Objective: To determine whether participation in a consensus conferenc
e on the assessment of dementia would influence conference participant
s with respect to their recommendations to primary care physicians for
the assessment of dementia. Design: Questionnaire before and after th
e conference. Setting: Canadian Consensus Conference on the Assessment
of Dementia, held in Montreal, Oct. 5 and 6, 1989. Participants: All
38 experts representing relevant health disciplines who participated i
n the consensus conference; 36 completed both questionnaires. Outcome
measures: Participants' opinion before and after the conference as to
how frequently each of 28 manoeuvres (12 blood tests, 4 neurologic ima
ging procedures, 4 types of consultation and 8 ''other'' tests) should
be ordered by primary care physicians as part of an assessment of a p
atient with dementia suspected on clinical grounds. Results: For 18 (6
4%) of the 28 manoeuvres (10 of the 12 blood tests, 3 of the 4 neurolo
gic imaging procedures and 5 of the 8 ''other'' tests), there was a sh
ift in opinion after the conference toward recommending that primary c
are physicians them less often, for 10 of these 18 (5 blood tests and
5 ''other'' tests) the shift was statistically significant. For the re
maining 10 manoeuvres (36%) the shift in opinion was toward a recommen
dation that primary care physicians order them more often; the shift w
as not statistically significant for any of these 10 manoeuvres. Concl
usion: Expert members of a consensus conference are influenced by the
process of having participated in such a conference and are capable an
d willing to change their initial recommendations when confronted with
relevant data.