ASTHMA AND ASTHMA-LIKE SYMPTOMS IN ADULTS ASSESSED BY QUESTIONNAIRES - A LITERATURE-REVIEW

The first widely used questionnaire in respiratory epidemiology was the questionnaire from the Medical Research Council (MRC) of Great Britain. In the first version, from 1960, there were only a few questions about wheezing, but in later editions, more questions about asthma and asthma-like symptoms were added. The MRC questionnaire initiated the development of other questionnaires such as the European Community for Coal and Steel (ECSC) questionnaire of respiratory symptoms and the questionnaire from the American Thoracic Society and the Division of Lung Diseases (ATS-DLD-78). In Tucson, Ariz, a questionnaire was developed in the 1970s that was focused on the subject's own report of asthma. In Great Britain, a questionnaire was developed in the 1980s with the intention of finding the most valid symptom- based items for identifying asthma, 'the IUATLD (1984) questionnaire.' When judging the validity of a questionnaire, it is essential to understand sensitivity and specificity. Sensitivity is the fraction of the truly diseased subjects found to be diseased using the questionnaire. Specificity is the fraction of the truly healthy subjects found to be healthy using the questionnaire. Regarding questionnaires dealing with asthma, the situation is confusing because of the absence of any gold standard for asthma. The most usual mode of validation has been to test the questionnaire against the results of a clinical physiologic investigation, often a nonspecific bronchial challenge test. Another approach has been to compare the answers from the questionnaire with the clinical diagnoses of asthma. When validated in relation to bronchial challenge tests, the questions about self-reported asthma have a mean sensitivity of 36 percent (range, 7 to 80 percent) and a mean specificity of 94 percent (range, 74 to 100 percent). The questions about 'physician-diagnosed asthma' have even higher specificity, 99 percent. When validated in relation to a clinical diagnosis of asthma, the mean sensitivity for the question about self-reported asthma was 68 percent in the reviewed studies (range, 48 to 100 percent). The specificity was 94 percent (range, 78 to 100 percent). One problem in using the presence of bronchial hyperreactivity (BHR) as a gold standard for asthma is that many people with BHR report no respiratory complaints. In other words, the presence of BHR is a measure with high sensitivity but low specificity for asthma. The effect of using a methacholine challenge test as a standard for the disease will thus be an underestimation of the sensitivity of the questionnaire. The problem with using validation in relation to a physician's diagnosis of asthma is that the bias is probably considerable between different physicians. Hence, the best way to identify subjects with asthma when validating a questionnaire is to use a combination of clinical physiologic investigations and a clinical judgement of the symptoms. In epidemiologic studies of asthma, a disease with a low prevalence (<5 percent), the specificity of the diagnostic test is of great importance. A low specificity, below 98 percent, will generate many false-positive cases. This will be deleterious for both the exposure-disease analyses and the comparisons of prevalences between different populations. In epidemiologic studies of asthma, incidence studies are preferable. Hence, questions that take temporal aspects into consideration have to be developed.