Dementia is estimated to affect nearly 1 million people in the UK, with this number projected to rise to 1.4 million by 2040 (Alzheimer’s Society 2024). Dementia is also the leading cause of mortality in the UK, causing more than 12% of deaths in 2025 (Office for National Statistics 2026). As the scale of dementia increases, it is forecast that the already considerable economic burden will also increase, rising to £90 billion in the UK by 2040 (Alzheimer’s Society 2024). Early diagnosis and treatment can improve patient outcomes and help reduce the economic burden, predominantly through delaying care home admission.
Alzheimer’s disease is the most common cause of dementia, estimated to cause between 60 and 70% of cases (World Health Organization 2025). It is characterised by the pathological build up of amyloid-beta plaques and neurofibrillary tangles, and an associated clinical picture of progressive cognitive decline. Current guidance from the National Institute for Health and Care Excellence (NICE 2018) recommends treating mild to moderate Alzheimer’s disease with acetylcholinesterase inhibitors – either donepezil, rivastigmine or galantamine, the last of which is the subject of the review in this month’s Cochrane Corner, by Lim et al (Reference Lim, Schneider and Loy2024).
Galantamine was launched in Europe in 2000 and it works by increasing acetylcholine and modulating nicotine receptors. It is available in tablet, capsule and liquid form. In addition to the original twice-daily immediate-release formulation, there is now a once-daily prolonged-release capsule. The usual starting dose for the latter is 8 mg once daily and can be increased to 24 mg over time. Although the NICE guidance ((NICE 2018) suggests any acetylcholinesterase inhibitor as monotherapy, galantamine is often used second- or third-line where donepezil has failed, with local guidelines typically recommending donepezil first, because of its lower cost (Department of Health and Social Care 2026) and the simplicity of the titration and once-daily dosing (Jackson Reference Jackson, Ham and Wilkinson2004).
The review also considers mild cognitive impairment (MCI), a clinical syndrome that results in subsequent dementia in 10% of cases after 12 months. Individuals with MCI are a highly heterogeneous group and some even revert to ‘normal’ cognitive function. No medications are licensed for MCI in the UK, and NICE does not recommend drug treatment, but several related technology appraisals are underway.
The Cochrane Review
Lim et al’s (Reference Lim, Schneider and Loy2024) Cochrane Review updates an earlier review by Loy & Schneider (Reference Loy and Schneider2006).
The research question
The research question – regarding galantamine’s clinical effects and potential moderators of effect in their study sample – was clearly formulated using the PICO format (population; intervention; comparison; outcome), as shown in its ‘Summary of findings’ tables.
The study sample is defined as ‘people with dementia due to Alzheimer’s disease’ and ‘people with mild cognitive impairment’. The authors specify no criteria for MCI. Alzheimer’s disease is classified as ‘probable’ or ‘possible’ according to either the National Institute of Neurological and Communicative Diseases and Stroke/Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria, DSM-III-R or DSM-IV. Although standardised criteria aid comparability across studies, including participants with ‘possible’ Alzheimer’s disease in the meta-analysis introduces diagnostic heterogeneity, potentially reducing the internal validity of the review.
Eligible studies administered galantamine (any dose) to the treatment group and placebo to controls; cognitive function was the primary outcome.
Results
The identified studies
Following a comprehensive search strategy, 21 studies were identified for inclusion. All 21 were double-blind (Box 1), parallel-group randomised controlled trials (RCTs) comparing galantamine with placebo in people with either (a) possible or probable Alzheimer’s disease or (b) MCI. Of these, 11 studies had featured in the previous review (Loy Reference Loy and Schneider2006) and several had been published or updated since. Of the 21 studies, 2 did not provide sufficient outcome data to be included in the meta-analysis. Studies ranged in size from 18 participants recruited across two centres (Kadir Reference Kadir, Darreh-Shori and Almkvist2008) to 2051 participants recruited from 127 centres across 13 European countries (Hager Reference Hager, Baseman and Nye2014). The review authors note that 16 of the 21 studies were sponsored by pharmaceutical companies (Johnson & Johnson or its subsidiary, Janssen Pharmaceuticals, which was the original manufacturer of galantamine).
Double-blind studies
‘ Blinding’, often known as ‘masking’, is a technique used in clinical trials to reduce bias. It refers to those involved in the trial not knowing, or being ‘blind’ to, which group or ‘arm’ of the trial the participant has been assigned to. In a ‘double-blind’ trial, neither the participant nor the researchers know which arm of the trial the participant has been assigned to – for example, in the case of this Cochrane Review, whether the patient is taking galantamine or placebo.
