Stepwise health surveillance for bronchial irritability syndrome in workers at risk of occupational respiratory disease.
Post WK., Venables KM., Ross D., Cullinan P., Heederik D., Burdorf A.
OBJECTIVES: Questionnaires, lung function tests, and peak flow measurements are widely used in occupational health care to screen for subjects with respiratory disease. However, the diagnostic performance of these tests is often poor. Application of these tests in a stepwise manner would presumably result in a better characterisation of subjects with respiratory disease. METHODS: Cross sectional data from workers exposed to acid anhydrides, to laboratory animals, and to flour dusts were used. Sensitivity and specificity were calculated from cross tables of different (combinations of) tests for bronchial hyperresponsiveness and bronchial irritability in the past four weeks (BIS). From sensitivity and specificity likelihood ratios were computed and change in probability of BIS was calculated. RESULTS: The prevalence of BIS was 7%, 7%, and 5%, respectively. In all groups questionnaire data provided excellent sensitivity but poor specificity, which was inherent on the broad definition of symptoms. Adding the forced expiratory volume in one second/forced vital capacity (FEV1/FVC) ratio yields almost perfect specificity, and peak expiratory flow (PEF) variability is intermediate in populations in which smoking induced or non-allergic respiratory diseases predominates. In occupational groups in which asthma is a problem, adding PEF measurements will optimise sensitivity and specificity in detection of BIS. The probability of BIS for subjects with a negative combined test outcome was lower than the probability before testing. Subjects with a positive combined test outcome had a probability of BIS after the tests at least three times the probability before. CONCLUSIONS: Combined testing yields better sensitivity and specificity. An advantage of combined testing is an economy in the effort to screen for subjects with BIS. Combined testing resulted in more detailed estimation of the probability of BIS.