Fifty-nine RCTs, reporting seventy-five different outcomes, were included in the review.
Donepezil versus placebo (24 RCTs, n=7,556)
A significant change in ADAS cognitive subscale (ADAS-cog) scores was found in favour of donepezil (10mg/daily) in patients with AD (all severity levels) (WMD -2.80, 95% CI: -3.28, -2.33; 5 RCTs) and in patients with donepezil mild/moderate VD (WMD -2.17, 95% CI: -2.99, -1.34; 2 RCTs). No evidence of significant statistical heterogeneity was found. No significant between group difference was found in mild cognitive impairment. Significant statistical heterogeneity was found (I2 = 75.5%). RRs for improvement from baseline (RR 2.01, 95% CI: 1.58, 2.57; 3 RCTs) and improvement or stabilisation from baseline (RR 1.50, 95% CI: 1.20, 1.89; 1 RCT) in patients with AD (all severity levels, based on the CIBIC-plus) were in favour of donepezil. No evidence of statistical heterogeneity was found. No significant between group difference was found on the CIBIC-plus in patients with mild/moderate VD. Evidence of moderate statistical heterogeneity was found (I2 = 55.1%). Borderline statistical significance was found in favour of donepezil for ADLs in patients with VD (WMD-0.78, 95% CI: -1.58, 0.01; 2 RCTs). No significant between group difference was found for behaviour (Neuropsychiatric Inventory (NPI)). Donepezil was associated with diarrhoea, nausea, anorexia, dizziness, vomiting in patients with Alzheimer dementia. For vascular dementia, it was associated with abnormal dreams, diarrhoea, nausea, and muscle and leg cramp in patients. In general, the quality of reporting harms was low to moderate.
Donepezil versus galantamine (2 RCTs, n=251) or rivastigmine (1 RCT, n=not reported)
One study found statistical differences in favour of galantamine in a subgroup of patients (MMSE scores between 12 and 18) for changes in cognition (ADAS-cog and MMSE) and scores on the Screen for Caregiver Burden. The second study was pilot study (results not reported). Adverse events most frequently reported were nausea, vomiting, agitation, headache and falls. The rates for all these harms were marginally higher in the galantamine group. No difference was found in frequency of serious events.
When donepezil was compared with rivastigmine in patients with moderately severe AD, no significant between group difference was found on measures of cognition and behaviour, but statistically significant changes in global function were found in favour of rivastigmine. Subgroup analysis found significant differences on some measures of behaviour and function in patients 75 years or older compared to younger patients. In general, patients receiving rivastigmine reported more adverse events but the frequency of serious events did not differ between groups. The quality of the harms was well reported.
Galantamine versus placebo (10 RCTs, n=3997)
Significant improvement on the ADS-cog score was found in favour of galantamine (24mg/daily) in patients with mild/moderate AD (WMD -2.45, 95% CI:-3.48, -1.42; 6 RCTs) and in patients with AD and VD (WMD -2.70, 95% CI: -3.95, -1.45; 1 RCT). Evidence of significant statistical heterogeneity was found for mild/moderate AD (I2 = 75.5%). The RR for improvement or stabilisation from baseline (CIBIC-plus) was statistically significant in patients with mild/moderate AD (RR 1.22, 95% CI: 1.12, 1.33; 4 RCTs) and mild/moderate AD or VD (RR 1.25, 95% CI1.08, 1.45; 1 RCT). No evidence of significant statistical heterogeneity was found. A significant improvement in ADLs were found in favour of galantamine in patients with AD (all severity) (WMD 1.84, 95 % CI: 0.68, 3.00); no evidence of significant statistical heterogeneity was found. Galantamine was associated with anorexia, dizziness, nausea, vomiting and weight loss. Rates of withdrawal due to adverse events ranged from 4 to 17% in the placebo group and 8 to 54% in the galantamine groups. Most studies did not report using a standardised instrument to collect harms data.
Rivastigmine versus placebo (9 RCTs, n= 2164)
Significant statistical improvement on the ADAS-cog was found in favour of rivastigmine (6mg/daily and 12mg/daily) in patients with AD (all severity levels) (WMD -3.91, 95% CI: -5.48, -2.34). However, evidence of significant statistical heterogeneity was found (I2 = 90.8%). A significant improvement from baseline on the CIBIC-plus in favour of treatment was found when compared with placebo for patients with AD (all severity levels) (RR 1.76, 95% CI: 1.35, 2.29; 3 RCTs). One study looked at improvement or stabilisation on the same scale in the same patient group but no significant between group difference was found. Statistically significant effect in favour of rivastigmine was found on the Nurses Observation Scale for Geriatric Patients mood subscale (NOSGER-mood) (WMD 3.75, 95% CI: 2.66, 4.85) and the geriatric depression Scale (GDS) (WMD 0.22, 95% CI: 0.15, 0.28); evidence of moderate statistical heterogeneity was found on the NOSGER-mood (I2=36.5%). Rivastigmine was associated with increased abdominal pain, anorexia, dizziness, fatigue, headache, malaise, nausea and vomiting. Rates of withdrawal due to adverse events ranged from 0 to 11% in the placebo groups and 12 to 29% in the rivastigmine groups. Quality of reporting harms varied widely.
Tacrine versus placebo (6 RCTs, n=1203)
One trial found a statistically significant improvement on ADAS-cog for patients receiving tacrine. No significant between group differences were found in the other trials. Elevated alanine aminotransferase levels or other hepatic abnormalities were reported in the treatment group compared to controls in 6 trials. Tacrine was associated with gastrointestinal problems, dizziness, nausea and vomiting. The proportion of patients withdrawing due to adverse events ranged from 0 to 12% in the placebo group compared with 0 to 55% in the tacrine groups. Overall, the quality of procedures used to collect harms data was moderate.
Tacrine versus idebenone (1 RCT, n=not reported)
Results not reported.
Memantine versus placebo (5 RCTs, n=1944)
Significant improvement on the ADAS-cog was found for patients with mild to moderate VD receiving memantine compared with placebo (WMD -2.20, 95% CI: -3.24, -1.15; 2 RCTs). The authors state that a significant difference was found on the CIBIC-plus in patients with Alzheimer dementia and vascular dementia, and that sensitivity analysis did not significantly change the summary effect size, however, only the results of one subgroup are presented (AD (all severity levels), RR 1.25, 95% CI: 1.35, 2.29). No evidence of statistical heterogeneity was found. A significant difference in favour of memantine was found on the NOSGER-mood (VD, all severity levels) and the NPI (AD, all severity levels); no evidence of statistical heterogeneity was found. Adverse events included gastrointestinal symptoms, dizziness, and headaches in patients receiving memantine. However, agitation was found to be more frequently reported in the placebo group.