Nine studies (n=14,027) were included in the review: 7 prevention studies (n=13,920) and 2 treatment studies (n=107). The prevention studies comprised 4 cohort studies and 3 case-control studies; the treatment studies were both RCTs.
Unadjusted pooled data (4 cohort and 2 case-control studies) showed a significantly greater incidence of dementia in participants who had not used statins in comparison with those who had (OR 0.67, 95% CI: 0.54, 0.82, p=0.042). When pooled adjusted RRs were calculated, no significant differences were apparent when using either fixed-effect or random-effects models. However, significant heterogeneity was detected in the pooled adjusted RRs. There were no significant differences between statin users and non-users in the incidence of AD, based on unadjusted and adjusted pooled effect sizes using random-effects and fixed-effect models. Statistical tests for heterogeneity were not significant. When studies were grouped according to study design, significant differences in favour of statins were found for case-control studies of dementia (RR 0.28, 95% CI; 0.15, 0.54) and case-control studies of AD (RR 0.61, 95% CI: 0.42, 0.88); no statistically significant differences in the incidence of dementia or AD were observed between statin users and non-users in cohort studies.
One RCT of atorvastatin (80 mg) showed a significant difference in ADAS-cog scores at 6 months in favour of the statin, compared with placebo, in patients with mild to moderate dementia (p<0.03); however, no significant differences were present at 12 months (p=0.055). In the second RCT, no significant differences in ADAS-cog scores were found between simvastatin (20 to 80 mg) and placebo after 26 weeks. Overall, the pooled mean difference in ADAS-cog scores was not significantly different between statin and placebo groups.