Statistical analysis

Data analysis was performed using the Statistical Package for Social Sciences software (Standard Version 11.5.0; SPSS Inc., Chicago, IL, USA). All statistical tests were 2-tailed, and an alpha value of .05 indicated significance. The t test was used to assess differences between the two independent groups (i.e., the favorable and poor response groups) regarding age, educational level, initial CDR-SB score, initial MMSE score, second CDR-SB score, and second MMSE score. The chi-squared test was used to compare the case–control and therapeutic groups with regard to the ApoE ε4 genotype and sex. In addition, the chi-squared test was used to compare the therapeutic groups with regard to cilostazol use.

Multiple logistic regression models were fit to the data to calculate the odds ratios (ORs) and 95% confidence intervals (CIs) of the association between the therapeutic response and cilostazol use. This model was adjusted for age, sex, educational level, initial MMSE score, initial CDR-SB score, and ApoE ε4 status.

The dependent variable in each logistic regression model was the response (favorable or poor), and either of the therapeutic indicators was examined separately. Independent variables, including age, educational level, initial CDR-SB score, and initial MMSE score, were treated as continuous variables with 1-year increments for age and educational level and 1-score increments for CDR-SB and MMSE scores. This contrasted with the dichotomous categorical variables including cilostazol use, sex, and ApoE ε4 status. The R squared for the logistic regressions is 31.12%. The lack of fit chi-squared is not significant (Prob > ChiSq = 0.2791).