Using t tests and chi-square tests, demographic and
clinical variables were compared between subjects with CDR Global 0 and those
with CDR Global 0.5. We also compared the two CDR groups by cognitive and
retention status: cognitively stable versus cognitive decline groups (cognitive
decline group = subjects who had a “cognitive evaluation trigger”
at any time point) and “completers” versus
“non-completers”(non-completer = subjects who did not complete all
four annual visits). Inter-class coefficients (ICC) were calculated to assess
three-month test-retest reliability for subject and study partner CFI. Ten
thousand bootstrap samples were generated to compare the ICCs between subject
and study partner CFI. Generalized estimating equation (GEE) method was used to
examine the extent to which CFI score predicted cognitive decline as measured by
the presence and absence of a “cognitive evaluation trigger,”
while controlling for other variables and accounting for the within-subject
correlation. To examine the short-term prediction ability of the CFI, the CFI
score from the previous year and the change in CFI score from current year to
previous year were used as independent variables. mMMSE and FCSRT scores in the
previous year, CDR Global score and age at baseline, ethnicity, and
subjects’ primary language were used as covariates in the GEE analysis.
To examine the long-term prediction ability of the CFI, the CFI score from two
years ago and the change in CFI score from current year to two-year-ago were
used as independent variables. mMMSE and FCSRT scores from two years prior, CDR
Global score and age at baseline, ethnicity, and subjects’ primary
language were used as covariates in the GEE analysis. Statistical significance
was defined a priori at p = 0.05.
clinical variables were compared between subjects with CDR Global 0 and those
with CDR Global 0.5. We also compared the two CDR groups by cognitive and
retention status: cognitively stable versus cognitive decline groups (cognitive
decline group = subjects who had a “cognitive evaluation trigger”
at any time point) and “completers” versus
“non-completers”(non-completer = subjects who did not complete all
four annual visits). Inter-class coefficients (ICC) were calculated to assess
three-month test-retest reliability for subject and study partner CFI. Ten
thousand bootstrap samples were generated to compare the ICCs between subject
and study partner CFI. Generalized estimating equation (GEE) method was used to
examine the extent to which CFI score predicted cognitive decline as measured by
the presence and absence of a “cognitive evaluation trigger,”
while controlling for other variables and accounting for the within-subject
correlation. To examine the short-term prediction ability of the CFI, the CFI
score from the previous year and the change in CFI score from current year to
previous year were used as independent variables. mMMSE and FCSRT scores in the
previous year, CDR Global score and age at baseline, ethnicity, and
subjects’ primary language were used as covariates in the GEE analysis.
To examine the long-term prediction ability of the CFI, the CFI score from two
years ago and the change in CFI score from current year to two-year-ago were
used as independent variables. mMMSE and FCSRT scores from two years prior, CDR
Global score and age at baseline, ethnicity, and subjects’ primary
language were used as covariates in the GEE analysis. Statistical significance
was defined a priori at p = 0.05.
