Epidemiology works in a public domain, gathering the results of survey
s and trials into forms of knowledge which are made available to many
stakeholders. Health policy makers, lawyers, the media, medical techno
logy companies, and those who use and deliver health services all have
legitimate interests in epidemiology. There is unfortunately no commo
n language in which each of these stakeholders can express their inter
est in the outcomes of epidemiological studies. The largest and most i
mportant gap exists between those who use computational data and those
who use cultural acid linguistic models to generate their explanation
s. Methods have been described, however, which allow the identificatio
n of all legitimate stakeholders before epidemiological studies are un
dertaken. Identifying the stakeholders. however, will serve no purpose
unless there is a prior commitment by epidemiologists to respect both
reductionist and narrative accounts of truth. (C) 1998 Elsevier Scien
ce Ltd. All rights reserved.