OBJECTIVE: To determine the relative importance of familial, dietary, behavioral, psychological and social risk factors for predicting body mass index (BMI) change, and onset of overweight and obesity among adolescent girls. METHODS: Data from the NHLBI Growth and Health Study (n = 2 150), a longitudinal cohort of girls, were used to identify the most important predictors of change in BMI percentile between the ages of 9 and 19 years, and second, risk for becoming overweight and obese. Forty-one baseline predictors were assessed using a tree-based regression method (Random forest) to rank the relative importance of risk factors. RESULTS: The five factors that best predicted change in BMI percentile (p < 0.05) were related to family socio-economic position (income and parent education) and drive to restrict eating and weight (body dissatisfaction, drive for thinness and unhappiness with physical appearance). The factors that were statistically significant (p < 0.05) predictors of both onset of overweight and obesity were income, ineffectiveness and race. CONCLUSIONS: Family socio-economic position and emotion regulation appeared as the top predictors of both BMI change and onset of overweight and obesity. Our results build upon prior findings that policies to prevent the onset of obesity during adolescence be targeted towards girls from lower socio-economic position households. Our findings also suggest several novel psychological factors including ineffectiveness as predictors of obesity during adolescence. These predictive findings offer a direction for future inquiry into adolescent obesity etiology using causal methods.
OBJECTIVE: To determine the relative importance of familial, dietary, behavioral, psychological and social risk factors for predicting body mass index (BMI) change, and onset of overweight and obesity among adolescent girls. METHODS: Data from the NHLBI Growth and Health Study (n = 2 150), a longitudinal cohort of girls, were used to identify the most important predictors of change in BMI percentile between the ages of 9 and 19 years, and second, risk for becoming overweight and obese. Forty-one baseline predictors were assessed using a tree-based regression method (Random forest) to rank the relative importance of risk factors. RESULTS: The five factors that best predicted change in BMI percentile (p < 0.05) were related to family socio-economic position (income and parent education) and drive to restrict eating and weight (body dissatisfaction, drive for thinness and unhappiness with physical appearance). The factors that were statistically significant (p < 0.05) predictors of both onset of overweight and obesity were income, ineffectiveness and race. CONCLUSIONS: Family socio-economic position and emotion regulation appeared as the top predictors of both BMI change and onset of overweight and obesity. Our results build upon prior findings that policies to prevent the onset of obesity during adolescence be targeted towards girls from lower socio-economic position households. Our findings also suggest several novel psychological factors including ineffectiveness as predictors of obesity during adolescence. These predictive findings offer a direction for future inquiry into adolescent obesity etiology using causal methods.
Authors: Maria Cecília F Assunção; Ludmila C Muniz; Samuel C Dumith; Valerie L Clark; Cora L P Araújo; Helen Gonçalves; Ana M B Menezes; Pedro C Hallal Journal: J Adolesc Health Date: 2012-11-10 Impact factor: 5.012