Childhood obesity is an important public health problem, with a rapidly increasing frequency worldwide. Identification of critical periods for the development of childhood and adolescent obesity could be very useful for targeting prevention measures. Weight status in early childhood is a poor predictor of adult adiposity status, and most obese adults were not obese as children. We first proposed to use the body mass index (BMI) charts to monitor individual BMI development. The adiposity rebound (AR) corresponds to the second rise in BMI curve that occurs between ages 5 and 7 years. It is not as direct a measure as BMI at any age, but because it involves the examination of several points during growth, and because it is identified at a time when adiposity level clearly change directions, this method provides information that can help us understand individual changes and the development of health risks. An early AR is associated with an increased risk of overweight. It is inversely associated with bone age, and reflects accelerated growth. The early AR recorded in most obese subjects and the striking difference in the mean age at AR between obese subjects (3 years) and non-obese subjects (6 years) suggest that factors have operated very early in life. The typical pattern associated with an early AR is a low BMI followed by increased BMI level after the rebound. This pattern is recorded in children of recent generations as compared to those of previous generations. This is owing to the trend of a steeper increase of height as compared to weight in the first years of life. This typical BMI pattern (low, followed by high body fatness level) is associated with metabolic diseases such as diabetes and coronary heart diseases. Low body fatness before the AR suggests that an energy deficit had occurred at an early stage of growth. It can be attributable to the high-protein, low-fat diet fed to infants at a time of high energy needs, the former triggering height velocity and the latter decreasing the energy density of the diet and then reducing energy intake. The high-fat, low-protein content of human milk may contribute to its beneficial effects on growth processes. Early (pre- and postnatal) life is a critical period during which environmental factors may programme adaptive mechanisms that will persist in adulthood. Under-nutrition in fetal life or during the first years after birth may programme a thrifty metabolism that will exert adverse effects later in life, especially if the growing child is exposed to overnutrition. These observations stress the importance of an adequate nutritional status in childhood and the necessity to provide nutritional intakes adapted to nutritional needs at various stages of growth. Because the AR reflects particular BMI patterns, it is a useful tool for the paediatrician to monitor the child's adiposity development and for researchers to investigate the different developmental patterns leading to overweight. It contributes to the understanding of chronic disease programming and suggests new approaches to obesity prevention.
Although obesity, particularly childhood obesity, is increasing worldwide, no decisive evidence emerges to account for this trend. Factors such as nutrition in childhood and adulthood provide an inconclusive explanation for adult health outcomes. By contrast, it seems that risk factors occurring in early life could play a major role.1, 2 Identification of early markers of adult body adiposity is useful for clinicians to monitor the child fatness development and for researchers to investigate the origin of obesity. Childhood weight status is a poor predictor of adult adiposity and most obese adults were not obese as children.3 Other anthropometric parameters must be investigated. Age at adiposity rebound (AR), recorded on individual BMI growth curves, has been identified as an indicator predicting adult fatness.4 This indicator may help identify the critical periods of adiposity development and the determinants of future obesity and health risks.
Assessment of the age at adiposity rebound
Nutritional status during growth is currently assessed on the basis of weight and height. Weight for age and weight for height are recommended for young children by the World Health Organization.5 However, the first indicator ignores height and the second ignores age. Weight-for-height indices according to age include the three parameters (weight, height and age) simultaneously. As it is highly correlated with weight and body fat, and weakly related to stature, the weight/(height)2, Quételet or body mass index (BMI) has been selected to predict adiposity in children.6 In the early 1980s, the first BMI distribution reference charts,6 revised 10 years later,7 were constructed for all ages throughout childhood (Figure 1) and adulthood.7 This method is now used worldwide.
