Data analysis
Pearson chi-square tests were used for discrete
variables and Student’s t-tests for continuous variables,
separately, to compare the differences
between dementia cases and matched controls. The
examined variables included: medical comorbidities,
psychiatric illnesses, and health service utilization,
individually. Next, we applied conditional
logistic regression to examine the associations
between bipolar disorder and the risk for dementia
with and without covariates adjustment.
Covariates included in the multiple conditional
logistic regression models were those that showed
a significant crude odds ratio (OR) in univariate
conditional logistic analyses. To test whether there