Comparison between type A and type B early adiposity rebound in predicting overweight and obesity in children: a longitudinal study Johanna Roche1,2,3*†, Sylvain Quinart1,4†, David Thivel3,5, Stéphanie Pasteur4, Frédéric Mauny6,7, Fabienne Mougin1, Sandrine Godogo4, Mélaine Rose4, Florence Marchal4, Anne-Marie Bertrand4, Marc Puyraveau6,7 and Véronique Nègre4,8 1Research Unit EA3920 ‘Prognostic Markers and Regulatory Factors of Cardiovascular Diseases’, University of Bourgogne Franche-Comté, 25000 Besançon, France 2Brain Function Research Group, School of Physiology, Faculty of Health Sciences, University of the Witwatersrand, 2193 Johannesburg, Gauteng, South Africa 3Research Unit EA 3533 'Laboratory of the Metabolic Adaptations to Exercise under Physiological and Pathological Conditions' (AME2P), Clermont-Auvergne University, 63000 Clermont-Ferrand, France 4Pediatric Obesity Prevention and Rehabilitation Department, RéPPOP-FC, University Hospital of Besançon, 25000 Besançon, France 5Auvergne Research Center for Human Nutrition (CRNH), 63000 Clermont-Ferrand, France 6uMETh, Inserm CIC 1431, CHU Besançon, 25000 Besançon, France 7UMR Chrono-environnement 6249, CNRS/Université Bourgogne Franche-Comté, 25000 Besançon, France 8Specialized Obesity Center (CSO), University Hospital of Nice, 06000 Nice, PACA, France (Submitted 10 July 2019 – Final revision received 31 January 2020 – Accepted 2 March 2020 – First published online 16 March 2020) Abstract Early adiposity rebound (EAR) predicts paediatric overweight/obesity, but current approaches do not consider both the starting point of EAR and the BMI trajectory. We compared the clinical characteristics at birth, age 3–5 and 6–8 years of children, according to the EAR and to its type (type A/type B-EAR).We assessed the children’s odds of being classified as overweight/obese at age 6–8 years, according to the type of EAR as defined at age 3–5 years. As part of this two-wave observational study, 1055 children were recruited and examined at age 3–5 years. Antenatal and postnatal information was collected through interviews with parents, and weight and height from the health records. Type A and type B- EAR were defined in wave 1 according to the BMI nadir and the variation of BMI z-score between the starting point of the adiposity rebound and the last point on the curve. At 6–8 years (wave 2), 867 childrenwere followed up; 426 (40·4 %) children demonstrated EAR. Among them, 172 had type A-EAR, higher rates of parental obesity (P< 0·05) and greater birth weight compared with other children (P< 0·001). Odds for over- weight/obesity at 6–8 years, when adjusting for antenatal and postnatal factors, was 21·35 (95 % CI 10·94, 41·66) in type A-EAR children and not significant in type B-EAR children (OR 1·76; 95 % CI 0·84, 3·68) compared with children without EAR. Classification of EAR into two subtypes provides physicians with a reliable approach to identify children at risk for overweight/obesity before the age of 5 years. Key words: Early adiposity rebound: Childhood: Obesity: BMI trajectory: Perinatal risk factors Due to the high prevalence of paediatric obesity worldwide, despite a trend towards a stabilisation of this epidemic(1), early clinical screening must be improved in a preventive perspective. Amongst other factors, the detection of an adiposity rebound (AR), defined as ‘the point at which the BMI increases after its nadir’(2,3), before the age of 5·5 years, named ‘early AR’(2,4) (EAR), has been defined as a reliable indicator for future obesity(2,4–6). Adipose tissue undergoes physiological changes during the early years of life, with a rapid increase during the first year due to the growing size of the adipocytes and then a decrease during the following years, to remain stable until the age of 6 years. Thereafter, body fat shows a second phase of rapid growth, named ‘AR’, whichwas first reported and documented by Rolland-Cachera et al.(4). According to these andother authors, there is an association Abbreviations: AR, adiposity rebound; CUG, catch-up growth; EAR, early adiposity rebound; IOTF, International Obesity Task Force. * Corresponding author: Johanna Roche, email johanna.roche@wits.ac.za † These two authors contributed equally to the work and must be considered as two first authors. British Journal of Nutrition (2020), 124, 501–512 doi:10.1017/S0007114520000987 © The Authors 2020 D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 mailto:johanna.roche@wits.ac.za https://doi.org/10.1017/S0007114520000987 https://crossmark.crossref.org/dialog?doi=10.1017/S0007114520000987&domain=pdf https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 between the age at which AR occurs and adiposity at age 16 years(4,7). Later studies also documented the importance of EAR, highlighting its association not only with cardiometabolic health development(8) but also with maturation disorders, among others(9,10). Several studies have investigated the potential risk factors associated with the occurrence of this EAR, particularly focusing on antenatal and postnatal factors(11,12). For instance, maternal and paternal obesity are now recognised as strong risk factors for accelerated weight growth trajectory(13,14). Undeniably, the description of EAR and its associated antena- tal and postnatal risk factors constituted a major step for the understanding and management of childhood overweight, whose prevalence is still dramatically increasing worldwide(15). In France, 17 % of children and adolescents aged 6–17 years are overweight, including 4 % who are obese(16). However, the cur- rent definition of the EAR does not seem to be sufficiently dis- criminating. Accordingly, although Hughes et al. reported that 58 % of children identified as having an EAR before the age of 3·6 years were actually overweight at 15 years old, only 22 % of those identified with an EAR by the age of 5 years actually became overweight at 15 years old(17). In 2006, Rolland-Cachera et al.