Objective: The present field study examines the use of BMI in comparison with estimates of percent fat mass to screen for overweight and obesity in children.
Design: Cross-sectional field study.
Setting: Four waves of children 1996-1999 at Kiel, North West Germany.
Subjects: A representative large sample of 2,286 5-7 year old children representing 40% of the total child population examined by school physicians within the same period in Kiel.
Main outcome measures: BMI was compared with anthropometric measures (TSF, BSF, SIF, SSF) and bioelectrical impedance analysis (BIA). The 90th and 97th BMI percentiles were used as cut offs for overweight and obesity, respectively.
Results: BMI reached a low sensitivity to identify overweight children when compared with the two estimates of % FM (0.60 to 0.78 for girls, 0.71 to 0.82 for boys, respectively). The specificity of BMI was 93 to 95 %. By contrast, BMI reached higher sensitivity to screen for obese children of 0.83 to 0.85 for boys and 0.62 to 0.80 for girls at a concomitant specificity of 0.95 to 0.98 for boys and 0.96 to 0.97 for girls as defined by assessment of body fat mass. Comparing nutritional status of overweight children classified as overweight or non overweight by BMI shows that BMI only identified obese but not-overweight children.
Conclusion: BMI can be used to screen for obese children. In contrast BMI has a poor sensitivity to screen for overweight children. Body composition analysis should be used to screen for children at risk of becoming obese.