Comparison of body mass index values proposed by Cole et al. (2000) and Must et al. (1991) for identifying obese children with weight-for-height index recommended by the World Health Organization

Abstract

AbstractObjectives:To calculate the sensitivity, specificity and agreement of body mass index (BMI) values proposed by Cole et al. (Br. Med. J. 2000; 320: 1) and Must et al. (Am. J. Clin. Nutr. 1991; 53: 839 & 54: 773) with weight-for-height index in the nutritional evaluation of children.Design:Criterion standards for diagnostic tests.Setting:North-east and south-east Brazil.Subjects:Two thousand nine hundred and twenty children studied in Life Pattern Research performed by the Brazilian Institute of Geography and Statistics in 1997. Main outcome measures are the sensitivity, specificity and agreement of BMI values proposed by Must et al. (1991) and Cole et al. (2000).Results:Sensitivity of values proposed by both authors was around 90%. Specificity was almost 100% considering weight-for-height index as the gold standard. The agreement of both values with weight-for-height index, based on kappa results, was good and in pre-school children it was excellent.Conclusions:Values proposed by Cole et al. (2000) and Must et al. (1991) should be used carefully to screen obesity in childhood but can be used to ‘diagnose’ overweight children with a very low chance of having false-positive results. Although the values proposed by both authors performed similarly, use of Cole et al.'s values should be encouraged. The latter cover children from 2 to 6 years old; their values are presented for six-month age intervals; they are based on a larger sample from six different countries; and they are related to the definition of adult obesity.