Obesity can be defined as a pathological excess of body fat 1289. However, fat is not homogeneous, and abdominal fat has emerged as most clinically relevant 10. Differences between males and females in adult body fat have long been recognized, the android, male, fat pattern represented by relatively greater fat in the upper body, the gynoid, female, pattern by relatively greater amounts in the hip and thigh areas 11. Body fat distribution is partially hormonally controlled, and differences between males and females in fat mass have generally been considered to become manifest during puberty 112. However, recent studies of pre-pubertal children have also found higher levels of fat and evidence of the gynoid pattern among females at much younger ages 13. Studies which have assessed the degree of body fat associated with significantly elevated cardiovascular risk factors in children and adolescents, suggest that health risks become evident at lower levels for males than females 13.

Given its association with health risks, it has been suggested that there should be a greater emphasis than is current on the assessment of fat, particularly abdominal fat, for both clinical and epidemiological purposes 14. However, the most accurate measures of body fatness are generally expensive, require laboratory conditions and/or are associated with reduced participant acceptability. Most epidemiological studies therefore rely on anthropometry, which can include measurements of skinfolds, or of waist and hip circumferences representing abdominal fat. The most frequently used measure is weight in relation to height, usually represented by the body mass index (BMI), defined as weight (kg)/height squared (m²) 13.

BMI has, despite many recognized limitations 15, been described as crucial to obesity classification and monitoring 14. Given appropriate cut-offs, it can be used to classify obese children and adolescents with high specificity, thus identifying few false positives 16. However, BMI varies according to both gender and age, and much less is known about specific BMI levels associated with health risks among children and adolescents than among adults 8. BMI-for-age gender-specific reference charts with recommended cut-offs defining obesity have been developed in a number of countries. In addition, the International Obesity task Force (IOTF) has produced charts based on pooled international data using centile curves based on the recommended definitions of adult overweight and obese, aiming to compare across populations and employ a consistent definition throughout the lifespan 171819. Although the majority of studies adopt one or other of these definitions, alternative cut-offs continue to be used 20, but with decreasing frequency. The range of definitions means that it is often difficult to compare obesity rates between studies, because the methods used impact on the proportion of children defined as obese 1321.

Table 1, which shows obesity rates at several time points and in a large number of different countries, demonstrates the impact of different definitions.

Despite different definitions, the wide range in prevalences between countries is clear. This was also seen in a study of European, Israeli and US adolescents, who self-reported height and weight in 1997–98. These data are not shown in Table 1 since the definition was not 'standard', being the 95^th BMI centile for age and gender derived for the whole sample. Overall rates were therefore around 5%, with the lowest rates in Lithuania, and the highest in the United States 22. Very similarly, a comparison of obesity prevalence, based on the IOTF definition among adolescents from 34 countries who self-reported height and weight in 2001–02 found the lowest rates in Lithuania and Latvia, and the highest in Malta and the United States. This study noted particularly high prevalences in countries in North America, Great Britain and south-western Europe 23. A review of surveys conducted within Europe during the 1990s, suggested higher levels of childhood overweight and obesity in southern countries and lower levels in the central and eastern countries which experienced political and economic transition over that period 24. Within developing countries, high prevalences are found in the Middle East 25.

Returning to Table 1, a further issue demonstrated is that of globally increasing rates, regardless of definition. Obesity has been described as the most common pediatric disease in most of the world, with the exceptions of sub-Saharan Africa and the former Soviet Union 26. Within Europe, the prevalence rates of overweight and obesity are not rising at a constant rate, but are accelerating 27. In respect of secular changes, what Table 1 cannot show is that in addition to increases in obesity rates, the distribution of BMI has become more skewed, with the greatest increases at the highest levels 2829. There is also evidence of changing body composition, with increasing fat and reductions in muscle 3031.

In respect of gender, Table 1 suggests no consistent male or female excess. Differences are generally small, and there is no suggestion that either males or females consistently predominate within particular age groups, or according to particular definitions. Although international patterns are, in general, also difficult to discern, Table 1 suggests that within Asia, a male excess may exist in Korea, Taiwan and China, in contrast to a female excess in Middle Eastern countries. A proposal that the male excess in China may result from gender differences in child care associated with the one-child policy 32, has been challenged 33. An apparent female excess in some UK studies using the IOTF definition is artefactual, since this definition has been shown to have lower sensitivity in UK males than females 34. The study which derived 95^th BMI centiles for age and gender within its international adolescent sample found higher rates among males in some countries, while in others there was a female excess. However, in almost all cases the gender differences were not statistically significant 22. Similarly, the review of surveys conducted within Europe in the 1990s found the number of countries with higher prevalences for overweight and obesity among females was almost the same as that with higher prevalences for males 24. It is possible that gender differences may emerge in future; a study conducted in the US found significantly greater increases in rates among males between 1986–98 35. It has therefore been suggested that prevalence estimates for both males and females should always be presented 2.

