Comparative Effectiveness and Safety of Cognitive Enhancers for Treating Alzheimer's Disease

Systematic Review and Network Metaanalysis

Andrea C. Tricco, PhD; Huda M. Ashoor, BSc; Charlene Soobiah, HBSc; Patricia Rios, MSc; Areti Angeliki Veroniki, PhD; Jemila S. Hamid, PhD; John D. Ivory, MSc; Paul A. Khan, PhD; Fatemeh Yazdi, MSc; Marco Ghassemi, MSc; Erik Blondal, HBSc; Joanne M. Ho, MD; Carmen H. Ng, MSc; Brenda Hemmelgarn, MD; Sumit R. Majumdar, MD; Laure Perrier, PhD; Sharon E. Straus, MD


J Am Geriatr Soc. 2018;66(1):170-178. 

In This Article

Abstract and Introduction


Background/Objectives To examine the comparative effectiveness and safety of cognitive enhancers for Alzheimer's disease (AD).

Design Systematic review and Bayesian network metaanalysis (NMA).

Setting MEDLINE, EMBASE, Cochrane Library, CINAHL, Ageline (inception–March 2016).

Participants Individuals with AD in randomized controlled trials (RCTs), quasi-RCTs, and nonrandomized studies.

Intervention Any combination of donepezil, rivastigmine, galantamine, or memantine.

Measurements Two reviewers independently screened titles, abstracts, and full-texts; abstracted data; and appraised risk of bias.

Results Twenty thousand three hundred forty-three citations were screened, and 142 studies were included (110 RCTs, 21 non-RCTs, 11 cohort studies). NMA found that donepezil (Mini-Mental State Examination: mean difference (MD) = 1.39, 95% credible interval (CrI) = 0.53–2.24), donepezil+memantine (2.59, 95% CrI = 0.12–4.98), and transdermal rivastigmine (2.02, 95% CrI = 0.02–4.08) improved cognition more than placebo. NMA found that donepezil (Alzheimer's Disease Assessment Scale–cognitive: MD = −3.29, 95% CrI = −4.57 to −1.99) and galantamine (MD = −2.13, 95% CrI = −3.91 to −0.27) improved cognition more than placebo. NMA found that donepezil+memantine (MD = −5.23, 95% CrI = −8.72 to −1.56) improved behavior more than placebo. NMA found that donepezil (MD = −0.32, 95% CrI = −0.46 to −0.19), donepezil+memantine (MD = −0.57, 95% CrI = −0.95 to −0.21), oral rivastigmine (MD = −0.38, 95% CrI = −0.56 to −0.17), and galantamine (MD = −3.79, 95% CrI = −6.98 to −0.59) improved global status more than placebo. NMA found that galantamine decreased the odds of mortality (odds ratio = 0.56, 95% CrI = 0.36–0.87). No agent increased risk of serious adverse events, falls, or bradycardia. Some increased risk of headache (oral rivastigmine), diarrhea (oral rivastigmine, donepezil), nausea (oral rivastigmine, donepezil, galantamine), and vomiting (oral rivastigmine, donepezil, galantamine).

Conclusion An exhaustive review of the literature involving 142 studies demonstrated that cognitive enhancers in general have minimal effects on cognition according to minimal clinically important difference and global ratings. The drugs appear safe, but this must be interpreted cautiously because trial participants may have less comorbidity and fewer adverse effects than those treated with these drugs in clinical practice.


Alzheimer's disease (AD) is the commonest cause of dementia, affecting 46 million people worldwide in 2015.[1] The economic burden of AD is substantial, at an estimated $818 billion, or more than 1% of global gross domestic product. It is expected that the economic burden of AD will increase along with the anticipated rise in prevalence.[1,2]

The cholinergic deficit hypothesis for AD has been a major therapy target.[3] Cholinesterase inhibitors (donepezil, galantamine, rivastigmine) are believed to improve cognitive impairment in mild to moderate AD by inhibiting neuronal acetylcholine breakdown.[4] Galantamine is unique in its additional ability to modulate nicotinic receptor activity.[4] Another hypothesis for neuronal loss and subsequent cognitive impairment in AD is glutamatergic-mediated excitatory cytotoxicity.[3] The cognitive enhancer memantine is believed to block flow-through channels of N-methyl-d-aspartate receptors–glutamate receptors involved in cognition.[5]

Although the safety and efficacy of cognitive-enhancing medications have been examined in previous systematic reviews, only randomized controlled trials (RCTs) were included, limiting applicability of the findings to real world people. Furthermore, a network metaanalysis (NMA) has not been conducted. NMA provides summary treatment effects for pairwise treatment comparisons even if these treatments have never been compared directly in a study and provides the opportunity to rank each included treatment for the outcome examined.[6–8] NMA involves the simultaneous analysis of direct evidence from head-to-head comparisons of the treatments of interest and indirect evidence–when the treatments of interest are compared through one or more common treatment comparators.[9,10] For example, consider the scenario of a head-to-head study directly comparing Treatment A with placebo and a head-to-head study directly comparing Treatment B with placebo. An indirect treatment effect estimate comparing A with B can be inferred based on the treatment effects of A versus placebo and B versus placebo (where placebo is the common treatment comparator). The potential utility of NMA and the generalizability of its results rest on the validity of the required assumptions.[11] Transitivity is a critical NMA assumption that the included trials have similar characteristics across treatment comparisons.[6,12,13] The statistical notion of transitivity is the consistency assumption, in which different sources of evidence (e.g., direct and indirect evidence) are in agreement before analyzing them jointly. The validity of the consistency assumption is fundamental to the reliability of the results. NMAs are being found in increasing numbers in the literature because of their ability to examine all potential treatments.[14] We examined the comparative effectiveness and safety of these agents in the treatment of individuals with AD through a systematic review and NMA.