Sir.—I read with interest the article by Shaw et al1 on assessing bronchiolitis. It is always a challenge to predict which infants with wheezing will have rocky courses and which can be discharged to outpatient follow-up, so a predictive test would be welcome. Two aspects of this article concerned me. One involves the potential for expectation bias in the study, and the other involves the generalizability of the study.
As pointed out by Sackett and colleagues,2 a diagnostic test must be evaluated without influencing or being a component of the outcome (the criterion or "gold" standard) it is being used to predict. In this instance, eventual overall severity of illness was the standard against which the predictive test of outpatient assessment was judged. However, the outpatient assessment was made using criteria that may have directly influenced the judgment of illness severity. For example, the ability to