The development of the BMI during growth parallels the development of more direct measures of body adiposity such as skin folds. On the average, a rapid increase of the BMI occurs during the first year of life. The BMI subsequently declines and reaches a minimum around the age of 6 years, before beginning a sustained increase up to the end of growth. The point of minimal BMI value (the nadir of the BMI curve) is the start of the AR.4, 8
Different methods are used to assess age at AR. It was first assessed by visual inspection.4, 8 This involves identifying an upward trend in the BMI after the nadir. In some cases, the descending phase of BMI is followed by a plateau. In this case, the age at AR corresponds to the start of the steep BMI increase (e.g. 8 years in case 4; Figure 1). In order to identify the upward trend in BMI, Dorosty et al.9 specified that all consecutive measurements of the BMI after the nadir should show an increase, and required that any increase in the BMI after the nadir had to equal or exceed 0.1 kg/m2. Gasser et al.10 determined the AR from the velocity curve of the BMI. Age at AR corresponded to the point where the velocity curve became positive after a fall of BMI in infancy and early childhood. Polynomial models to describe the patterns of change in BMI are also used.11, 12, 13, 14, 15 Age at minimum BMI is derived from these models. Comparing different approaches, a recent study15 concluded that estimating AR visually appears to best reflect the physiological basis of the AR. Using this method, mean age at AR in the French reference population was 6.2±1.6 years.4, 8 This is similar to the generally reported mean values, varying between 5 and 7 years according to the population studied and the method used.11, 12, 13, 14, 15, 16
Prediction of adult fatness
AR and adult fatness
On average, AR takes place by the age of 6 years, but in individual cases, as shown in Figure 1, it may occur earlier or later.8 These individual BMI patterns explain why, before AR, the child's BMI predicts adult fatness only poorly. Case 3 in Figure 1 (a thin child at 4 years becoming overweight after an early AR) points out that many cases of overweight that are diagnosed at adolescence actually have their origin much earlier in life. The BMI chart can be a very useful tool for the paediatrician to monitor the BMI development in children.
Several studies have investigated the association between the age at AR and adult adiposity.11, 12, 13, 14, 16, 17, 18 Using the French sample6, 19 of the International Growth Study,20 we established that age at AR was significantly associated with BMI level at later ages:4 the earlier the AR, the higher the BMI or the subscapular skin fold at age 21 years.8 Regression analysis performed on this sample indicated that each year decrease in age at AR was associated with a 0.84 kg/m2 (0.82 in boys and 0.85 in girls) increase in the predicted BMI level at age 21 years (P<0.001) (unpublished data). In a study conducted in the US, each year decrease in age at AR was associated with a 2.5 kg/m2 increase in the predicted BMI level at age 19–23 years.17
Whether an early AR reflects increased fat or lean body mass was questioned.21 A study following children between 5 and 9 years found that differences in BMI during the AR period were caused specifically by alteration in body fat assessed by dual X-ray absorptiometry (DXA) rather than by alteration in lean mass. Children undergoing early AR gained fat at a faster rate than children who rebounded at a later age.22 From this observation, it is justified to go on using the term ‘Adiposity rebound’4 rather than using ‘BMI rebound’ as proposed later on.17
We also investigated the association between AR and bone age: an early AR was associated with advanced skeletal maturity.4 This association, also reported by Williams et al.,13 is consistent with the association between rapid growth and later high BMI23, 24, 25, 26 and, as a rule, with the accelerated growth of all body tissues observed in obese children.27 Accelerated growth is often reported in low birth weight infants.23 Using the data of the French reference study,19 no association was recorded between the age at AR and birth weight (r=−0.08; P=0.23) (unpublished data). An early AR then reflects rapid post-natal growth, rather than catch up growth following low birth weight. In a study conducted in New Zealand, AR arose earlier in children who were tall at 3 years, and an early AR was associated with early menarche, suggesting that the timing of rebound is an indicator of physical maturity.28
Comparison between indicators predicting adult adiposity
It has been suggested that an early AR predicts later fatness because it identifies children whose BMI is high21, 29 and/or crossing centiles upwards.29 BMI at the time of AR (5–7 years) and BMI centile crossing were proposed as more direct indicators than AR for predicting later fatness.3, 10, 13, 17, 21, 29 Actually, different aspects must be considered.
