ABSTRACT
Epidemiologic methods have recently come under increased pressure to provide critical, decision-making information in the political, regulatory, and legal arenas. The quest for higher certainty, faster results, and lower cost tempts many epidemiologists to consider the use of plentiful and inexpensive data from existing monitoring networks and surveys. This paper discusses some of the troubling methodologic questions raised by the use of pre-existing data bases. Since yielding to temptation might not always be wrong (or at least unprofitable), the paper also identifies potential strengths of the approach in studying environment/disease associations. For purposes of illustration, emphasis is maintained on exposure data bases and on studies of the effects of ambient air pollution. Many of the points apply to data bases on health effects and to studies of other types of environmental exposures.
