A model-based approach for assessing bronchodilator responsiveness in children: The conventional cutoff revisited

An increase in FEV1 >=12% has been proposed in international guidelines as a clue to airway reversibility for diagnosing asthma in both adults and children. However, the validity of this cut-off has been questioned in the pediatric population. The aim of this study was to provide evidence that different cut-off values in BDR may be associated with better performance in discriminating among outpatient children with naïve asthma (A) and without asthma (NA). We compared three approaches: i) the conventional cutoff (12%); ii) the cut-off estimated by Youden's criteria; and iii) the cut-off based on a model-driven approach. we found that the conventional cut-off of 12% showed poor sensitivity in discriminating A and NA. The cut-off of 6.5% obtained maximizing Youden's J statistic showed higher sensitivity than the conventional one; however, the average correct classification rates obtained using the two criteria mentioned were less than 63%, highlighting poor discriminating performance. A model-based approach identifying three different categories of BDR - low (<7.9%), intermediate (7.9%-14.7%) and high (>=14.7%) - yielded correct classification rates higher than 80%. The model-based approach made it possible to develop a dynamic nomogram, which graphically returns the prediction probability of asthma, overcoming the elevated risk of misclassification associated with the use of the conventional cut-off of 12%.