Conflicting results were also obtained when clinical scoring systems were applied to children with asthma in acute-care settings. Some investigators have been able to distinguish between patients who need to be admitted to the hospital and those who can be discharged on the basis of a spirometric threshold, but this again has not been a universal finding. Similarly, variable results have been obtained when the value of arterial blood gas measurements were used in predicting patient outcome. In a recent study, Geelhoed et al measured oxygen saturation by pulse oximetry in a group of children with acute asthma, and found that an initial saturation of 91 percent differentiated patients with a favorable vs unfavorable outcome. ventolin inhalers
Most of the above cited works recently have been evaluated by Wasson et al, and severe methodologic defects have been demonstrated. These defects included failure in definition of outcome, and lack of blind assessment of the predictive variables and the outcome decision. The purpose of this study was to assess the usefulness of historic data, clinical score, spirometry, oxygen saturation, or any combination of these variables in predicting the outcome of acute asthma episodes in children. In this study, we used the “CART” statistical software based on a tree-structured form of discriminant analysis, to build and assess clinical prediction rules. As compared to more traditional methods such as multivariate discriminant analysis, this method can produce more flexible decision rules. In particular, it is more effective at detecting interactions among the predictors.