Identifying population differences in lung function: results from the Allied Dunbar national fitness survey

In order to identify valid population differences in lung function (e.g. occupational, ethnic), it is necessary to adjust for known confounding variables (e.g. age, body size). The present paper proposes appropriate methods for analysing forced expiratory volume (FEV1), forced vital capacity (FVC) and maximum oxygen uptake (VO2 max), recorded as part of the Allied Dunbar national fitness survey (ADNFS). The ADNFS randomly selected subjects from 30 regional sites throughout England. Measurements of FEV1, FVC and complete records of other relevant information were available on 2672 subjects. Traditional analyses of co-variance (ANCOVA) were found to be inappropriate to investigate population differences in FEV1 and FVC, due to a significant increase in error variance with age. However, by fitting a multiplicative model with allometric body size components to the FEV1 and FVC measurements using weighted log-linear regression, valid and plausible associations with body size, age, smoking, and physical activity, together with 'gender specific' regional differences in lung function were identified. Further insight was obtained when FEV1 and FVC were included into the multiplicative model to predict VO2 max. The apparent advantage of being taller when predicting VO2 max, was explained more accurately by the subjects' superior FVC. In summary, by fitting the multiplicative 'allometric' model using weighted log-linear regression, valid population differences in lung function were identified. Regions containing a higher proportion of working-class, unemployed or less affluent subjects were found to be associated with below average lung function performances.