AD after controlling for age, sex, depression, and education, but only for those with
normal cognition at baseline (Geerlings et al. 1999 ). The opposite was seen in the
Schofield et al. (1997a,b) study, in which SMC predicted dementia diagnosis only
for those who were impaired at baseline (Schofield et al.1997a).
Summary
Most studies that use a single or few questions to assess SMC demonstrate that
SMC are related to decline in objective cognitive performance over time. Of the
exceptions, one showed a positive relationship once the data were reanalyzed using
structural equation modeling and one showed a positive relationship only for those
with impaired cognitive performance at baseline. When validated questionnaires are
used to measure SMC, the data are more equivocal. In three studies, SMC assessed
by the Retrospective Functioning, Mnemonics Usage, and Frequency of Forgetting
subscales of the MFQ were unrelated to change in objective cognitive performance
over time. In contrast, two studies that used a sum total score for SMC found a
relationship between SMC and decline in objective cognitive performance over
time, and one study found that changes in a Frequency of Forgetting subscale are
related to declines in objective memory over time. The data on whether SMC reflect
decline from prior levels of objective function are similarly equivocal, with several
studiesfinding no effect and several showing that SMC do reflect decline from prior
levels of function. The positive and negative studies noted here included those
within both questionnaire categories—a single or few questions and validated
questionnaires. They also included those that tested a range of cognitive domains.
Thus, these disparatefindings cannot be attributed solely to methodological dif-
ferences. Finally, longitudinal studies almost universallyfind that baseline SMC
predict later cognitive impairment or dementia, although two studies suggest that
baseline cognitive status may affect this relationship.
Conclusions
The answer to the seemingly simple question“Do SMC relate to actual memory
function?”is complicated and depends on how SMC are ascertained, how actual
memory function is assessed, and the age and cognitive health of the population
studied. The literature in young and middle-aged adults is small, but reveals that
whether a single question about current memory problems is posed, or a validated
questionnaire examining types of cognitive problems is used, SMC do relate to
objective cognitive function. This relationship is most apparent when a compre-
hensive battery of cognitive tests that are challenging enough to produce a range of
scores is used, as SMC relate not only to verbal memory, but also to attentional
processes such as encoding and working memory. A small neuroimaging study
292 M.T. Weber and P.M. Maki