While gender differences may be small overall, patterns may differ according to ethnic group. Among females, British studies almost always find the highest rates of obesity among black females, with the lowest levels generally found among South Asians 3637. Among a sample large enough to sub-divide into black African and black Caribbean groups, one UK study of 11–13 year olds found higher BMIs among both groups than in their white, Indian, Pakistani, Bangladeshi or mixed ethnicity peers 38. These studies have generally found smaller, and less consistent ethnic differences in obesity rates among males. However, a US study found the same pattern among both males and females, the highest rates occurring in Mexican Americans at ages 6–11, and Non-Hispanic blacks at ages 12 to 17 39. Different results in respect of rates according to ethnicity by gender in studies conducted in the UK and US may result in part from the different groups included; there are few US data on the health and weight status of Asian Americans 40.

In respect of socioeconomic status (SES), studies suggest higher rates of obesity among low income groups in richer countries, and high income groups in poorer ones 141. It is suggested that this is because in developing nations, higher SES individuals have become globalized, with easy access to relatively cheap, calorie-dense foods, while those of lower SES remain localized and undernourished 42. Within the US, increases have also been greatest among children and adolescents from the lowest income families, so increasing SES disparities 35. However, not all studies in developed countries find SES differences 364344, and among those that report separately, there is some evidence that SES differences may be clearer among females than males 3745.

Like height, weight is heritable. One recent review suggests that twin and adoption studies point to a genetic contribution for BMI of 40–70% 46, while a more extensive, but earlier, review of familial resemblance suggests that genetic factors explain 50–90% of BMI 47. The results of genetic studies, where presented separately for males and females, appear broadly similar. Overall, the findings of such studies mean that genetic factors determine individual susceptibility to gain weight. Such 'thrifty' genes provide an evolutionary advantage in times of famine, when humans have to stockpile energy to survive, but a disadvantage when food is plentiful 4849. However, as several authors point out, while the propensity for obesity may have existed for a long time, the recent rapid rise in rates demonstrates the central role of environmental factors 464750.

Obesity is related to an imbalance between energy input and output, the size of which may be very small if over a long period 51. One review suggests that in children, an imbalance of around 2%, which is the equivalent of around 30 calories or 15 minutes of tv instead of play a day, may lead to obesity 52. Behavioral determinants therefore include excess energy intake and/or inadequate energy expenditure 53, although the emphasis given to these 'Big Two', and the neglect of other plausible contributors to the secular increase in obesity has recently been questioned 54.

In respect of energy intake, dietary surveys do not suggest a secular increase among children and adolescents. However, the results of such studies may be confounded by an increasing trend towards greater under-reporting, to the extent that reported intake may be below the estimated required physiological minimum, especially among older girls. Regular consumption of high energy-dense fast-foods and sugary drinks which are associated with less satiation and so insufficient compensation via subsequent reductions in intake, increased portion size, eating outside the home and snacking have been particularly implicated in promoting weight gain. This is especially the case among older children, who are less influenced by biological cues of satiety 5055. In respect of energy expenditure and physical activity, assessment is difficult 56, and evidence for secular trends is scarce because of the absence of suitable baseline data 57. However, a decline in UK adolescent energy intake from the 1930s to the 1980s with no concurrent change in body mass, increasingly pervasive electronic and screen-based entertainment, greater car travel and reduced walking or cycling, together with evidence of reductions in fitness in developed countries all point to lower activity levels 57585960.

Gender differences in many areas of nutrition emerge in childhood and adolescence. The greater fat-free mass of males, particularly post puberty, requires a higher energy intake. Thus the US Government suggests approximate daily requirements for moderately active 4–8 year olds as being 1,500 kcal for both males and females, compared with 2,600 kcal for males and 2,000 kcal for females aged 14–18 61. Studies of adolescents have found that females are more likely to pay attention to foods as a way to influence health and to meet nutritional recommendations, while males eat more fast foods. It has been suggested that these differences arise not only because of differences in Western societies' perception of ideal body weights, but also because some foods are gendered; for example meat may represent strength and virility. In addition, the menstrual cycle has been associated with cravings for foods rich in fat and carbohydrate 1262.

