Prediction of AHR by a multivariate score
Admission symptoms related to AHR and asthma have a wide variability, and the diagnostic role of a single symptom is limited. The combination of single items can improve the power of prediction(7, 24). We combined the five items, extracted by multivariate analysis as independent predictive factors, to construct a weighted multivariate score which reached an AUC > 0.7 which is acceptable for a prognostic model(39). To our knowledge, we generated a novel algorithm to predict AHR in children without an obstructive pattern. In contrast to the previously established predictive scores like the asthma predictive index (API), we aimed for the diagnosis of current asthma and not for the prediction of the development of asthma in later life(24). This algorithm could allow a reliable identification of those patients with an increased probability of AHR, and where an extended diagnostic using MCC might be helpful, against those patients who are less likely to have AHR and so prevent the mental stress and even the exhaustion of such patients caused by the severe symptoms induced by the performance of the MCC.