 min) than high SES boys (23 min). There is, as far as we know, only one other study of secular trends in children's sleep, in Swiss children. Here too there were reductions in sleep time, though not as pronounced as in the Australian sample.

6 Is it such a bad thing to be overweight?
“The data linking overweight and death, as well as the data showing the beneficial effects of weight loss, are limited, fragmentary and often ambiguous.”
Editorial, New England Journal of Medicine (1998).

It is popularly believed that fatter people have reduced life expectancy as a result of a greater likelihood of suffering from metabolic and cardiovascular disorders. In 2005, researchers at the Centres for Disease Control published an analysis of data from the NHANES surveys which challenged this belief. The data suggested that overweight people (BMI 25-30) actually live longer than people of normal weight (BMI 18.5-25). Relative to people of normal weight, the risk of death of overweight individuals was less than 80%. Furthermore, there has been an historical trend for the relative risk of death in overweight and moderately obese individuals to decrease since the 1970s. These results caused a great deal of surprise amongst public health researchers, but a number of researchers, among them Glen Gaesser and Ancel Keys, had previously questioned the link between moderate levels of overweight and increased mortality and morbidity.
These surprising findings may be due to the greater likelihood of targeted interventions for this group — greater use of statins and anti-hypertensives, for example, or more lifestyle interventions. However, they do show that the notion of a “natural” ideal BMI is untenable: mortality curves are specific to a particular society, which chooses to treat certain groups in certain ways. They also suggest that while real risks exist for the grossly obese, moderate overweight may not pose such a public health problem.

7 Conclusion
“Better to walk in darkness than to follow a false light.”
Voltaire

The received wisdom about the aetiology of childhood obesity is not unequivocally borne out by the evidence, and both media and academic sources are often abusively selective in their use of the literature, not infrequently with political ends in view. There is no doubt that children are getting fatter. It is likely that this is associated with reduced aerobic performance. An objective look at the evidence doesn’t allow us to conclude with confidence that the historical reason for increased fatness has primarily been increased energy intake, or alternatively decreased energy expenditure. If energy expenditure has decreased, it may not be primarily due to reduced sport and play, but rather to increased sedentarism. Whatever the historical pattern of change, however, there are not necessarily any lessons here for interventions. Even if children have become fatter mainly due to decreases in energy expenditure without concomitant increases in intake, it may be impossible to recreate the social and built environments which made higher levels of energy expenditure possible or necessary decades ago. Furthermore, it would be a preposterously narrow point of view which would ignore the benefits of a healthy diet unrelated to obesity. So both energy intake and energy expenditure should remain as elements of any intervention strategy. In the meantime, politicians, the media and academics should be much less confident about ascribing blame to individuals, parents, corporations or governments.

References
Dollman, J., & Olds, T.S. (2006). Secular changes in fatness and fat distribution in Australian children matched for body size. International Journal of Pediatric Obesity (in press).
Flegal, K., et al. (2005). Excess deaths associated with underweight, overweight and obesity. JAMA, 293(15), 1861-1867.
Harten, N., & Olds, T.S. (2004). Patterns of active transport in 9-12 year old Australian children. Australian and New Zealand Journal of Public Health, 28(2), 167-172.
Jannssen, I., et al. (2005). Comparison of overweight and obesity prevalence in school-aged youth from 34 countries and their relationships with physical activity and dietary patterns. Obesity Reviews, 6, 123-132.
Olds, T., Tomkinson, G., Léger, L., & Cazorla, G. (2006). Worldwide variation in the performance of children and adolescents: An analysis of 109 studies of the 20 m shuttle run test in 37 countries. Journal of Sports Sciences, 24(10). 1025-1038.
Tomkinson, G.R., Léger, L., Olds, T., & Cazorla, G. (2003). Secular trends in the fitness of children and adolescents 1980-2000 — an analysis of 20 m shuttle run studies. Sports Medicine, 33(4), 385-400.


Contact
Prof. Dr. Tim Olds
School of Health Sciences
University of South Australia
Adelaide
AUSTRALIA
Email: Timothy.Olds@unisa.edu.au





http://www.icsspe.org/portal/index.php?w=1&z=5