Gender differences in energy expenditure can also be attributed to both biological and social factors. Total energy expenditure (TEE) is strongly correlated with body weight, largely because of the relationship between resting energy expenditure and body weight 63. Differences in body composition at puberty are associated with higher TEE among males than females, however there is limited and inconsistent data on male-female differences in TEE pre-pubertally 6364. Of all the components of TEE, physical activity is most amenable to modification 50, and a review of prospective observational studies has concluded that increased physical activity and decreased sedentary behavior protect against relative weight and fatness gains in childhood and adolescence 59.

It has been suggested that the clearest biological correlate of physical activity in children and adolescents is gender 65, and also that the effect of energy expenditure on the etiology of obesity may vary by gender, ethnicity or age 515263. Gender differences in physical activity begin in childhood. A longitudinal study found that while boys' daily TEE increased continuously between ages 5 and 10, that of girls increased from around 1,400 kcal at 5 to 1,800 kcal at 6, but by age 9 had reduced to 1,600 kcal. This reduction was explained by a 50% reduction in physical activity between ages 6 and 9 52. One review cites data suggesting that levels of moderate or vigorous physical activities in US 11–12 year old boys are nearly twice those of girls, while age-related decreases in activity levels in adolescence are higher in girls. Thus a 1990 UK study found decreases between ages 11 and 18 in mean self-report daily activity causing at least slight breathlessness from 83 to 57 minutes among males and from 49 to 26 among females. A review of the world literature suggested annual declines in physical activity of up to 2.7% among boys and 7.4% among girls between the ages of 10 and 17 566566. Such findings have been related to differences in motor skills development, body composition, socialization towards sports and physical activity and freedom to engage in activities independently outside the home 5665.

Sedentary behavior has also been associated with body composition and BMI. For example, children who watch more tv have been shown to have higher skinfold thicknesses, while one study has suggested that approximately 17% of early adult overweight may be attributable to watching tv for two or more hours daily in childhood (reported in 57). Given this, it is interesting that a review of studies of screen-based media use in contemporary youth, most from Europe or the US, found that males are also higher users of tv and video games. For example, 30% of males and 25% of females watched tv for more than four hours a day, while 30% males and 7% females played video games for more than four hours a week. This suggests that females may spend time in sedentary activities not measured by most current instruments 67.

Obesity clusters in families, and parental obesity is the most important risk factor for obesity in children 68. While implicating genetic factors, the environments of children and, to a lesser extent, adolescents, are also generally provided by their parents 6970. There is evidence that familial patterns of adiposity may be partly attributable to similarities in diet. There is also evidence that self-regulation of energy intake by infants and children can be over-ridden by parental control, first among those who are formula fed and who may be encouraged to finish the bottle, then as well-meaning parents attempt to restrict children's eating. Gender differences have also been reported here: maternal restriction of snacks has been found to promote their over-consumption in an unrestricted setting by females, but not males; greater parental control has been linked to increased adiposity in females, but not males; mothers who use more control have been found to have daughters, but not sons, with poor energy regulation 697172. In addition, there is some evidence that parents are less likely to encourage sons to lose weight, perhaps because of the larger, more muscular ideal male body shape 73. Parents also appear to be strong influences on physical activity in childhood, and again, there is evidence of gender differences; for example, stronger relationships with parental activity for girls 6574.

Childhood physical health consequences were, until fairly recently, largely unrecognized. In fact, there are few organ systems that severe obesity does not affect 1. Associated outcomes include: cardiovascular risk factors, identified in children as young as 5 years; type 2 diabetes; non-alcoholic fatty liver disease; asthma; sleep-disordered breathing; systemic inflammation; and orthopedic problems. There is a marked tendency for obesity, particularly adolescent obesity, to track into adulthood, and so be associated with adult obesity-related disorders. However, even after controlling for current risk factors including weight, childhood obesity is associated with increased risk of all-cause and coronary heart disease mortality 12757677787980.

