Epidemiology is reductionist in that it usually relies on creating cat
egories of people or risk factors. Classification must be undertaken a
s part of any study, however by the act of choosing groups, individual
s are (potentially) consigned to either a higher or lower risk group.
We discuss this from an ethical perspective and consider: (a) whether
the groupings commonly chosen genuinely represent the risk factor of i
nterest, (b) the implications for individuals when consigned to groups
and (c) the implications for epidemiology. (C) 1998 Published by Else
vier Science Ltd. All rights reserved.