Background: Evaluation and ethical review of epidemiological research projects raises the problem of the limits between research and non-research. This ambiguous boundary reflects the status of this discipline at the crossroads between research and practical action. The question then is: in the field of health research, what gives data collection and analysis its quality of scientific activity?
Methods: A conceptual and empirical study has been conducted about the practices of epidemiological research evaluation, centred on the case of the French Consultative Committee for the data processing in health research (CCTIRS), which is a consultative board that permits the National commission for the personal data protection (CNIL) to take decision about health research protocols that process personal data. The study was realized from 2003 to 2006.
Results: It is shown that the evaluation of such research protocols processing personal data articulates intimately two kinds of criteria: methodology and relevance. By studying and characterizing the different kinds of protocols that are judged not to be "scientific research" (poor science, pseudo-science and non-science), it becomes possible to understand the motives that lead to distinguish between what is and what is not research in epidemiology. A special attention is given to two kinds of problematic cases: firstly, the case of conflict of interests into the protocols themselves (i.e. seeding trials or surveys); secondly, the problem of epidemiological registers and other databases which are not hypothesis-oriented. This last case leads to relate the conceptual frame of the committee with historical circumstances (the way which this discipline was introduced in France) and also mere epistemological considerations (the question of induction and generalizability).
Conclusion: The activity of this committee illustrates a differentiated conception of what is research in epidemiology, influenced by explanatory analytical research paradigms. Finally, the field of epidemiological research appears to be structured by some values that appear through the elaboration and the application of the ethical and regulatory texts.