A high BMI level at the age of 6 years is significantly associated with later high BMI, but also with previous high BMI levels. In the Zurich study,10 the correlation between BMI at 6 years and adult BMI was 0.60. It was of the same magnitude (r=0.66) between the BMI at 6 years and the BMI at 1 year. Using the data of the French longitudinal study of nutrition and growth (ELANCE) described earlier,30, 31 we have compared the BMI patterns according to either BMI at 6 years or age at AR (Figure 2). Figure 2a shows that a high BMI at the age of 6 years is associated with high BMIs at all ages, starting from birth. By contrast, an early AR is associated with a high BMI at adult age, but with normal or even low BMI level before the rebound (Figure 2b). The association between an early AR and low or average BMI before the AR was reported in various studies.8, 9, 12, 28 Consequently, as a rule, a child fat at age 6 years is more likely to be always fat, whereas a child with an early AR is more likely to have a normal or even low adiposity level in early life and to develop fatness only from the time of AR. The BMI trajectory of subjects who were thin in infancy and thereafter put on weight rapidly (Figure 2b) is associated with insulin resistance and coronary heart diseases.32, 33, 34 In a study conducted in India,33 despite an increase in BMI between the ages of 2 and 12 years, none of the subjects with impaired glucose tolerance or diabetes were obese at the age of 12 years. The mechanisms by which body fat is acquired seem to be at least as important as the consequences of excess fat per se in the pathogenesis of diabetes, hypertension and cardiovascular diseases.35
Consequently, as compared with a high BMI at 6 years or with centile crossing,29 the early AR, which is associated with a particular BMI trajectory, has different implications and is likely associated with different environmental factors.
BMI pattern in the obese
Mean age at AR in the obese generally occurs around the age of 3 years compared to 6 years in reference populations. In a study conducted in 62 obese children examined in a department of paediatric endocrinology, mean age at AR was 3.2 years.8 None of these children had an AR later than the age of 6 years. Two children had their AR at the age of 6 years (i.e. average) and more than half of them (55%) had an AR at, or before, the age of 3 years.8 These observations, together with similar data reported elsewhere16, 36 showing that most obese subjects displayed an early AR, are in contrast to the observation that most obese adults were not obese as children. An early AR is a better indicator of the timing of the origin of obesity than a high BMI, which may only appear after a progressive increase over many years.37, 38 The very early AR recorded in most obese subjects suggests that factors promoting obesity have operated very early in life and probably several years before the AR. This observation is consistent with studies showing that early (pre- and post-natal) life is a ‘critical period’ for later risks.1, 2
Secular trends in growth patterns
Childhood obesity is increasing but other growth parameters are also changing. Secular trends to increasing adult height, weight and BMI have been clearly documented. At birth, weight trends are small.39 In childhood, height trends tend to be greater than in adulthood, owing to the associated advance in maturation. The increase in height from one generation to the next occurs mainly in the first 2 years of life. The adult height trend matches that at age 2 years, so that the increment in adult height has already been achieved by age 2 years.39 Similar trends are observed in France40 where two 20-year follow-up studies were conducted under similar conditions 30 years apart.19, 31 Weight, height and BMI of subjects born in 1985 (ELANCE study) were converted into Z-scores for each age and sex, in relation to the population born in 1955 as the reference, using the LMS method41 (Figure 3). At the end of growth, subjects born in 1985 are taller, heavier and fatter31 than those born in 1955.19 Over the 30 years period, subscapular skin fold thickness at adult age increased from 8.8 to 14.2 mm in males and from 9.9 to 17.3 mm in females. Height trend is maximum between 4 and 10 years and decreases until adulthood. As reported earlier,39 the adult height trend is about the same as the trend at age 2 years. The decreased height trend observed from the age of 10 years is accounted for by the earlier growth cessation in children born in 1985, owing to their advanced maturation.31 At the age of 20 years, BMI is higher in the recent study, but this