| Literature DB >> 11998546 |
Alexander M Walker1, Robert P Wise.
Abstract
Four barriers confront the architects and users of surveillance systems. Bonferroni's Bind blunts our ability to act on real safety signals when the surveillance scheme is capable of producing realistic but wrong signals just by chance. The paradox of definitive data is that the results of a truly comprehensive scheme become untestable because there is no further relevant information to be had. Flexible priors augment the first two barriers by intensifying our instinctive cautions, and by giving almost any claim the appearance of empiric proof. The poverty of tabular summaries is that they presuppose the existence of relevant categories for tabulation, and that they fail to persuade. The resolution for proactive surveillance is sen/cen: high sensitivity to signals and vigorous censoring to filter out the noise.Mesh:
Year: 2002 PMID: 11998546 DOI: 10.1002/pds.675
Source DB: PubMed Journal: Pharmacoepidemiol Drug Saf ISSN: 1053-8569 Impact factor: 2.890