OBJECTIVE: To compare different approaches (visual estimation of individual BMI curves with polynomial models) to estimate age at adiposity rebound (AR), as different approaches might lead to different results. AR has been suggested as a critical period between intra-uterine life and early adulthood, and recent data showed that early age at AR is associated with higher body mass later in life. METHODS: Longitudinal anthropometric data from the DOrtmund Nutritional and Anthropometric Longitudinally Designed (DONALD) Study were used to obtain individual BMI growth curves. We then compared the visual estimation approach to polynomial models in three different scenarios reflected by different data sets: an idealistic, an realistic, and a realistic scenario with imputed values. RESULTS: In all three scenarios, the visual estimation yielded significantly higher estimates than the polynomial models of 2nd or 3rd order. Cross-tabulations of groups of age at AR (early, medium, and late) showed that truly concordant classification was low, ranging only from 51 to 63%. A closer examination of the data indicated that the differences in estimates were mainly due to differences in the underlying definitions: the polynomial models select the nadir in the growth curve as the age at AR, whereas the visual estimation deviates from this concept in those cases where there is plateau in the growth curve. In the latter instance, the turning point of the growth curve before its increase is selected as the age at rebound. CONCLUSIONS: Estimating AR with the visual approach appears to best reflect the physiological basis of the AR, and is also preferable, because it resulted in the lowest number of children with missing estimates for age at AR. Only when the underlying criteria for the estimation of AR with the visual approach were modified, could concordant results between the two approaches be obtained. Considering the underlying physiological basis, it became clear that approaches which determine AR by simply identifying the nadir in the BMI curve do not reflect AR appropriately. This refers to those cases in which the nadir in the growth curve and the turning point at the onset of the adiposity increase are not identical.
OBJECTIVE: To compare different approaches (visual estimation of individual BMI curves with polynomial models) to estimate age at adiposity rebound (AR), as different approaches might lead to different results. AR has been suggested as a critical period between intra-uterine life and early adulthood, and recent data showed that early age at AR is associated with higher body mass later in life. METHODS: Longitudinal anthropometric data from the DOrtmund Nutritional and Anthropometric Longitudinally Designed (DONALD) Study were used to obtain individual BMI growth curves. We then compared the visual estimation approach to polynomial models in three different scenarios reflected by different data sets: an idealistic, an realistic, and a realistic scenario with imputed values. RESULTS: In all three scenarios, the visual estimation yielded significantly higher estimates than the polynomial models of 2nd or 3rd order. Cross-tabulations of groups of age at AR (early, medium, and late) showed that truly concordant classification was low, ranging only from 51 to 63%. A closer examination of the data indicated that the differences in estimates were mainly due to differences in the underlying definitions: the polynomial models select the nadir in the growth curve as the age at AR, whereas the visual estimation deviates from this concept in those cases where there is plateau in the growth curve. In the latter instance, the turning point of the growth curve before its increase is selected as the age at rebound. CONCLUSIONS: Estimating AR with the visual approach appears to best reflect the physiological basis of the AR, and is also preferable, because it resulted in the lowest number of children with missing estimates for age at AR. Only when the underlying criteria for the estimation of AR with the visual approach were modified, could concordant results between the two approaches be obtained. Considering the underlying physiological basis, it became clear that approaches which determine AR by simply identifying the nadir in the BMI curve do not reflect AR appropriately. This refers to those cases in which the nadir in the growth curve and the turning point at the onset of the adiposity increase are not identical.
Authors: Izzuddin M Aris; Sheryl L Rifas-Shiman; Ling-Jun Li; Ken P Kleinman; Brent A Coull; Diane R Gold; Marie-France Hivert; Michael S Kramer; Emily Oken Journal: Int J Epidemiol Date: 2019-02-01 Impact factor: 7.196
Authors: C Börnhorst; A Siani; M Tornaritis; D Molnár; L Lissner; S Regber; L Reisch; A De Decker; L A Moreno; W Ahrens; I Pigeot Journal: Int J Obes (Lond) Date: 2016-11-30 Impact factor: 5.095
Authors: N Kato; T Isojima; S Yokoya; T Tanaka; A Ono; H Yokomichi; Z Yamagata; S Tanaka; H Matsubara; M Ishikuro; M Kikuya; S Chida; M Hosoya; S Kuriyama; S Kure Journal: Int J Obes (Lond) Date: 2017-10-03 Impact factor: 5.095
Authors: Edward H Ip; Sarah A Marshall; Santiago Saldana; Joseph A Skelton; Cynthia K Suerken; Thomas A Arcury; Sara A Quandt Journal: J Pediatr Date: 2017-02-24 Impact factor: 4.406
Authors: S Péneau; R González-Carrascosa; G Gusto; D Goxe; O Lantieri; L Fezeu; S Hercberg; M F Rolland-Cachera Journal: Int J Obes (Lond) Date: 2016-03-22 Impact factor: 5.095
Authors: Izzuddin M Aris; Sheryl L Rifas-Shiman; Ling-Jun Li; Ken Kleinman; Brent A Coull; Diane R Gold; Marie-France Hivert; Michael S Kramer; Emily Oken Journal: J Pediatr Date: 2018-06-28 Impact factor: 4.406
Authors: Leonardo Trasande; Chris Cronk; Maureen Durkin; Marianne Weiss; Dale A Schoeller; Elizabeth A Gall; Jeanne B Hewitt; Aaron L Carrel; Philip J Landrigan; Matthew W Gillman Journal: Environ Health Perspect Date: 2008-09-12 Impact factor: 9.031