Authors' Response: A further plea for adherence to the principles underlying science in general and the epidemiologic enterprise in particular

In a recent Commentary in the Journal of the National Cancer Institute, we argued that false positive findings are common in epidemiologic studies of environmental and occupational determinants of cancer and that, given the non-experimental nature of epidemiologic research, this should come as no surprise.1 We presented several examples of likely false positives and called for, among other things, increased epistemological modesty in the reporting and interpretation of study findings, particularly when they are new, with little supporting biologic rationale, and arise from a plethora of comparisons. Vineis, in an article in this issue of the journal,2 has misconstrued and misrepresented our recommendations for straightforward scientific skepticism and epistemic humility. (We will focus on the substance of the Vineis article and leave to readers the judgement about whether its underlying tone was appropriate to scientific discourse.) A number of his points obfuscate rather than clarify the issues we raised. Although Vineis suggests otherwise, we are well aware of the principles and practical issues involved in assessing causality in exposure–disease relations. Nowhere have we argued that epidemiologists ‘just restrict themselves to a mere description of “possible” risk factors’; in fact we encourage inferences about exposures as ‘causes’ of diseases, but only when such conjectures are based on valid, reliable evidence rather than on sporadic, unreplicated, ‘cherry-picked’ findings.3,4 We look forward to a cessation of the recent development in which epidemiologic studies cannot be criticized without initiating letters and other ‘thought pieces’ accusing researchers who express methodological skepticism or reservation of representing ‘the vision of some powerful groups with vested interests’, or being involved in manufacturing uncertainty.