Discriminating asthma and COPD based on bronchodilator data: An improvement of the methods

Abstract

The degree of bronchodilation is usually expressed as a percentage of the predicted or baseline value. It has been shown that the relation between pre- and post-dilation lung function values is not adequately described by this approach: the sensitivity/specificity in separating asthmatics from cases of chronic obstructive pulmonary disease (COPD) could be improved. The alternative method we investigate is based on a logistic regression approach incorporating pre- and post-dilation lung function, age, sex and height. The discriminatory power of forced expiratory volume in one second (FEV1) increase as a percentage of the predicted or baseline FEV1 is compared to our approach using two databases containing bronchodilator data and diagnoses (asthma, chronic bronchitis or emphysema, the latter two grouped as (OPD). The increase as a percentage of the predicted or baseline approach and our alternative method show areas under the receiver operator curve (ROC) (males/females) of 0.552/0.629, 0.523/0.550 and 0.867/0.879 in one database. In the other database the same trend is present although less obvious: 0.628/0.730, 0.592/0.737 and 0.709/0.749. This increase in discriminatory power is obtained in the optimal use of all available information, especially age, which is not used in the increase of the predicted/baseline FEV1 rule. © 2005 IOP Publishing Ltd.