ObjectiveTo determine the performance of spirometry and respiratory oscillometry (RO) in the prediction of severe asthma exacerbations (SAEs) in children. MethodsIn a prospective study, 148 children (age 6–14 years) with asthma were assessed with RO, spirometry and a bronchodilator (BD) test. Based on the findings of spirometry and the BD test, they were classified into three phenotypes: air trapping (AT), airflow limitation (AFL) and normal. Twelve weeks later, they were re-evaluated in relation to the occurrence of SAEs. We analysed the performance of RO, spirometry and AT/AFL phenotypes for prediction of SAEs by means of positive and negative likelihood ratios, ROC curves with the corresponding areas under the curve (AUCs) and a multivariate analysis adjusted for potential confounders. ResultsDuring the follow-up, 7.4% of patients had SAEs, and there were differences between phenotypes (normal, 2.4%; AFL, 17.9%; AT, 22.2%, P = .005). The best AUC corresponded to the forced expiratory flow between 25% and 75% of vital capacity (FEF25-75): 0.787; 95% confidence interval, 0.600−0.973. Other significant AUCs were those for the reactance area (AX), forced expiratory volume in the first second (FEV1), the post-BD change in forced vital capacity (FVC), and the FEV1/FVC ratio. All of the variables had a low sensitivity for prediction of SAEs. The AT phenotype had the best specificity (93.8%; 95% CI, 87.9–97.0), but the positive and negative likelihood ratios were both significant only for the FEF25-75. In the multivariate analysis, only some spirometry parameters were significative for prediction of SAEs (AT phenotype, FEF25-75 and FEV1/FVC). ConclusionsSpirometry performed better than RO for prediction of SAEs in the medium term in schoolchildren with asthma.