Some of the physical consequences of obesity are gendered, with some evidence that adult mortality risks are higher in respect of adolescent obesity among males 7576. There have been suggestions of a stronger relationship between obesity and asthma in females, but results in this respect are inconsistent 8182. Among females, obesity is associated with early onset of puberty and menarche which have, in turn, been linked to breast cancer, although the relationships are complex 83. Obesity has also been related to other cancers of the female reproductive system, increased risk of spontaneous abortion and menstrual problems, particularly polycystic ovary syndrome which can lead to infertility 11112496884. The relationship between obesity and female puberty led to the suggestion that a threshold level of fatness is required for the female growth spurt and menarche. This is not now thought to be the case 8586; indeed, there is a growing body of evidence suggesting the opposite, that puberty effects levels of fatness 15.

Links have also been suggested between increasing levels of obesity, its association with female puberty and claims of recent reductions in age of menarche. Reviews conflict, some suggesting 'clear evidence' that the age of menarche is falling 87 and others that while menarche may be stable, the first signs of puberty, such as breast budding may now be occurring earlier 8586. However, it may be the case that there has been no recent secular change in either, but rather that obesity is associated with premature development of pubic hair and apparent breast tissue separate from true puberty 88.

Since markers of puberty are more overt and recordable for females, there is less information about its timing for males 87. However, some reviews of obesity and puberty appear to extrapolate the female findings to males or children/adolescents in general 6885. In fact, although the evidence is much sparser, there is some evidence that in males, obesity is associated with later sexual maturation 11586.

The immediate psychosocial consequences of childhood and adolescent obesity have been long-recognized 89. Stigmatization and discrimination by peers is well-documented; from very young ages, obese children are characterized in negative ways, less preferred as friends and more likely to be the targets of teasing or bullying 1275. There is also evidence of bias and stereotyping by teachers and even some parents 90. While it could be argued that such behaviors might diminish as obesity becomes more commonplace, this does not appear to have occurred 1.

Evidence in respect of the psychological consequences of child and adolescent obesity is mixed 1. While some reviews conclude that it is associated with psychological or psychiatric problems, increasing with age and particularly among girls 78, others suggest that research has failed to find consistent differences in the global self-esteem 91, depression or anxiety 92 of obese or overweight children and adolescents compared with the rest of the population. Such conflicting results may have arisen in part because obese clinic samples tend to have poorer psychological well-being and quality of life than community samples. While obese community samples have lower body satisfaction and physical self-competence than non-obese, few are depressed or have low global self-esteem 9394. Small differences in their psychological well-being may also be explained, at least partly, by weight-related teasing or bullying 95. There have been suggestions of associations between depression and obesity, and that major depression in adolescents, or young females, predicts increased adult BMI 9596. However, the relationship between obesity and depression has been described as unproved 94.

Reductions in body satisfaction and well-being amongst the obese are greater among adolescents than children, and among females than males. Reflecting this, 12 of 18 recent studies of body dissatisfaction and 7 of 28 of self-esteem reported in one review included female participants only 94. The consensus is that gender differences in body satisfaction emerge around 8–10 years of age 73. While both males and females with a high BMI wish to be thinner, the picture is more complex for males because of the larger muscular male ideal. In relation to stereotypes and attitudes, some studies have demonstrated equivalent stereotyping of obese peers, regardless of gender. However others have suggested that females express stronger dislike of obese peers than males, and that obese females are rated more negatively than obese males. There is also some evidence that females may be more vulnerable to obesity-related victimization, but again, this is not a consistent finding 9097. The picture in relation to body satisfaction and obesity is also complicated by ethnic differences. For example, black girls with high BMIs are less likely than those from other ethnic groups to consider themselves overweight or desire to be thin 9098.

Gender differences in weight-control methods are also evident. Males are more likely to exercise, begin dieting at a higher BMI than females, and tend to focus on increasing upper body size while reducing fat 99. In contrast to professionally administered, 'sensible' weight loss programs, 100, the use of unhealthy quick-fix dieting practices by some adolescents has been linked to a range of negative physiological and psychological outcomes 101102. Although such behaviors are significantly more likely among females 102103, males are not immune to disordered eating behaviors 99104. The finding of one study that femininity as represented by self-ascribed expressive personality traits, was predictive of eating problems among both male and female adolescents 105, highlights the need to consider such characteristics within each gender, rather than just compare the behaviors of males and females.

Finally, to return to the longer-term consequences, longitudinal studies in both the US and UK have demonstrated that child and adolescent obesity is associated with adverse social and economic outcomes such as reduced years of education, income and marriage rates in young adulthood. Again, there are gender differences, these effects being stronger for females 12757678.