Objectives: Measurement of fractional exhaled nitric oxide (FENO) might substitute bronchial provocation for diagnosing asthma. However, optimal FENO thresholds for diagnosing asthma remain unclear. We reanalyzed data collected for a systematic review investigating the diagnostic accuracy of FENO measurement to exploit all available thresholds under consideration of pretest probabilities using a newly developed statistical model.
Study design and setting: One hundred and fifty data sets for a total of 53 different cutoffs extracted from 26 studies with 4,518 participants were analyzed with the multiple thresholds model. This model allows identifying thresholds at which the test is likely to perform best.
Results: Diagnosing asthma might only be possible in a meaningful manner when the pretest probability of asthma is at least 30%. In that case, FENO > 50 ppb leads to a positive predictive value of 0.76 [95% confidence interval (CI): 0.29-0.96]. Excluding asthma might only be possible, when the pretest probability of asthma is 30% at maximum. Then, FENO < 20 ppb leads to a negative predictive value of 0.86 (95% CI 0.66-0.95).
Conclusion: The multiple thresholds model generates a more comprehensive and more clinically useful picture of the effects of different thresholds, which facilitates the determination of optimal thresholds for diagnosing or excluding asthma with FENO measurement.
Keywords: Asthma; Diagnostic accuracy; Fractional exhaled nitric oxide; Receiver operating characteristic analysis; Sensitivity; Specificity.
Copyright © 2017 The Authors. Published by Elsevier Inc. All rights reserved.