(2) described two profiles of early rebounders, namely ‘fat children’, who start their EAR at the top of their corpulence chart, and ‘lean children’, who start their EAR at the bottom of their corpulence chart (using French curves(18)). On the other hand, Thibault et al., in 2010, identified in a French cohort of 1424 children aged 8–9 years that the com- bination of the criterion ‘overweight at 3–4 or 5–6 years’ and ‘increase in BMI> 1 kg/m2 between 3–4 and 5–6 years’ was a good predictor of the risk of overweight at 8–9 years(19). Based on these studies, and given that only a modest proportion of children identified with EAR actually become overweight/ obese(17), we aimed to refine the definition of EAR using French curves related to the International Obesity Task Force (IOTF) references(18). Thus, we considered both the BMI z-score at the starting point of the EAR (BMI nadir) and the BMI trajec- tory. Indeed, it remains unclear whether the velocity of the BMI curve during EAR systematically predicts future overweight/ obesity or not. Therefore, in order to properly identify children most at risk for future overweight/obesity, we proposed a clas- sification of EAR into two classes, based on both initial BMI and BMI trajectory, namely: type A-EAR and type B-EAR. The objectives of the present study were: 1) to compare the clinical characteristics at birth, at age 3–5 years and at age 6–8 years of children, according to the presence or absence of EAR and, according to the type of EAR (type A and type B-EAR) and 2) to assess the children’s odds of being classified as overweight/obese at age 6–8 years, according to the type of EAR as defined at age 3–5 years. We further hypothesised that children exhibiting a type A or type B-EAR would exhibit similar clinical characteristics at birth, and that exhibiting a type A-EAR at age 3–5 years would be associated with a greater risk of overweight/obesity at age 6–8 years when adjusting for antenatal and postnatal factors, compared with children from the non-EAR group. Materials and methods Design and population This longitudinal, multicentre, observational study was con- ducted in the Department of Haute Saone, France, among children born between 1 January 2003 and 1 May 2005. As described below, two waves of evaluations were performed: the first at 3–5 years of age and the second at 6–8 years of age. Fig. 1 presents the flow chart of the study. Wave 1: a total of 1159 French children were enrolled in public and private pre-schools, as part of the annual medical n 1055 Missing or uninterpretable data (n 104) First wave: Protection of Mother and Infant service (PMI) - March to July 2008 Children born between 1 January 2003 and 1 May 2005 (n 1159) Second wave: School Health Service (SHS) - March to July 2011 Lack of follow-up data (n 134) n 921 n 867 Missing or uninterpretable data (n 54) Fig. 1. Flow chart of the study population. 502 J. Roche et al. D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 examination at school. Between March and July 2008, data collection including antenatal and postnatal information was collected from the children’s health record and from interview with the parents. Anthropometric measurements were performed by health professionals (nurses from the Protection of Mother and Infant service). Wave 2: 3 years later, between March and July 2011, we per- formed a second set of measurements similar to those of wave 1, as part of the annual medical examination. In total, 921 children from the initial overall sample completed the second wave of measurements. During eachwave, anthropometric measurements of the chil- dren were performed and antenatal and postnatal information were collected during an interview with at least one parent. We excluded children with secondary obesity such as syn- dromic, iatrogenic or endocrine obesity and children whose parents failed to attend the initial medical exam. This study was approved by the Ethics Committee of the University Hospital of Besançon, France (under the number 13/22), by the French national advisory committee for the processing of data in health research (number 14.705) and by the national commission for data protection (number 914632). Legal representatives of participating children were informed of the procedures and providedwritten informed consent before enrolment of their child in the study. Antenatal and postnatal information Data related to birth and the first years of life were retrospectively collected from the children’s health record and during the inter- views with parents in the first wave: Gestational age (extremely preterm (<28 weeks), very preterm (28–31 weeks), moderate preterm (32–33 weeks), late preterm (34–36 weeks) and term babies (≥37 weeks)) and body weight status at birth (hypotrophy, eutrophy or macrosomia, respectively, defined as birth weight <10th percentile, 10th–90th percentile or>90th percentile of the reference values) were collected and classified according to French referen- ces(20,21). The mother’s smoking status during pregnancy was collected during the interview. Body weight and height of the father and themother (apart from pregnancy) were self-reported by the parents and used to calculate their BMI. The presence or absence of gestational diabetes and pregnancy-induced hypertension was self-reported by the parents. Postnatal data concerned how the infant was fed (infant formula feeding or maternal breast-feeding and, if yes, for how long). Body weight and height of the child, taken from the child’s health record, at birth and at 3, 9, 12, 24, 36 and 48 months, were collected to establish a graph of BMI. Anthropometric measurements and body weight determination Anthropometric measurements were conducted by health professionals during wave 1 and wave 2, as part of the annual medical examination. Body weight was measured to the nearest 0·1 kg using a calibrated scale, and height was determined to the nearest 0·01 m using a standing stadiometer for each child. BMI was calculated as body weight divided by the square of the height in m (kg/m2). BMI z-score, specific for age and sex, was calculated by the LMS method(22). Gestational age at birth for birth weight was calculated according to French reference data(20). Body weight status was classified into one of the three categories according to the IOTF referen- ces(22): thinness (IOTF-30). A classification according to the WHO references(23) was also performed, and the corresponding data are presented in online Supplementary Tables S1–S3. Determination of the early adiposity rebound. BMI generally peaks during the first year of life and subsequently declines, reaching a minimum around the age of 6 years(2). The point of minimal BMI value is the start of the AR. Each AR starting before the age of 5·5 years is usually considered as EAR(4). EAR was evaluated by two investigators: one trained and one untrained, in a blinded fashion, using BMI z-scores and French curves which refer to the IOTF references(18). Determination of early adiposity rebound type A and type B. EAR was sub-classified into two types, namely: type A-EAR and type B-EAR. Type A-EAR was defined according to two situations relative to the position of the BMI nadir in relation to the median (corre- sponding to a BMI z-score = 0 or to the 50th percentile): 1) If the BMI nadir is equal to or above the median (BMI z-score ≥ 0), a type A-EAR is defined when there is an increase of at least 0·5 units in BMI z-score between the starting point of the AR (BMI nadir) and the last point. 