[Design and implementation of a multicenter case-control study on bladder cancer in Spain]

The aim of a case-control design is to make a measurement of the highest possible accuracy, reducing to a minimum the possible selection, information and confusion biases. The design, methods and procedures of a multicentric study carried out in 14 hospitals from 4 autonomous communities in Spain are discussed. Overall 254 incident cases, 243 prevalent cases, 583 hospital controls and 530 population controls were included. The proportion of nonresponders was somewhat higher in population controls, particularly in large cities. Statistically significant differences between cases and controls were not found for socioeconomic levels, educational level, number of jobs, degree of response, duration of the interview, and errors of transcription and codification of information. Also, significant differences were not found between the population and hospital controls regarding diet habits and cigarette smoking. It is considered that a non-biased measurement of the association can be obtained from cases and hospital controls. The collected information was satisfactory. In spite of the lack of adequate facilities for research in hospitals, the development of multicentric etiologic studies is feasible.