[Quality assurance of data collection and data processing in epidemiologic study data]

Quality assurance of the data generating processes in epidemiologic studies is a prerequisite for the internal validity of study results. This paper presents practical aspects of such a quality assurance system pertaining to the planning, data gathering, data entry and data processing phase of a study. It is concerned with data obtained in the framework of a project rather than with data accumulating continuously in private practices, research institutes or veterinary faculties. During the planning phase of a project, standard operating protocols should be developed that assure a reliable performance of observation, coding and data entry. The data base structure, consisting of tables, input validation rules and queries, should be predefined and well documented. A data safety concept will provide the necessary integrity, physical safety and availability of the data. The paper presents technical solutions to common data processing problems with emphasis on re-coding and relational data base facilities (Microsoft-ACCESS) using a hypothetical study on risk factors for mastitis.