2) If the BMI nadir is below the median (BMI z-score < 0), a type A-EAR is defined when there is an increase equal to or greater than 1 unit in BMI z-score between the starting point of the AR (BMI nadir) and the last point. Type B-EAR was defined according to one situation: If the BMI nadir is below the median (BMI z-score < 0), a type B-EAR is defined when there is an increase greater than 0·5 but lower than 1 unit in BMI z-score between the starting point of the AR (BMI nadir) and the last point (see Fig. 2). Subjects were thus classified as non-EAR group or EAR group. Within the EAR group, two subgroups were distinguished, namely type A-EAR group and type B-EAR group. Determination of the catch-up growth. Catch-up growth (CUG) was defined at 2 years when a difference of >0·67 standard deviations was observed between the 24-month BMI z-score and the birth-BMI z-score (11). Statistical analysis All statistical analyses were performed with SAS software, version 9.4 (SAS Institute, Inc.). Discrete variables are expressed as numbers and percentages, and continuous variables as mean values and standard deviations unless otherwise stated. Comparisons between non-EAR and EAR groups and between non-EAR, type A-EAR and type B-EAR groups were performed using the χ2 or Fisher’s exact test, Student’s t test or ANOVA according to the type of variable. Post hoc tests were performed as appropriate (Scheffe’s test). Type A and type B early adiposity rebound 503 D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://doi.org/10.1017/S0007114520000987 https://doi.org/10.1017/S0007114520000987 https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 To determine the predictive factors of overweight and obesity in children, we first performed bivariate logistic analysis. Variables with a level of significance of ≤0·20 in the bivariate analysis were included in the multivariate model, which was analysed with stepwise logistic regression. Interaction effects were sought for all variables included in the model. Model dis- crimination was assessed by the C-index, which is equal to the AUC. Calibration was assessed by the Hosmer–Lemeshow 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Age (years) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Age (years) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Age (years) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 ∆+1·79 ∆+0·88 ∆+1·26 3e 10e 25e 50e 75e 90e 97e IOTF30 IOTF25 3e 10e 25e 50e 75e 90e 97e IOTF30 IOTF25 3e 10e 25e 50e 75e 90e 97e IOTF30 IOTF25 Age (years) B M I ( kg /m 2 ) 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 B M I ( kg /m 2 ) 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 B M I ( kg /m 2 ) 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 B M I ( kg /m 2 ) (a) (b) (c) (d) Fig. 2. Examples of non-early adiposity rebound (EAR) and EAR types presented on a French curve, according to the starting point of the EAR and the BMI trajectory. (a) NoEAR; (b and c) typeA-EAR; (d) typeB-EAR.Δ, Difference in BMI z-score between lastmeasure and starting point of theEAR; IOTF, InternationalObesity TaskForce. 504 J. Roche et al. D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 goodness-of-fit statistic. All reported P values are two-sided. A P value < 0·05 was considered statistically significant. Results Characteristics of the population In total, 1159 children were enrolled in the study; of these, 104 subjects were excluded from further analysis because of missing data or uninterpretable BMI trajectories. A total of 1055 children (509 girls and 546 boys) with a mean age of 4·4 (SD 0·4) years and a mean BMI of 15·7 (SD 1·5) kg/m2 were thus retained for analy- sis. At wave 2, a total of 921 subjects were seen. Fifty-four sub- jects were excluded, and 867 subjects (422 girls and 445 boys) were retained for final analysis, at amean age of 7·5 (SD 0·4) years and a mean BMI of 16·5 (SD 2·3) kg/m2. Among the 1055 interpreted BMI trajectories, agreement between the two investigators for EAR determination was found for 960 (91 %) children. A third observer was called on to resolve discrepancies for ninety-five subjects. EAR, determined during the first wave, was observed in 426 children (40·4 % of the population; 225 girls, 201 boys) who con- stitute the EAR group, while 629 subjects (284 girls and 345 boys) did not exhibit an EAR (non-EAR group, Table 1). Among the EAR group, 172 children (40·4 %; 93 girls, 79 boys) were identified as presenting type A-EAR and 254 subjects (59·6 %; 132 girls, 122 boys) exhibited type B-EAR (Table 2). Non-early adiposity rebound and early adiposity rebound groups Antenatal and postnatal information. Gestational age, size at birth including height, body weight and body weight status at birth were not different between the two groups. Mother’s smok- ing status during pregnancy, BMI of the parents, prevalence of pregnancy-induced hypertension, breast-feeding and gesta- tional diabetes were also similar between groups (Table 1). Characteristics at age 3–5 years. Children from the EAR group presented higher BMI (P< 0·0001) than children from the non- EAR group. A similar prevalence of CUGwas observed in the two groups. In the EAR group, only two children were thin (0·5 %), v. 35 (5·6 %) in non-EAR children. Among those with EAR, 57 (13·4 %) children were already overweight and 22 (5·2 %) were obese. Regarding the non-EAR group, 27 (4·3 %) children were overweight with a BMI trajectory at the top of the chart since early infancy, but none was obese (Table 1). Data according to the WHO references are shown in online Supplementary Table S1. Characteristics at age 6–8 years. Children from the EAR group presented higher BMI (P< 0·0001) than those from the non-EAR group. In the EAR group, no child was thin, while 5 (1·0 %) chil- dren from the non-EAR group were thin. In the EAR group, 77 (21·7 %) childrenwere overweight and 35 (9·9 %)were obese. Among the non-EAR group, only 41 (8·0 %) children were overweight and 4 (0·8 %) were obese (Table 1). Data according to the WHO references are shown in online Supplementary Table S1. Non-, type A- and type B-early adiposity rebound groups Antenatal and postnatal information. Subjects