The Accuracy of a Screening Tool in Epidemiological Studies—An Example of Exhaled Nitric Oxide in Paediatric Asthma

Abstract

Diagnostic tests are widely used in medicine, especially in the clinical setting. The doctor’s decision regarding the treatment process is mostly based on the result of the diagnostic test. The value of the test is expressed by its accuracy. It is easier to verify the accuracy of a diagnostic test in a clinical setting in comparison to an epidemiological setting. Moreover, a very good test may not work in epidemiological settings in the same effective way as in a clinical setting, especially because the accuracy is affected by the prevalence of the disease. The aim of the study is to assess the accuracy of FeNO measurement in different respiratory disorders or symptoms, including their prevalence, in a childhood population. The secondary aim is to suggest the optimal FeNO cut-off for epidemiological screening for respiratory diseases and symptoms. Methods: The cross-sectional study included 447 children (50.8% boys and 49.2% girls) aged 6–9 years. An adapted version of the ISAAC questionnaire was used for the assessment of the respiratory status. FeNO was measured with an electrochemical device (Niox Mino) according to ERS/ATS recommendations. For interpretation, the FeNO cut-off values of 20 parts per billion (ppb), 25 ppb and 35 ppb were applied taking the real-life prevalence of the disease or symptoms into consideration and also for simulated prevalences of 20%, 30%, 40%, 50% for the interpretation of the accuracy of the test. The accuracy was calculated according to the following formula: Accuracy = (Prevalence) (Sensitivity) + (1- Prevalence) (Specificity). The area under the curve was calculated based on logistic regression. Results: For all respiratory outcomes, FeNO accuracy decreased with increasing prevalence, and in general the area under the curve (AUC) was low. The highest FeNO accuracy was found for any asthma diagnosis (with possible coexisting diseases/symptoms), i.e., 78.6%, 92.8% and 88.5% for FeNO cut-offs >19 ppb, >24 ppb and >34 ppb, respectively. The AUC was 0.628. For the same FeNO cut-offs, the accuracy of an asthma diagnosis without any coexisting diseases and symptoms was 81.2%, 87.5%, 92.9%, respectively, with an AUC of 0.757. Conclusion: FeNO accuracy decreases with increasing prevalence of the respiratory disease and symptoms. The best accuracy for the FeNO cut-off in the screening of asthma for epidemiological purposes is 35 ppb. For isolated asthma, the best accuracy for FeNO was 92.9%.