trend appears from the age at AR only. Before the AR, BMI was lower in children born in 1985, with a maximum trend at age 2 years. The lower BMI in early life can be explained by the greater gain in height as compared to weight. Mean age at AR occurred at 6.2 years in children born in 1955 and at 5.6 years in those born in 1985. As compared to subjects born in 1955, the growth pattern of subjects born 30 years later typically shows a rapid height gain and a low BMI in early life, followed by an early AR and subsequent high fatness.31 This BMI trajectory during childhood is associated with insulin resistance and coronary vascular diseases in later life.32, 33, 34 Similar secular trends are reported in other countries. Anthropometric measurements recorded in a longitudinal study conducted in German children from 197842 were compared with more recent data recorded from 2001.43 At the age of 2 years, BMI was lower in the second study (P<0.034), which was due to a gain of height as compared to the earlier study. In England, height increased in different age groups between 1972 and 1994, weight for height also showed a slight decline in the youngest (5- or 6-year-old) children, it remained unchanged in the 7- and 8-year- old and showed an increase in the 9- and 10-year- old.44
Factors associated with an early AR and increased fatness
Positive energy balance results from high energy intake or low physical activity. Results of a longitudinal study showed an increased fatness development in children with a sedentary lifestyle. This increase appeared after an early AR.45
Nutritional studies have investigated the influence of early nutrition on either later fatness development or on the age at AR. The association between nutritional intakes at the age of 2 years and fatness development was investigated in our longitudinal study of nutrition and growth.30 Both high protein (% energy) and high energy intakes were associated with increased BMI at 8 years, but following different BMI patterns. BMI values in subjects consuming a high-energy diet at 2 years were high at all ages with average age at AR. BMI values in subjects consuming a high-protein diet tended to be lower in early ages and became higher after an early AR occurred. The BMI pattern of subjects consuming a high-energy diet is similar to the pattern in subjects with a high BMI level at 6 years (Figure 2a), whereas the BMI pattern of subjects consuming a high-protein diet is similar to the pattern in subjects with an early AR (Figure 2b). No association was found between the percentage of energy from the other nutrients (fat and carbohydrates) and age at AR or fatness level at 8 years. Other studies found an association between high protein intake in infancy and later high fatness,24, 46 whereas others did not.9, 47 A study conducted in Danish children found an association with measures of body size (weight and height) but not with body fat.48
We previously proposed that the characteristics of childhood obesity, that is, increased stature and body mass (lean and fat)27 were the consequence of high protein intakes inducing changes in hormonal status.49, 50 High plasma insulin like growth factor-1 (IGF1) concentrations and reduced growth hormone (GH) secretion (spontaneous or in response to a wide variety of stimuli) are characteristic features of children with simple obesity.51 Increased IGF1 could promote hyperplasia in all tissues, including adipose tissue, and decreased GH levels could decrease lipolysis, promoting the development and maintenance of large fat stores. A study conducted in 2.5-year-old Danish children found that animal protein (from milk but not from meat or vegetables) was positively associated with serum IGF-1 concentrations and height.52
Our results regarding the association between nutritional intakes in infancy and growth patterns have underlined the inadequate nutrient balance of the infant diet in industrialized countries.50 By the age of 1 year, the infant diet is characterized by high intake of protein (∼4 g/kg body weight or 16% of total energy) and low intake of fat (∼28%). Protein intake represents about 3–4 times the protein needs. The nutrient imbalance is remarkable considering that human milk provides 6% of energy as proteins and 52% as fat.53 The protective effect of human milk reported in some studies could be explained by its low protein and high fat content. Breast-fed infants have a slower growth.23 A similar observation was recorded in our longitudinal study described earlier.30 Length increase between birth and 2 years was 36.8±2.8 cm in breast-fed infants and 38.1±2.9 in those who were not breast fed (P=0.03) (unpublished data). There is now increasing evidence that a rapid growth (weight and/or length) in infancy or early childhood can predispose to later risks.26 It is associated with large subsequent weight gain17, 23, 24, 54 and a central body fat pattern23, 25 and also with a risk of diabetes,55 cardiovascular disease56 and cancer.57, 58