with type A-EAR had a greater gestational age compared with children from the non-EAR group (P< 0·001) and compared with those from the type B-EAR group (P< 0·001). At birth, height was greater in the type A-EAR group com- pared with the type B-EAR group (P< 0·001) and to the non- EAR group (P< 0·001). Birth weight was significantly higher in the type A-EAR group compared with the type B-EAR and non-EAR groups (P< 0·001, Table 2). Characteristics at age 3–5 years. Age was similar between the three groups. BMI was significantly greater in the type A-EAR group compared with the type B-EAR and the non-EAR groups (P< 0·0001, Table 2). Rates of CUG were 49 and 40·9 % in the type A-EAR and non- EAR groups, respectively, v. 28·5 % in the type B-EAR group (P< 0·0001). Concerning body weight status, in the type A-EAR group, 57 (33·1 %) children presented overweight and 22 (12·8 %) pre- sented obesity and no child presented thinness. In the type B-EAR group, 2 (0·8 %) children presented thinness and no child presented overweight or obesity (Table 2). Data according to the WHO references are shown in online Supplementary Table S2. Characteristics at age 6–8 years. Age was similar between the three groups. BMIwas higher in the type A-EAR group compared with the type B-EAR and non-EAR groups (P< 0·0001, Table 2). Concerning body weight status, in the type A-EAR group, 56 (40·0 %) children were overweight and 33 (23·6 %) children presented obesity. In the type B-EAR group, 21 (9·8 %) children were overweight and 2 (0·9 %) children presented obesity. In both groups, no child presented thinness (Table 2). Data accord- ing to the WHO references are shown in online Supplementary Table S2. When combining the type A and type B-EAR groups, 112 chil- dren were overweight/obese. 79·5 % of them (n 89) were from the type A-EAR group and 20·5 % (n 23) from the type B-EAR group. Risk factors for overweight and obesity at age 6–8 years Bivariate analysis. Gestational age, body weight status at birth (eutrophy, macrosomia and hypotrophy), gestational diabetes and pregnancy-induced hypertension were not associated with overweight/obesity at age 6–8 years. Risk factors for overweight/obesity at age 6–8 years by bivari- ate analysis were exposure to prenatal smoking (OR 1·98; 95 % CI 1·22, 3·21), maternal obesity (OR 3·00; 95 % CI 1·82, 4·95), paternal overweight (OR 1·80; 95 % CI 1·20, 2·71), paternal obesity (OR 3·51; 95 % CI 1·99, 6·19) and CUG (OR 2·40; 95 % CI 1·65, 3·50). Risk of overweight/obesity was decreased by breast- Type A and type B early adiposity rebound 505 D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://doi.org/10.1017/S0007114520000987 https://doi.org/10.1017/S0007114520000987 https://doi.org/10.1017/S0007114520000987 https://doi.org/10.1017/S0007114520000987 https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 Table 1. Characteristics of the non-early adiposity rebound (EAR) and EAR groups according to the International Obesity Task Force (IOTF) definition (Mean values and standard deviations; numbers and percentages) Non-EAR group EAR group Pn % n % Antenatal and postnatal information n 629 426 Sex (% boys) 54·9 47·2 0·015 Gestational age (weeks) 0·235* Mean 39·2 39·3 SD 2·1 2·1 Gestational age 0·439‡ <28 weeks 2 0·4 3 0·8 28–32 weeks 7 1·3 4 1·1 32–36 weeks 34 6·3 16 4·2 >37 weeks 497 92·0 358 93·9 Size at birth Height (cm) 0·261* Mean 49·3 49·5 SD 2·9 2·8 Body weight (kg) 0·952* Mean 3·2 3·2 SD 0·6 0·5 Body weight status at birth 0·577† Eutrophy 498 79·7 341 80·4 Macrosomia 36 5·8 29 6·8 Hypotrophy 91 14·5 54 12·8 Mother’s smoking status during pregnancy 0·317† Not smoker 359 80·3 248 83·2 Smoker 88 19·7 50 16·8 BMI of the mother 0·476† <25 kg/m2 420 72·4 271 69·0 25–30 kg/m2 99 17·1 78 19·8 ≥30 kg/m2 61 10·5 44 11·2 BMI of the father 0·180† <25 kg/m2 313 56·4 192 50·3 25–30 kg/m2 192 34·6 151 39·5 ≥30 kg/m2 50 9·0 39 10·2 Gestational diabetes 0·515† No 382 91·8 259 93·2 Yes 34 8·2 19 6·8 Pregnancy-induced hypertension 0·620† No 390 94·4 263 95·3 Yes 23 5·6 13 4·7 Breast-feeding 0·443† No 265 44·0 168 40·7 <6 months 256 42·4 192 46·5 ≥6 months 82 13·6 53 12·8 Clinical parameters at age 3–5 years Age (years) 0·513* Mean 4·4 4·4 SD 0·4 0·5 BMI (kg/m2) <0·0001* Mean 15·3 16·3 SD 1·2 1·7 Catch-up growth 0·197† No 302 59·1 247 63·3 Yes 209 40·9 143 36·7 Body weight status <0·0001‡ Thinness (IOTF-30) 0 0 22 5·2 Clinical parameters at age 6–8 years n 513 354 Sex (% boys) 54·6 46·6 0·021 Age (years) 0·641* Mean 7·5 7·5 SD 0·4 0·4 506 J. Roche et al. D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 Table 1. (Continued ) Non-EAR group EAR group Pn % n % BMI (kg/m2) <0·0001* Mean 15·9 17·4 SD 1·6 2·9 Body weight status <0·0001‡ Thinness (IOTF-30) 4 0·8 35 9·9 * Student’s t test for quantitative variables for comparison between non-EAR and EAR groups. † χ2 Test for qualitative data analysis, for comparison between non-EAR and EAR groups. ‡ Fisher’s exact test for qualitative data analysis, for comparison between non-EAR and EAR groups. Table 2. Characteristics of the non-early adiposity rebound (EAR), type A and type B-EAR groups according to the International Obesity Task Force (IOTF) definition (Mean values and standard deviations; numbers and percentages) Non-EAR group (0) Type A-EAR group (1) Type B-EAR group (2) Dunn’s test (P) n % n % n % P 1 v. 0 1 v. 2 2 v. 0 Antenatal and postnatal information n 629 172 254 Sex (% boys) 54·9 45·9 48·0 0·046 Gestational age (weeks) 0·003† *** *** NS Mean 39·2 39·8 39·1 SD 2·0 1·3 2·4 Gestational age (weeks) 0·088§ <28 2 0·4 0 0 3 1·3 28–32 7 1·3 0 0 4 1·8 32–36 34 6·3 3 2·0 13 5·7 >37 497 92·0 149 98·0 209 91·3 Size at birth Height (cm) 0·015† *** *** NS Mean 49·3 50·0 49·1 SD 2·9 2·0 3·2 Body weight (kg) <0·0001† *** *** NS Mean 3·2 3·4 3·1 SD 0·6 0·4 0·6 Body weight status at birth 0·070‡ Eutrophy 498 79·7 143 83·6 198 78·3 Macrosomia 36 5·8 15 8·8 14 5·5 Hypotrophy 91 14·5 13 7·6 41 16·2 Mother’s smoking status during pregnancy 0·484‡ Not smoker 359 80·3 152 84·4 96 81·4 Smoker 88 19·7 28 15·6 22 18·6 BMI of the mother 0·026‡ <25 kg/m2 420 72·4 92 60·1 179 74·6 25–30 kg/m2 99 17·1 38 24·8 40 16·7 ≥30 kg/m2 61 10·5 23 15·0 21 8·7 BMI of the father 0·027‡ <25 kg/m2 313 56·4 69 45·4 123 53·5 25–30 kg/m2 192 34·6 60 39·5 91 39·5 ≥30 kg/m2 50 9·0 23 15·1 16 7·0 Gestational diabetes No 382 91·8 100 92·6 159 93·5 0·777‡ Yes 34 8·2 8 7·4 11 6·5 Pregnancy-induced hypertension 0·737‡ No 390 94·4 104 96·3 159 94·6 Yes 23 5·6 4 3·7 9 5·4 Breast-feeding 0·713‡ No 265 44·0 63 38·6 105 42·0 <6 months 256 42·4 79 48·5 113 45·2 ≥6 months 82 13·6 21 