Background: Studies of the aetiology of fatal diseases often rely on data obtained from relatives, which can cause loss of precision and introduce bias. We assessed the quality of such information on demographics, occupation, smoking and alcohol habits.
Methods: We compared contemporary interviews, based on a structured questionnaire, with male workers from the man-made vitreous fibre production industry in four European countries and their relatives. The participation rate was 63% (74 pairs of workers and relatives).
Results: Only minor differences in the ability to answer the questions appeared among workers and relatives, except for specific occupational questions. There was moderate to excellent agreement for demographics, residential and work history (kappa or intraclass correlation range: 0.44-0.98). For smoking habits, beer and wine consumption the agreement was good to excellent (range: 0.59-0.99). In particular, number of different residential areas, jobs, industries, and duration of wine drinking were significantly underreported by the relatives. No general determinant for reduced agreement appeared.
Conclusions: In general, the quality of information obtained from relatives appeared good. However, information on specific occupational exposures may be improved by supplementing the information from relatives with details obtained from colleagues, occupational hygiene experts or occupation-exposure matrices.