Double-blinding is important because it reduces the likelihood that the hopes or expectations of those involved in the study can influence the results – either participants, who might perceive or report symptoms differently if they know they are taking a drug rather than a placebo or vice versa, or researchers, whose interpretation or assessment of findings might be similarly vulnerable to influence.
Double-blinding might last for the duration of a clinical trial or for a defined period, and this should be clearly defined in the study’s methodology.
The characteristics of study samples varied, with one study recruiting only people with severe Alzheimer’s disease (Burns Reference Burns, Bernabei and Bullock2009), whereas others included only participants with mild Alzheimer’s disease. Some studies (Burns Reference Burns, Bernabei and Bullock2009; Hager Reference Hager, Baseman and Nye2014) included participants with cerebrovascular disease, whereas Blautzik et al (Reference Blautzik, Keeser and Paolini2016) required radiological confirmation of its absence. Dosing regimens also differed significantly between studies, with administration one to three times daily, and use of both immediate- and prolonged-release formulations. Studies varied in length, with the longest lasting for 2 years, and in timing and nature of outcome measures. The review focused on cognitive function, which was the most reported outcome among the studies.
Analysis of findings
Risk of bias was assessed using six standard domains (randomisation, allocation concealment, masking (‘blinding’) of participants and personnel, masking of outcome assessors, completeness of outcome data, and selective reporting) applied to each of the 21 studies, giving 126 separate assessments. Of these assessments, 78 (61.9%) were judged low risk. Only five assessments across all studies were deemed high risk, four of which concerned incomplete outcome data (attrition bias) due to marked participant drop-out. Two of these four also used a ‘last observation carried forward’ (LOCF) approach (Box 2) in their intention-to-treat (ITT) analyses, which has the potential to bias outcomes in favour of treatment, given the progressive nature of cognitive impairment in dementia.
The ‘last observation carried forward’ approach
If participants in a clinical trial drop out before the trial has been completed, it is good practice to perform an intention-to-treat analysis, to minimise bias in reporting. In the ‘last observation carried forward’ (LOCF) approach, the researchers use the last data captured for that participant at all future end-points. This can potentially introduce bias into results, particularly for outcome data likely to have changed at points of further assessment (e.g. cognitive function in dementia).
Analyses were performed separately for each outcome and measurement instrument, with further subgroup analyses by dosing regimen. For the main outcome, cognitive function, the most often used instrument was the Alzheimer’s Disease Assessment Scale – Cognitive Subscale (ADAS-Cog) (Box 3). Except for the 6 mg thrice daily dosing at 3 months, all data demonstrated better cognitive outcomes with galantamine than with placebo, with a mean difference of 2.86 points (95% CI −3.29 to −2.43). Both ITT and observed case analyses support this conclusion, but the effect was stronger in the ITT analysis. The authors report that this effect meets the 2.6 point minimum clinically important difference (MCID) (Box 4). However, MCID estimates for the ADAS-Cog vary, and the threshold they use (from Watt Reference Watt, Veroniki and Tricco2021) is lower than many other published values, potentially overestimating the possible clinical benefit. Where the Mini-Mental State Examination (MMSE) was used, galantamine also slowed cognitive decline compared with placebo at 24 months (mean difference 0.73, 95% CI 0.34–1.12).
The ADAS-Cog
The Alzheimer’s Disease Assessment Scale – Cognitive Subscale (ADAS-Cog) (Rosen Reference Rosen, Mohs and Davis1984) is a measurement instrument purpose-designed for assessing cognitive function (memory, language, and praxis) in clinical trials for Alzheimer’s disease. Any such clinical trial taking place in the USA is mandated to use the ADAS-Cog. It can, however, also be used in clinical practice.
The instrument takes approximately 30 min to administer and would ordinarily be administered by a clinician or researcher. It includes 11 tasks and yields a score between 0 and 70, where higher scores indicate greater impairment.
Minimum clinically important differences
The term minimum clinically important difference (MCID) is used to describe the smallest possible change in an outcome that would be deemed meaningful to the patient. It is a patient-centred way of assessing outcomes and establishing whether they might hold any significance for patients in real-world clinical practice.