Paradoxically, the proportion of fat in the infant diet is low at a period of high energy needs. Subsequently, the percentage of energy provided by fat increases. In our longitudinal study of nutrition and growth,59, 60 the percentage of fat intake increased from the age of 10 months to 6 years (27.4 at 10 months, 32.6 at 2 years, 36.5 at 4 years and 37.3 at 6 years), then plateaued until 16 years and reached 38% at 20 years (Figure 4), whereas it should be high in early infancy and should then gradually decrease during the first years of life.53
Similar fat intake patterns were reported in other countries.53 The BMI pattern associated with an early AR (low BMI level followed by high BMI after the AR) can be the consequence of reduced energy balance owing to high-protein, low-fat intake in early life followed by increasing fat intake with age. In the recent years, most studies report that infants have been fed a diet with a higher proportion of energy coming from protein and lower intakes of fat than would have been the case in the past.61, 62, 63, 64, 65 These changes could account for the secular trends of accelerated linear growth and decreased BMI in the first 2 years of life as shown in Figure 3. Decreased fat intake in French children was mainly owing to a decreased consumption of whole milk, compensated for by low-fat milk and dairy products.64 The daily consumption of whole milk in 2-year-old children was 206 g in 1973 and 90 g in 1986, whereas the consumption of low-fat milk (half skimmed and skimmed milk) was 44 g in 1973 and 216 g in 1986. The decline in energy intake in infants and children over the last decades is very likely the consequence of the low-energy protein-rich diet they currently consumed.50
Undernutrition at any time in early life (during fetal life as reflected by low birth weight1 or before an early adiposity rebound32, 33, 34) may predispose to later risks. A relative energy deficit66 may programme ‘thrifty metabolism’ and mechanisms of adaptative thermogenesis.35 Lucas67 proposed the term ‘programming’ for the process in which the programming stimulus exerts long-term effects when applied at a critical or sensitive period. Adaptation to low fat intake in early life may have adverse effects when, later in life, children will eat a more abundant high-fat diet. The low-fat diet given to infants may result from strategies to prevent obesity or cardiovascular diseases in industrialized countries, despite general agreement that dietary fat content should not be reduced in early life,53, 68 and from poor living conditions in the developing world.
In order to prevent obesity, it is of great interest to identify subjects at risk before adiposity reaches high levels. Age at AR predicts later fatness, but because it involves the examination of several points during growth, the AR mainly reflects the growth pattern rather than absolute fatness level. Considering the age at AR allows a clear understanding of changing BMI pattern in individuals. The early AR recorded in most obese subjects suggests that determinants of obesity have operated early in life. An early AR is associated with low fatness before the rebound and high fatness after the rebound. Similar BMI patterns are observed in relation with various other situations (secular trends, sedentary life style, high protein intake in early childhood). The pattern of low fatness level followed by increased fat gain has important health implications, as it is typically recorded in subjects with diabetes or cardiovascular diseases. Low fatness before the AR suggests that a period of energy deficit had occurred at an early stage of life, which could have ‘programmed’ adaptive metabolism, becoming detrimental later on. These observations stress the importance of an adequate nutritional status in early childhood and the necessity to provide nutritional intakes adapted at various stages of growth.
The AR seems to be a useful indicator to predict adiposity for the use of paediatricians and researchers. The association recorded between BMI patterns and various environmental conditions suggests new approaches to investigate the origin of obesity and metabolic diseases and to improve prevention strategies starting from early life.
About this article
International Journal of Obesity (2018)