Steven Stellman - Workshop on guidelines to the epidemiology of weak associations. Confounding

Version 1

      Publication Details (including relevant citation   information):

      Stellman, S. D. 16 (2) 165-82-

      Abstract: The magnitude of confounding is   examined in nine case studies of two "weak" relationships:   between artificial sweeteners and bladder cancer, and between   oral contraceptives and cervical dysplasia. Confounding had   little or no influence on the results of any published study. The   responsible epidemiologist must always consider the possibility   of confounding, no less when associations are weak than when they   are strong. Identification of potentially confounding variables   is an integral part of good epidemiologic practice. Rarely,   however, does confounding itself, especially from unidentified   sources, live up to its reputation for introducing seriously   spurious associations. An investigator is more likely to be led   astray by undetected biases than by pure confounding.

      Address (URL): http://www.ncbi.nlm.nih.gov/pubmed/3588560