Does self reported BMI really reflectthe proportion of overweight and obese children?

Background. Monitoring the prevalence of overweight and obesity is crucial to understanding whether adolescents’ self reported anthropometric measures can be used to estimate weight problems. This study assesses reliability and validity of self reported height, weight and BMI in a representative sample of Italian adolescents and examines independent predictors of reporting errors. Methods. The Health Behaviour in School Aged Children questionnaire was used to collect data among 11, 13 and 15 year old pupils in the Piedmont region (Italy). Weight and height were self assessed by students and measured in school by trained personnel. Overweight and obesity were identified using Cole classification. Agreement was assessed by Pearson correlation coefficient and weighted kappa statistic. Validity was evaluated by sensitivity and specificity to determine how accurately students classified themselves into BMI categories. Multivariate logistic regression was performed to study predictors of bias between self rated and measured BMI categories. Results. Italian students underestimate their weight by an average of 1.2 kg and underreport BMI by 0.6 kg/m2. Girls tend to underreport their weight more than boys. Self reported data are highly specific in identifying not overweight (94.5%) and obese children (99.3%) but sensitivity is modest. Girls are systematically less sensitive than boys. Body self image may influence self reported data and fat perceiving children are less accurate in reporting their anthropometric measures. Conclusion. Self reported data may be used to analyse general trends on aggregate data but not to assess correctly overweight and obesity prevalence, especially at the individual level.