Analysis 5 examines the importance of individual components of the clinical score, along with the other presentation variables. It should be noted that the value of pulsus paradoxus as a predictive variable is hampered by the fact that it was not measured in 79 patients (39.5 percent), particularly in the younger children. Dyspnea was the sole variable chosen from all presentation variables, as a prediction rule, with a sensitivity of 88 percent, but, specificity of only 71 percent. buy ampicillin
The relative importance of all the variables on presentation were as follow: dyspnea (100), accessory muscle use (81), Sa02 (71), wheezing (53), respiratory rate (44), FEV\ (14). The relative importance of the variables on disposition were dyspnea (100), wheezing (88), accessory muscle use (82). When the overall clinical score was included in the analysis with the individual components of the score, the overall clinical score was selected, leading to the same decision trees as in analyses 1 to 4. When each of the changes in the clinical score, Sa02 or pulmonary function values from presentation to discharge were added to the analysis, they were not chosen as predicting variables in the decision tree.
When the 54 patients who performed pulmonary function were analyzed separately, the CART decision rules did not change.