12·9 32 12·8 Type A and type B early adiposity rebound 507 D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 feeding (duration of at least 6 months), compared with infants receiving infant formula (OR 0·42; 95 % CI 0·21, 0·84). Concerning EAR, children in the EAR group had an increased risk of overweight/obesity at age 6–8 years (OR 4·81; 95 % CI 3·29, 7·03) compared with those in the non-EAR group. In sub- group analyses, compared with those in the non-EAR group, children in the type A-EAR group had an increased risk of over- weight/obesity at age 6–8 years (OR 18·15; 95 % CI 11·45, 28·76), whereas those in the type B-EAR group did not have a signifi- cantly increased risk (OR 1·25; 95 % CI 0·74, 2·13) of overweight/obesity at age 6–8 years (Table 3). Results of the bivariate analyses according to the WHO definitions are presented in online Supplementary Table S3. Stepwise multivariate analysis. Risk factors for overweight/ obesity at age 6–8 years by stepwise multivariate logistic regres- sion analysis were mother’s smoking status during pregnancy (OR 2·31; 95 % CI 1·15, 4·65), maternal obesity (OR 2·84; 95 % CI 1·33, 6·06), paternal overweight (OR 2·29; 95 % CI 1·12, 4·28), paternal obesity (OR 2·81; 95 % CI 1·12, 7·06) and CUG (OR 2·18; 95 % CI 1·23, 3·88) and type A-EAR (OR 21·35; 95 % CI 10·94, 41·66). The C-index for the model was 0·846 (Table 3). According to the WHO definitions, the risk factors for overweight/obesity at age 6–8 years by stepwise multivariate logistic regression analysis were macrosomia (OR 5·68; 95 % CI 2·33, 13·86), mother’s smoking status during pregnancy (OR 2·73; 95 % CI 1·46, 5·12), maternal obesity (OR 2·59; 95 % CI 1·28, 5·24), paternal overweight (OR 2·14; 95 % CI 1·26, 3·63), type A-EAR (OR 13·96; 95 % CI 7·45, 26·18) and type B-EAR (OR 3·39; 95 % CI 1·94, 5·92). The C-index for the model was 0·831 (online Supplementary Table S3). Discussion The prevalence of paediatric overweight and obesity remains alarming worldwide(15), accompanied by early metabolic(24), sleep(25,26) and functional(27) impairments, amongst others, calling for effective preventive and treatment strategies. In the present study, 40·4 % of the children were classified as having an EAR. This prevalence is in line with other studies, such as Hughes et al.(17), who reported that 27·2 % of British children experienced EAR, and Ip et al.(12), who found a prevalence of about 58·3 % in Latino children. Although the occurrence of EAR is recognised as a predictive marker for future paediatric overweight and obesity, there is a need to reinforce our screeningmethods and to identify accurate and reliable predictors. In the present study, only 26·3 % of chil- dren identified as having an EAR actually became overweight/ obese at age 6–8 years. This low prevalence highlights the need Table 2. (Continued ) Non-EAR group (0) Type A-EAR group (1) Type B-EAR group (2) Dunn’s test (P) n % n % n % P 1 v. 0 1 v. 2 2 v. 0 Clinical parameters at age 3–5 years Age (years) 0·302† Mean 4·4 4·4 4·5 SD 0·4 0·5 0·4 BMI (kg/m2) <0·0001† *** *** NS Mean 15·3 17·8 15·3 SD 1·2 1·7 0·7 Catch-up growth 0·0001‡ No 302 59·1 79 51·0 168 71·5 Yes 209 40·9 76 49·0 67 28·5 Body weight status <0·0001§ Thinness (IOTF-30) 0 0 22 12·8 0 0 Clinical parameters at age 6–8 years n 513 140 214 Sex (% boys) 54·6 46·7 46·4 0·070 Age (years) 0·476† Mean 7·5 7·5 7·5 SD 0·4 0·4 0·4 BMI (kg/m2) <0·0001† *** *** NS Mean 15·9 19·5 16·0 SD 1·6 3·1 1·7 Body weight statusb <0·0001§ Thinness (IOTF-30) 4 0·8 33 23·6 2 0·9 * P< 0·05, ** P< 0·01, *** P< 0·001. † ANOVA for quantitative data analysis and ‡ χ2 test or § Fisher’s exact test for qualitative data analysis for comparison between non-EAR, type A-EAR and type B-EAR groups. Post hoc (Scheffe’s test) for comparison between groups. 508 J. Roche et al. D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://doi.org/10.1017/S0007114520000987 https://doi.org/10.1017/S0007114520000987 https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 to refine the definition of EAR, in order to detect only children at high risk of later overweight/obesity. Based on our clinical expertise, the present work proposes a sub-classification of EAR, considering not only early occurrence of EAR but mainly its velocity and the shape of its trajectory. Indeed, it seems important to differentiate between children whose BMI z-score is equal to or below 0 at the occurrence of the EAR and who show an alarming increase in their growth curve (type A); and those who have a lower BMI z-score with a low-to-moderate increase in their curve (type B). In this con- text, we aimed to compare the clinical characteristics at birth, age 3–5 and 6–8 years of children, according to the presence or absence of EAR, and according to the type of EAR (type A and type B-EAR), and to assess the children’s odds of being classified as overweight/obesity at age 6–8 years, according to the type of EAR defined at age 3–5 years. Regarding the clinical characteristics at birth, children with type B-EAR exhibited slightly lower height and body weight comparedwith children from the type A-EAR group and children from the non-EAR group, despite similar gestational age and maternal smoking status during pregnancy. Although these children could not be considered as small for gestational age, we could nonetheless expect to observe a substantial prevalence of CUG in this group, since 70–90 % of small for gestational age children usually experience this catch-up(28–30). Surprisingly, only 28·5 % of children with type B-EAR experienced CUG, compared with almost half of the type A-EAR group. This is of particular importance since CUG is recognised as a risk factor for overweight/obesity in childhood, as previously reported(31,32). Regarding the antenatal and postnatal factors, we found that the prevalence of both maternal and paternal obesity was greater in children with type A-EAR compared with the other children, which confirms the strong relation between parental obesity, EAR and risk of later overweight/obesity of the offspring. When assessing the children’s risk factors of being classified with overweight/obesity at age 6–8 years, some of the previously reported early-life risk factors for obesity during childhood, such as gestational age, body weight status at birth, gestational Table 3. Association between risk factors and overweight and obesity at age 6–8 years*† (Odds ratios and 95% confidence intervals) Unadjusted OR 95% CI P Final model adjusted OR 95% CI P Gestational age (weeks) 0·746 Did not enter – <32 0·72 0·16, 3·27 32–36 0·75 