Abstract
First of all, the set of patients used to generate the discriminant function must be different from the set used to test it. [...]the hypothesis-testing variables should be independent of the hypothesis-generating variables. Several attempts at objective documentation of such an early response have failed.1 2 3 I know of only one study that showed a prompt effect, and it was conducted in patients with mild asthma, who may not have needed steroids.4 Collins et al. did show a clear-cut improvement in forced expiration after six hours.2 To the Editor: We agree with the comments of Dr. Coifman and Dr. Franklin that an alternative to hospitalization in patients with a predictor index ≥4 who appear to respond to emergency-room treatment (relapse group) may be the addition of a short-term course of corticosteroids. [...]despite this difference, considerable overlap occurred and peak flow at discharge was less predictive of outcome than was initial peak flow, which was less predictive than the initial index scores. [...]we believe that the initial severity of acute asthma as measured by the predictor index better predicts outcome than response to therapy.