Spirometric data from 1586 healthy children, who did not smoke, were analysed to examine the effects of overweight as measured by the body mass index (weight/height2) on lung function. Overweight (72 boys, 88 girls) was defined as on or above the 90th percentile weight for height. After having controlled for the confounding variables of height and age, there were positive partial correlations between body mass index and lung function in girls whose weight was normal, in overweight girls, and in boys whose weight was normal, but not in overweight boys. In contrast to adults, body mass index has a positive effect on lung function in girls, whatever their weight. No such correlation between body mass index and lung function was seen in overweight boys. The observations may be accounted for by distinct sex dependent patterns of fat distribution in children. The effects of morbid obesity on respiratory function has been studied thoroughly in adults. Fat subjects tend to have lower lung volumes and less chest wall compliance.' 2 Information on lung function in children with normal lungs who are obese, but not morbidly so, is, however, sparse. The distribution of fat in adolescents differs from that in adults, and is sex dependent.3 Obesity, therefore, by its effect on the chest and diaphragm, may alter the lung function of children and adolescents differently from adults. Quetelet's index (weight/height2) is a body mass index for assessing obesity or overweight.4 We used it to evaluate the association between overweight and standing height and lung function in normal as well as nonmorbidly obese children. The measurements of lung function have an exponential relationship with age, height, weight, and body mass index, so they were all transformed logarithmically (base= 10) in statistical analyses. In bivariate linear regression analysis, lung function tests were the dependent variables and height and body mass index were the independent variables. The measurement 'age' was not included in the multiple regression analysis because of its high collinearity with stature (r>0 9). The comparative importance of a particular independent variable as a predictor of the dependent variable was estimated by the unit free standardised regression coefficients (mean and variance of the independent variable=0 and 1, respectively). The

No abstract

No abstract