0·31, 1·81 >37 1 Body weight status at birth 0·075 Did not enter – Eutrophy 1 Macrosomia 1·53 0·81, 2·90 Hypotrophy 0·59 0·33, 1·07 Mother’s smoking status during pregnancy 0·006 0·003 Not smoker 1 1 Smoker 1·98 1·22, 3·21 2·31 1·15, 4·65 BMI of the mother (kg/m2) <0·0001 0·037 <25 1 1 25–30 1·36 0·85, 2·17 0·81 0·37, 1·78 ≥30 3·00 1·82, 4·95 2·84 1·33, 6·06 BMI of the father (kg/m2) <0·0001 0·012 <25 1 1 25–30 1·80 1·20, 2·71 2·29 1·12, 4·28 ≥30 3·51 1·99, 6·19 2·81 1·12, 7·06 Gestational diabetes 0·699 Did not enter – No 1 Yes 1·16 0·55, 2·48 Pregnancy-induced hypertension 0·823 Did not enter – No 1 Yes 0·90 0·34, 2·37 Breast-feeding 0·048 0·580 No 1 <6 months 0·87 0·60, 1·27 ≥6 months 0·42 0·21, 0·84 Catch-up growth <0·0001 0·008 No 1 1 Yes 2·40 1·65, 3·50 2·18 1·23, 3·88 EAR <0·0001 Did not enter – No 1 Yes 4·81 3·29, 7·03 Type of rebound <0·0001 <0·0001 No EAR 1 1 Type A-EAR 18·15 11·45, 28·76 21·35 10·94, 41·66 Type B-EAR 1·25 0·74, 2·13 1·76 0·84, 3·68 *C-index = 0·846, Hosmer–Lemeshow = 0·7109. † The candidate variables for the multivariate model were those with a level of significance of ≤0·20 in the bivariate analysis. Type A and type B early adiposity rebound 509 D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 diabetes and pregnancy-induced hypertension, were not found to be associated with weight gain in later childhood. Moreover, multivariate logistic regression analysis did not allow us to observe a protective effect of breast-feeding. However, we did not assess whether breast-feeding was exclusive or not, which can explain why we did not observe a protective effect of breast-feeding against overweight/obesity. While previous stud- ies questioned the potential relationships between these factors and the development of paediatric obesity, results remain con- flicting and further evidence is needed(33–36). As expected(11,37), parental body weight status represents a strong risk factor for paediatric overweight/obesity and children born to obese mothers or fathers, respectively, have a 2·84- or 2·81-fold increase in the risk of exhibiting overweight/obesity at the age of 6–8 years. Parental obesity may enhance the risk of obesity not only through genes involved in homoeostatic regulation, appetite suppression and the control of energy bal- ance(38) but also through shared familial characteristics such as food preferences(39) or physical inactivity(40). We found that exposure to prenatal smoking was associated with an increased risk of overweight/obesity in children (adjusted OR 2·31; 95 % CI 1·15, 4·65). Other authors have reported that infants whose mothers smoked during pregnancy had not only a preserved ponderal index(41) but also more fat mass and less fat-free mass than infants of non-smokers(42,43). Maternal smoking may also result in lower fetal growth which, in turn, is counterbalanced by faster postnatal weight gain(44). This quicker postnatal weight gain is associated with the development of overweight(45) and, in accordance with the literature(11,37), we found that CUG represents a risk factor for overweight/obesity at age 6–8 years. Finally, we observed that children in the EAR group had a 4·8-fold increase in the risk of overweight/obesity at age 6–8 years compared with those in the non-EAR group. AR, which is the point at which the BMI increases after its nadir(2,3), is consid- ered ‘early’ in the French paediatric population when it occurs before the age of 5·5 years(2,4). Since its first definition in 1984 by Rolland-Cachera et al.(4), numerous studies have confirmed its strong relationship with later overweight/obesity(2,4–6), as well as with cardiometabolic health(46). Over the last decade, some authors(47,48) have suggested that EAR was a risk factor for later fat- ness only because it identifies childrenwhoseBMIwas alreadyhigh at the occurrence of the rebound, without considering the shape and trajectories of the curve. However, as emphasised by Rolland-Cachera & Péneau(49) and by Rolland-Cachera & Cole(50), BMI at the age of the AR is usually close to the 50th centile of the reference population (corresponding to a BMI z-score close to 0), showing that an EAR is not necessarily associated with high BMI values at rebound. Although its reliability to precisely deter- mine later overweight and obesity remains to be confirmed, EAR is a major tool for physicians to target children at risk. As such, by differentiating type A from type B-EAR according to the nadir of BMI and the trajectory of the children’s BMI curves, we found that children found to have type A-EAR show a 21-fold increase in the risk of overweight and obesity at age 6–8 years comparedwith children from the non-EAR group, after adjusting formother’s smoking status during pregnancy, for CUG and for the body weight of the mother and father. In children defined with type B-EAR, the risk for overweight and obesity was not increased, compared with children from the non-EAR group. Additionally, overweight and obesity were diagnosed at age 6–8 years among 51·7 % children with type A-EAR, while only 9·1 % were found to be overweight/obese in the type B-EAR group. In this way, among the 426 children initially identified with EAR, only 40·4 % of them (i.e. those who were classified as type A-EAR) were considered as being at high risk for future over- weight/obesity. This supports the utility of this classification, which is more sensitive for the identification of children at risk of overweight and obesity, and provides physicians with an easier way to efficiently screen for and manage paediatric obesity. As a reminder, we observed 91 % agreement between the two investigators for EAR determination. Importantly, it must be emphasised that while children pre- senting a type B-EAR did not exhibit a risk for overweight/ obesity compared with children from the non-EAR group, we cannot ensure that they are metabolically healthier than children with type A-EAR. Rolland-Cachera & Péneau(49) reported that the two main BMI trajectories, respectively, ‘lean’ and ‘fat’ children experiencing EAR may be associated with different health risks. Indeed, overweight children with high birth weight followed by high lean body mass may have lower health risks(49,51), while rapid weight gain after AR in thin infants is associated with insulin resistance and CHD(49,52). This study has some limitations that deserve to be considered when interpreting our results. First, antenatal and postnatal factors were collected following interviews with at least one parent, which might limit the accuracy of the information