Galantamine outperformed placebo in slowing the decline of functional disability and behavioural function – however, the authors deemed the evidence on global function to be of low certainty (rated using GRADE criteria), owing to ‘substantial statistical heterogeneity’. In MCI, galantamine showed no clear cognitive or functional benefit over placebo. The MCI evidence was drawn from only three studies (2076 participants), all of which had discontinuation rates of over 40% and were thus judged by the authors to have high risk of attrition bias.
The review outcomes also included adverse events. Not all adverse events were reported in all studies – for example, only 12 studies reported on deaths during the study. Their analysis showed that participants with Alzheimer’s disease taking galantamine were more likely to stop treatment than those taking placebo (22.7% v. 17.2%, odds ratio 1.41) and were more likely than those taking placebo to report nausea (20.9% v. 8.4%, odds ratio 2.89). Both results were based on evidence from over 3000 participants, rated by the authors as ‘high certainty’.
Discussion
Risk of bias
This updated Cochrane review demonstrates the value of revisiting earlier work, with a significantly expanded data-set allowing more confident conclusions. The design of the review is clear and the overall quality of evidence is high, with a low risk of bias across most domains. The authors acknowledge important limitations, including the dominance of Johnson & Johnson and its subsidiary Janssen in funding the included studies. This could be considered a potential source of bias in favour of positive trial outcomes – however, such funding patterns are common in studies of pharmacological treatments and so this is not a unique challenge for this review. They also note that using LOCF in ITT analysis introduces a potential favourable bias for the active treatment if it has a higher drop-out rate than placebo and is especially problematic when cognitive function is the observation being carried forward to future time points in a dementia study, where progressive cognitive decline is expected.
One design choice made by the authors that might potentially introduce positive outcome bias is their selection of MCID cut-offs for the instruments used. The authors adopted relatively generous interpretations (for example, considering a mean MMSE difference of 0.73 points (95% CI 0.34–1.12) at 24 months as clinically meaningful). They discuss several MCID thresholds for the ADAS-Cog but select the lowest.
In addition, use of language when describing cognitive outcomes could be more precise: the authors refer to ‘gains of 0.3 to 1.1 on the MMSE scale at 24 months’ and to galantamine ‘improving’ cognitive function. In both cases, referring to a less significant decline in cognition would be more accurate.
Adverse effects findings
The review’s adverse effects findings are interesting: although galantamine has a high drop-out rate, the placebo drop-out rate was also high (17.2%), suggesting that factors beyond tolerability contributed to patients stopping trial medications. At 6 months, all-cause discontinuations and discontinuations due to adverse events were higher in the galantamine groups, demonstrating a clear dose–response effect; however, by 24 months, discontinuation rates were similar between galantamine and placebo.
Validity and usefulness of findings
There are several factors potentially limiting the validity and usefulness of this review’s findings in clinical practice. First, in routine care, the most often used acetylcholinesterase inhibitor is donepezil, with galantamine typically a second- or third-line agent. Although essential to know that galantamine outperforms placebo, direct comparison with donepezil and rivastigmine would also be very useful.
Furthermore, the patient samples in many of the included studies are not representative of typical memory clinic populations. Many studies excluded individuals with comorbidities, which would exclude most patients newly diagnosed with dementia in a memory clinic setting. In the UK, around 54% of people over 65 have two or more health conditions (Kingston Reference Kingston, Robinson and Booth2018), with multi-morbidity expected to rise. One study (Burns Reference Burns, Bernabei and Bullock2009) stands out for its more inclusive sample, including care home residents with severe dementia and a higher mean age of 84. However, including people with severe dementia is debatable, as in the UK acetylcholinesterase inhibitors are recommended for mild to moderate dementia, with memantine preferred in severe disease.
Conclusions
This Cochrane review clearly demonstrates that galantamine has some positive effects relative to placebo for people with Alzheimer’s disease. It remains to be seen whether these effects are meaningful from the patient’s perspective, and whether any benefits outweigh the side-effect burden, given that galantamine-treated participants experienced significantly higher rates of nausea than those receiving placebo.
Although this is a robustly designed and executed Cochrane Review, the limited availability of head-to-head trials means we still lack good comparative evidence between the different acetylcholinesterase inhibitors. This is an area where future research could make a particularly valuable contribution to clinical decision-making, helping ensure that we offer patients the best of these now well-established and widely available treatments.
Author contributions
R.B. drafted the initial manuscript. V.V. reviewed and edited the initial manuscript, as well as all subsequent versions. Both authors have reviewed and approved the final manuscript submitted.
Funding
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Declaration of interest
None.
eLetters
No eLetters have been published for this article.