Background. This report identifies demographic and health-related base
line factors associated with non-compliance and attrition in elderly s
ubjects participating in a preventive intervention study over a 4-year
follow-up period. Methods. The Cox proportional hazards model and pol
ychotomous logistic regression were used to evaluate time to clinic as
sessment noncompliance and type of non-participation. Results. Of 1785
subjects, 35.6% were non-compliant during follow-up which meant they
were eligible to participate at the time they were contacted but would
not or could not return for an assessment. Refusals due to a lack of
continuing interest or being too busy, and those citing health problem
s constituted the major subgroups of non-compliers. Death, unable to c
ontact, and ineligible due to disenrolment or moving away accounted fo
r nearly 20%. Four-year compliance was achieved by 44.4% of the partic
ipants. Time to non-compliance was significantly shorter with older ag
e, lower annual income, poorer personal health ranking, and, to a marg
inal extent, depression. Lower annual income was associated with all r
easons for non-participation including disenrolment or moving. Other f
actors correlated with specific types of nonparticipation. Conclusions
. These results can be useful in designing more effective cohort maint
enance strategies. Greater resources may be allocated to subgroups who
are less likely to continue.