col- lected (e.g. gestational diabetes, pregnancy-induced hyperten- sion, BMI of the parents, etc.). Additionally, breast-feeding and smoking status during the pregnancywere also self-reported by the parents. Regarding breast-feeding, we cannot ensure that it was exclusive and this lack of information may explain why breast-feeding did not appear to be protective against over- weight and obesity in multivariate analysis. Second, assessment of fat mass and fat-free mass would have been relevant in order to refine the AR assessment(53), which was not possible due to the design of the study. Third, we evaluated EAR at the age 3–5 years instead of at the exact age of 5·5 years. This study was run in the context of pre- ventive medicine, in early childhood, at the time of pre-school, and data were collected as part of the annual medical examina- tion performed in pre-school children. Consequently, we were unable to perform data collection at the exact age of 5·5 years. However, the 3–5 year age group is of particular interest, since evidence shows that the earlier the EAR, the more severe the obesity (12,54). Finally, it would be relevant, in further studies, to transpose the present method to BMI curves by using percentiles to define type A and type B-EAR, in order to make it accessible to a wider public. In the current context of high prevalence of paediatric obesity worldwide, effective strategies for early diagnosis of overweight/ obesity are needed. Based on our clinical expertise, we propose a classification of EAR, using IOTF charts, including both the starting point of the EAR and the BMI trajectory, differentiating 510 J. Roche et al. D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 between children whose BMI z-score is ≥ 0 at the occurrence of the EAR and/or who show an alarming increase in their growth curve (type A-EAR), and children who have a lower initial BMI and who exhibit a low-to-moderate increase in their curve (type B-EAR). Following this approach, we were able to target almost 80 % of the children initially identified with EAR at 3–5 years, who became overweight/obese at 6–8 years. Differentiating type A from type B-EAR might represent a universal, easy and sensi- tive way for physicians to identify and improve the early screen- ing of children at high risk of obesity. Acknowledgements The authors would like to thank Fiona Ecarnot, PhD for editorial assistance and Dr Karine Scheuermaier and Nicolas Coutarel for scientific output. The authors are grateful to Dr Laurence Guillaume and her staff of the school health service and to Dr Dominique Arnoud and the staff of the service of protection of mothers and infants for technical assistance. The authors would like to address particular thanks to the study participants and their parents. This studywas funded by the regional council of the Franche- Comté region (Conseil Régional de Franche-Comté) via intramu- ral funding in the hospital (APICHU research grant). J. R. and S. Q. wrote the paper with support from D. T. and V. N. M. P., S. Q., V. N., D. T. and J. R. designed the work and the analysis. M. R., F. M., S. Q. and S. G. collected the data. S. G., A.-M. B., S. P., S. Q. and V. N. contributed data or analysis tools. M. P. and F. Mauny performed the analysis. F. Mougin, D. T. and V. N. provided critical revision of the article and V. N. supervised the project. All authors provided final approval of the version to be published. All authors have read and approved the manuscript. No author has any conflicts of interest to declare. Supplementary material For supplementary material referred to in this article, please visit https://doi.org/10.1017/S0007114520000987 References 1. Rokholm B, Baker JL & Sørensen TIA (2010) The levelling off of the obesity epidemic since the year 1999 – a review of evidence and perspectives. Obes Rev 11, 835–846. 2. Rolland-Cachera MF, Deheeger M, Maillot M, et al. (2006) Early adiposity rebound: causes and consequences for obesity in children and adults. Int J Obes (Lond) 30, Suppl. 4, S11–S17. 3. Dietz WH (1997) Periods of risk in childhood for the develop- ment of adult obesity – what do we need to learn? J Nutr 127, 1884S–1886S. 4. Rolland-Cachera MF, Deheeger M, Bellisle F, et al. (1984) Adiposity rebound in children: a simple indicator for predicting obesity. Am J Clin Nutr 39, 129–135. 5. Kato N, Isojima T, Yokoya S, et al. (2018) Earlier BMI rebound and lower pre-rebound BMI as risk of obesity among Japanese preschool children. Int J Obes 42, 52–58. 6. Whitaker RC, Pepe MS, Wright JA, et al. (1998) Early adiposity rebound and the risk of adult obesity. Pediatrics 101, E5. 7. Williams SM&Goulding A (2009) Patterns of growth associated with the timing of adiposity rebound. Obesity (Silver Spring) 17, 335–341. 8. Di Gravio C, Krishnaveni GV, Somashekara R, et al. (2019) Comparing BMI with skinfolds to estimate age at adiposity rebound and its associations with cardio-metabolic risk mark- ers in adolescence. Int J Obes (Lond) 43, 683–690. 9. Sun Y, Fang J, Yang R, et al. (2017) Prospective association between early adiposity rebound and adolescent development in girls. Zhonghua Yu Fang Yi Xue Za Zhi 51, 796–800. 10. Marakaki C, KarapanouO, Gryparis A, et al. (2017) Early adipos- ity rebound and premature adrenarche. J Pediatr 186, 72–77. 11. Reilly JJ, Armstrong J, Dorosty AR, et al. (2005) Early life risk factors for obesity in childhood: cohort study. BMJ 330, 1357. 12. Ip EH, Marshall SA, Saldana S, et al. (2017) Determinants of adiposity rebound timing in children. J Pediatr 184, 151.e2–156.e2. 13. Toschke AM, von Kries R, Beyerlein A, et al. (2008) Risk factors for childhood obesity: shift of the entire BMI distribution vs. shift of the upper tail only in a cross sectional study. BMC Public Health 8, 115. 14. Dorosty AR, Emmett PM, Cowin IS, et al. (2000) Factors associ- ated with early adiposity rebound. Pediatrics 105, 1115–1118. 15. Han JC, Lawlor DA & Kimm SY (2010) Childhood obesity. Lancet 375, 1737–1748. 16. Verdot C, Torres M, Salavane B, et al. (2017) Children and adults body mass index in France in 2015. Results of the Esteban Study and trends since 2006. BEH 8, 234–241. 17. Hughes AR, Sherriff A, Ness AR, et al. (2014) Timing of adiposity rebound and adiposity in adolescence. Pediatrics 134, e1354–e1361. 18. Programme National Nutrition Santé (2014) CALIMCO. 19. Thibault H, Meless D, Carriere C, et al. (2010) Early screening criteria for children at risk of overweight. Arch Pediatr 17, 466–473. 