[Bias due to non-responders in an epidemiological study (SAPALDIA)]

Within the Swiss Study on Air Pollution and Lung Diseases in Adults (SAPALDIA) 16267 adults aged 18 to 60 years from 8 different locations in Switzerland were randomly selected for answering a questionnaire about respiratory health and have a lung function examination with allergy test. 9561 subjects agreed with the examination (59%) (= responders, R). In order to study the possible influence of the bias introduced by non-responders (NR), 221 subjects who refused to participate among the 966 first subjects selected in Payerne were contacted by phone. 142 accepted a home visit and answered a shortened questionnaire about the main respiratory symptoms and diseases and indicated furthermore the reasons for their refusal. Non-responders have a lower mean educational level and belong to lower social classes than responders. The frequency of respiratory symptoms and diseases, allergies and smoking is similar in R and NR except a higher frequency of wheezing during the last 12 month (R: 12.5%, NR: 5.6%, p = 0.03). The level of carbon monoxide in expired air is higher in NR (17.6 ppm) that in R (11.9 ppm) (p = 0.01). A similar difference exists between NR (30.7 pp) and R (24.8 ppm) among current smokers (p < 0.01). The main reasons for refusal are lack of time (27.5%), lack of interest for medical study (22.6%), fear of health professionals (18.3%) or the existence of a another disease (9.9%). Furthermore, 2.8% of the subjects consider a medical study as useless and refuse principally any participation. The role of local press and media in the decision to participate seems to be important. Globally, the differences between R and NR are minimal and should not influence the validity of the results of the SAPALDIA study.