OBJECTIVE: To develop a tool to identify children with high risk of adult overweight (AO), especially before developing overweight, based on body mass index (BMI) standard deviation score(s) (SDS) changes between 2-6 years (y) of age. METHODS: We fitted a linear spline model to BMI SDS of 762 young Caucasian adults from the Terneuzen Birth Cohort at fixed ages between birth and 18 y. By linear regression analysis, we assessed the increase in explained variance of the adult BMI SDS by adding the BMI SDS at 2 y to the models including the BMI SDS at 4 y, 6 y and both 4 y and 6 y. AO risk was modelled by logistic regression. The internal validity was estimated using bootstrap techniques. Risk models were represented as risk score diagrams by gender for the age intervals 2-4 y and 2-6 y. RESULTS: In addition to the BMI SDS at certain ages, the previous BMI SDS during childhood is positively related to adult weight. Receiver Operating Curves analysis provides insight into sensible cut-offs (AUC varied from 0.76 to 0.83). The sensitivity and specificity for 2-6 y at the cut-off of 0.25 and 0.5 are respectively, 0.76 and 0.74, and 0.36 and 0.93, whereas the PPV is 0.52 and 0.67, respectively. CONCLUSIONS: The risk score diagrams can serve as a tool for young children for primary prevention of adult overweight. To avoid wrongly designating children at risk for AO, we propose a cut-off with a high specificity at the risk of approximately 0.5. After external validation, wider adoption of this tool might enhance primary AO prevention.
OBJECTIVE: To develop a tool to identify children with high risk of adult overweight (AO), especially before developing overweight, based on body mass index (BMI) standard deviation score(s) (SDS) changes between 2-6 years (y) of age. METHODS: We fitted a linear spline model to BMI SDS of 762 young Caucasian adults from the Terneuzen Birth Cohort at fixed ages between birth and 18 y. By linear regression analysis, we assessed the increase in explained variance of the adult BMI SDS by adding the BMI SDS at 2 y to the models including the BMI SDS at 4 y, 6 y and both 4 y and 6 y. AO risk was modelled by logistic regression. The internal validity was estimated using bootstrap techniques. Risk models were represented as risk score diagrams by gender for the age intervals 2-4 y and 2-6 y. RESULTS: In addition to the BMI SDS at certain ages, the previous BMI SDS during childhood is positively related to adult weight. Receiver Operating Curves analysis provides insight into sensible cut-offs (AUC varied from 0.76 to 0.83). The sensitivity and specificity for 2-6 y at the cut-off of 0.25 and 0.5 are respectively, 0.76 and 0.74, and 0.36 and 0.93, whereas the PPV is 0.52 and 0.67, respectively. CONCLUSIONS: The risk score diagrams can serve as a tool for young children for primary prevention of adult overweight. To avoid wrongly designating children at risk for AO, we propose a cut-off with a high specificity at the risk of approximately 0.5. After external validation, wider adoption of this tool might enhance primary AO prevention.
Authors: Marieke Welten; Marlou L A de Kroon; Carry M Renders; Ewout W Steyerberg; Hein Raat; Jos W R Twisk; Martijn W Heymans Journal: Diagn Progn Res Date: 2018-02-13
Authors: Marlou La de Kroon; Alet Wijga; Yvonne Vergouwe; Martijn W Heijmans; Vincent Wv Jaddoe; Jos Wr Twisk; Hein Raat Journal: JMIR Res Protoc Date: 2016-06-23