20. Salomon L-J, Bernard J-P, de Stavola B, et al. (2007) Birthweight and size: charts and equations. J Gynecol Obstet Biol Reprod (Paris) 36, 50–56. 21. Torchin H, Ancel P-Y, Jarreau P-H, et al. (2015) Epidemiology of preterm birth: prevalence, recent trends, short- and long- term outcomes (article in French). J Gynecol Obstet Biol Reprod (Paris) 44, 723–731. 22. Cole TJ & Lobstein T (2012) Extended international (IOTF) body mass index cut-offs for thinness, overweight and obesity. Pediatr Obes 7, 284–294. 23. World Health Organization (2007) BMI-for-age (5–19 years). http://www.who.int/growthref/who2007_bmi_for_age/en/ (accessed February 2019). 24. Gueugnon C, Mougin F, Simon-RigaudM-L, et al. (2012) Effects of an in-patient treatment program based on regular exercise and a balanced diet on high molecular weight adiponectin, resistin levels, and insulin resistance in adolescents with severe obesity. Appl Physiol Nutr Metab Physiol Appl Nutr Metab 37, 672–679. 25. Roche J, Gillet V, Perret F, et al. (2018) Obstructive sleep apnea and sleep architecture in adolescents with severe obesity: effects of a 9-month lifestyle modification program based on regular exercise and a balanced diet. J Clin Sleep Med 14, 967–976. 26. Roche J, Isacco L, Perret F, et al. (2019) Beneficial effects of a life- style intervention program on C-reactive protein: impact of cardiorespiratory fitness in obese adolescents with sleep disturb- ances. Am J Physiol Regul Integr Comp Physiol 316, R376–R386. 27. Chaplais E, Thivel D, Greene D, et al. (2015) Bone-adiposity cross-talk: implications for pediatric obesity. A narrative review of literature. J Bone Miner Metab 33, 592–602. Type A and type B early adiposity rebound 511 D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://doi.org/10.1017/S0007114520000987 http://www.who.int/growthref/who2007_bmi_for_age/en/ https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987 28. Karlberg J & Albertsson-Wikland K (1995) Growth in full-term small-for-gestational-age infants: from birth to final height. Pediatr Res 38, 733–739. 29. Itabashi K, Mishina J, Tada H, et al. (2007) Longitudinal follow- up of height up to five years of age in infants born preterm small for gestational age; comparison to full-term small for gestational age infants. Early Hum Dev 83, 327–333. 30. Argente J, Mehls O & Barrios V (2010) Growth and body com- position in very young SGA children. Pediatr Nephrol 25, 679–685. 31. Toschke AM, Grote V, Koletzko B, et al. (2004) Identifying children at high risk for overweight at school entry by weight gain during the first 2 years. Arch Pediatr Adolesc Med 158, 449–452. 32. Eriksson JG (2007) Epidemiology, genes and the environment: lessons learned from the Helsinki Birth Cohort Study. J Intern Med 261, 418–425. 33. Giles LC, Whitrow MJ, Davies MJ, et al. (2015) Growth trajecto- ries in early childhood, their relationship with antenatal and postnatal factors, and development of obesity by age 9 years: results from an Australian birth cohort study. Int J Obes (Lond) 39, 1049–1056. 34. Hakanen T, Saha MT, Salo MK, et al. (2016) Mothers with gestational diabetes are more likely to give birth to children who experience early weight problems. Acta Paediatr 105, 1166–1172. 35. Zheng J-S, Liu H, Ong KK, et al. (2017) Maternal blood pressure rise during pregnancy and offspring obesity risk at 4–7 years old: the Jiaxing Birth Cohort. J Clin Endocrinol Metab 102, 4315–4322. 36. Wang Y, Gao E, Wu J, et al. (2009) Fetal macrosomia and adolescence obesity: results from a longitudinal cohort study. Int J Obes 33, 923–928. 37. Weng SF, Redsell SA, Swift JA, et al. (2012) Systematic review and meta-analyses of risk factors for childhood overweight identifiable during infancy. Arch Dis Child 97, 1019–1026. 38. Lenard NR & Berthoud H-R (2008) Central and peripheral regu- lation of food intake and physical activity: pathways and genes. Obesity (Silver Spring) 16, Suppl. 3, S11–S22. 39. Francis LA, Lee Y & Birch LL (2003) Parental weight status and girls’ television viewing, snacking, and body mass indexes. Obes Res 11, 143–151. 40. Parsons TJ, Power C, Logan S, et al. (1999) Childhood predic- tors of adult obesity: a systematic review. Int J Obes Relat Metab Disord 23, Suppl. 8, S1–S107. 41. Miller HC, Hassanein K & Hensleigh PA (1976) Fetal growth retardation in relation to maternal smoking and weight gain in pregnancy. Am J Obstet Gynecol 125, 55–60. 42. Cliver SP, Goldenberg RL, Cutter GR, et al. (1995) The effect of cigarette smoking on neonatal anthropometric measurements. Obstet Gynecol 85, 625–630. 43. Harvey NC, Poole JR, Javaid MK, et al. (2007) Parental determi- nants of neonatal body composition. J Clin Endocrinol Metab 92, 523–526. 44. Oken E, Levitan EB & Gillman MW (2008) Maternal smoking during pregnancy and child overweight: systematic review and meta-analysis. Int J Obes (Lond) 32, 201–210. 45. Oken E & Gillman MW (2003) Fetal origins of obesity. Obes Res 11, 496–506. 46. González L, Corvalán C, Pereira A, et al. (2014) Early adiposity rebound is associated with metabolic risk in 7-year-old chil- dren. Int J Obes 38, 1299–1304. 47. Dietz WH (2000) ‘Adiposity rebound’: reality or epiphenom- enon? Lancet 356, 2027–2028. 48. Cole T (2004) Children grow and horses race: is the adiposity rebound a critical period for later obesity? BMC Pediatr 4, 6. 49. Rolland-Cachera MF & Péneau S (2013) Growth trajectories associated with adult obesity. World Rev Nutr Diet 106, 127–134. 50. Rolland-Cachera MF & Cole TJ (2019) Does the age at adiposity rebound reflect a critical period? Pediatr Obes 14, e12467. 51. Singhal A, Wells J, Cole TJ, et al. (2003) Programming of lean body mass: a link between birth weight, obesity, and cardio- vascular disease? Am J Clin Nutr 77, 726–730. 52. Eriksson JG, Forsén T, Tuomilehto J, et al. (2003) Early adipos- ity rebound in childhood and risk of type 2 diabetes in adult life. Diabetologia 46, 190–194. 53. Plachta-Danielzik S, Bosy-Westphal A, Kehden B, et al. (2013) Adiposity rebound is misclassified by BMI rebound. Eur J Clin Nutr 67, 984–989. 54. Rolland-Cachera MF, Deheeger M, Guilloud-Bataille M, et al. (1987) Tracking the development of adiposity from one month of age to adulthood. Ann Hum Biol 14, 219–229. 512 J. Roche et al. D ow nloaded from https://w w w .cam bridge.org/core . W ITS U niversity (JO U RN ALS) , on 18 D ec 2020 at 10:04:08 , subject to the Cam bridge Core term s of use, available at https://w w w .cam bridge.org/core/term s . https://doi.org/10.1017/S0007114520000987 https://www.cambridge.org/core https://www.cambridge.org/core/terms https://doi.org/10.1017/S0007114520000987