Physician-derived asthma diagnoses made on the basis of questionnaire dataare in good agreement with interview-based diagnoses and are not affected by objective tests

Background: Defining the phenotype is critical for investigating the geneti c etiology of asthma. As part of the Collaborative Study on the Genetics of Asthma (CSGA), the primary objective of which is to identify asthma suscep tibility loci, an algorithm was designed to determine diagnoses of definite asthma, probable asthma, less than probable asthma, or no asthma, A respir atory questionnaire was designed to assist in the process of characterizing the asthma phenotype. Objective: This study was designed to determine the validity of the CSGA al gorithm for the diagnosis of asthma, to determine agreement in assessing an asthma diagnosis between the information obtained by the CSGA questionnair e versus a patient interview by a panel of specialist physicians, and to de termine the degree to which objective tests would alter the questionnaire-b ased certainty of asthma diagnosis. Methods: An expert panel of asthma clinicians (n = 4) indicated to what deg ree they were certain that a subject (n = 48) had asthma as determined by u sing a 6-point Likert scale based on a 20-minute interview (phase I), a rev iew of the CSGA questionnaire (phase II), a review of the questionnaire plu s skin test and peripheral blood eosinophilia data (phase III), and a revie w of phase III information plus pulmonary data (spirometry and methacholine -reversibility testing; IV), Intraclass correlation coefficients (ICCs) wer e calculated between the physicians' interpretation of the likelihood of as thma based on the information they received during each of the phases and b etween the CSGA algorithm and each of the phases. Results: Interjudge reliability with regard to the degree of certainty with which an asthma diagnosis could be made by interview was excellent (ICC, 9 8; 95% confidence intervals [95% CIs], 0.87-0.99). We also found that the a greement between the physicians' interview with the patients (phase I) and the CSGA algorithm was good and at least as good with the addition of the C SGA questionnaire data and objective data (ICC, 0.65-0.75). Good agreement was also observed between the average. certainty score from the interview a nd the CSGA questionnaire (ICC, 92; 95% CI, 0.76-0.93), and ICCs determinin g the agreement on asthma diagnosis between phase I and phases III and IV, in which objective data were introduced, did not change from the ICCs compa ring phase I with phase II (ICC of 0.93 [95% CI, 0.79-0.96] and ICC of 0.91 [95% CI 0.73-0.95], respectively). Conclusion: We conclude that the CSGA algorithm is a valid tool for which t he diagnosis of asthma can be made at an acceptable level of certainty and that the CSGA questionnaire, interpreted by an asthma specialist, is a usef ul tool for the diagnosis of asthma in clinical or epidemiologic studies.