The number of subjects at each study wave including number lost to follow-up and the reasons for attrition are shown in . Out of the 2212 participants enrolled at the prevalence, 154 (7.0%) were excluded from the incidence rate calculation due to baseline dementia (n=65) or baseline CIND/MCI that did not convert back to normal (n=89), 159 (7.0%) participants died before the two-year follow-up evaluation, and 172 (7.8%) were lost to follow-up (61 refused the two year evaluation, 153 subjects were lost to follow-up at the two year evaluation and 103 of them were also lost at the five year evaluation. Eight of the 17 baseline CIND/MCI subjects who converted to normal at two years were also lost to follow-up). A total of 1668 participants completed the 2-year follow-up and a total of 1255 participants completed the 5-year follow-up. A total of 1727 participants (those completed the two year evaluation (1668) and 59 subjects rescreened at the 5-year evaluation) were included in the analysis.
shows the baseline characteristics of the four groups of study participants and thus provides information on factors associated with attrition. Subjects who Completed Follow-up tended to be younger, have more years of education, have better CSI-D Informant Scores (ratings of daily function), have better CSI-D Cognitive Scores, have higher rate of self-reported hypertension and have lower rates of self-reported stroke than the group with Prevalent Dementia or CIND/MCI. The subjects who Completed Follow-up also had better CSI-D Informant Scores (ratings of daily function), have better CSI-D Cognitive Scores, lower rates of heart attack and stroke than the group that Died before Follow-up. The group that Completed Follow-up had higher proportions of female subjects and better CSI-D Cognitive Scores than the group that was Lost to Follow-up.
Baseline characteristics of the four groups of study participants.
shows the clinical diagnoses as a function of CSI-D screening group (good, intermediate, or poor) at the 2-year and 5-year follow-up. A total of 88 participants were diagnosed as having incident CIND/MCI (26 in the first wave and 62 in the second wave). The CSI-D performance groupings were defined to detect dementia and not unexpectedly incident cases of dementia at each wave were predominately from the poor performance group. Participants diagnosed as Normal at the incident waves came from each performance group while the incident CIND/MCI cases tended to come from the poor and intermediate categories.
Clinical diagnosis by CSI-D screening performance groups by study follow-up.
The age-specific and age-standardized annual incidence rates of CIND/MCI are presented in . The overall age-standardized annual incidence of CIND/MCI was 4.95% (95% confidence interval [CI] 3.39–6.52). The incidence rate increased with increasing age, from 3.43% (CI 1.17–5.69) in the young-old (age 65–74 years), to 6.44% (CI 4.18–8.69) in the old-old (75–84 years), and to 9.62% (CI 5.62–13.62) in the oldest-old (85+ years). The age-standardized rate of CIND/MCI as a function of gender indicate the rate for women (4.73 [CI 2.03–7.43]) is slightly but not significantly higher than the rate for men (4.35 [CI 1.98–6.72]). The CIND/MCI subtype of medically unexplained memory loss (combined single- and multi-domain amnestic MCI) was the most common subtype accounting for 53 or 60% of all incident cases and an standardized annual incidence of 3.67% (CI 2.75–4.48). The remainder of the incident cases was composed of subjects with CIND/MCI due to medical illness 11 cases, stroke/cerebrovascular disease 9 cases, alcohol/substance use 6 cases, and other/unknown causes 9 cases.
Age-specific annual incident rates of CIND/MCI.
shows a comparison of the age-standardized incidence rates for CIND/MCI and dementia in this cohort. CIND/MCI does have a slightly higher overall annual incidence rate than dementia (4.95% vs. 3.24%). The increase in rates is greater in the younger age groups after which they appear to converge.
Comparison of age-specific and age-standardized incidence rates of CIND/MCI and dementia in this cohort.
T-tests and Fisher’s exact tests were used to identify risk factors for incident CIND/MCI compared to the combined group of non-cases (those diagnosed as Normal and the good performance group who was not clinically assessed) using demographic and health history factors. The health history was drawn from self- and informant-report at the baseline CSI-D and included: age at diagnosis, gender, education, occupation, rural residence to age 19 years, alcohol use, smoking, body mass index, and history of diabetes, hypertension, heart attack, stroke, head injury, cancer, Parkinson disease, and depression. Significant univariate effects were noted for age, years of education, rural residence to age 19 years, white collar occupation, history of head injury, and history of depression. These variables were entered into a stepwise logistic regression model and the final results revealed that older age at diagnosis, history of head injury, and history of depression were risk factors for incident CIND/MCI while more years of education was protective (see ).
Final results from stepwise logistic regression (Outcome is CIND/MCI vs. Normal subjects and good performers).