ROC curve analyses and optimal cut-off values
The performance of IOS parameters in predicting a positive BDR was
assessed calculating the ROC curve for each analyzed parameter, that
resulted significant only for R5 and X5 (figure 1).
A decrease in R5 of 25.7% on baseline exhibited the best combination of
sensitivity and specificity (0.71 [95%CI 0.36-0.95] and 0,79
[95%CI 0.62-0.9], respectively) to detect an increase of FEV1 ≥12%
and/or ≥200 mL with an area under the ROC (AUC) of 0.77 (p = 0.03).
Similarly, an increase in X5 of 25.7% on baseline resulted the optimal
cut-off in terms of sensitivity and specificity (0.86 [95%CI
0.49-0.99] and 0.69 [95%CI 0.51-0.83], respectively) with an AUC
of 0.75 (p = 0.04). Conversely, AX was not able to discriminate between
a positive and negative BDR test (AUC 0.53; p = 0.83).