Pharmacological Treatment of Dementia: A Scoping Review of Systematic Reviews

Dementia is a group of chronic diseases characterized by a constant decline in the function of multiple cognitive domains comprising memory impairment, behavioral problems, loss of initiative, loss of independence in daily activities and loss of participation in social activities. It can be due to the direct physiological effects of a general medical condition, to the persisting effects of a substance or to multiple etiologies [1]. These problems with cognitive function decrease the quality of life of dementia patients and their caregivers, and put pressure on family relationships and friendships [2]. It is a highly prevalent condition, with Alzheimer's disease (AD) being the most common cause. The prevalence of dementia increases from 0.9% in 65- to 69-year-olds to over 30% in patients aged 85 years and older [3]. According to the World Health Organization, the total number of people with dementia worldwide in 2010 is estimated at 35.6 million and is projected to nearly double every 20 years. Between 2000 and 2008, deaths due to AD have risen 66% in the United States [2]. The population-attributable risk of AD regarding mortality over 5 years in people aged 65 years is estimated to be between 5 and 15% [4]. The disease causes a high financial burden on health care services: the current global costs involved in dementia are estimated to be more than USD 600 billion per year [2].

One of the most important issues is that patients with dementia cannot be cured, but the process of cognitive deterioration can merely be delayed. In many countries, cholinesterase inhibitors and memantine are registered for the treatment of cognitive impairment in AD. Furthermore, a wide range of medication is used for the behavioral and psychological symptoms in dementia. In the past few years, a large amount of research has been published at a rapid rate concerning different aspects of the disease, ranging from diagnostic tests and treatment options to the organization of care.

To develop or practice best care, clinicians and guideline makers often use systematic reviews since these summarize available evidence in a systematic way with enhanced precision. A large number of systematic reviews on diagnostic and medical interventions for dementia are available. However, these separate systematic reviews do not directly provide overarching insights into the extent and range of established evidence in a specific field.

This scoping review was carried out for the Healthcare Insurance Board, a consulting agency that advises the Dutch government on implementing the Dutch statutory health insurance [5]. The aim was to get an overview of currently registered pharmacological therapies for dementia about which systematic reviews are available. An evidence-based approach offers the most objective way to determine high quality and safety standards in clinical practice. It facilitates the process of transferring results of clinical research to practice and it has the potential to reduce healthcare costs, for example by disinvestment of ineffective practices.

In this scoping review, we provide an overview of available systematic reviews on the pharmacological treatment of the most prevalent forms of dementia and identify the fields in which more research is necessary.

A scoping study is a relatively new methodology to systematically review an extensive body of literature that addresses a broad research question [6]. While a systematic review focuses on retrieving an answer to a well-defined question, a scoping review ‘maps' the relevant literature in a complete field of interest and describes only the main findings. According to Arksey and O'Malley [7], the aim of a scoping review is not to analyze or draw conclusions. Consequently, the present work should be seen as a first step to provide an overview of the existing literature and to identify the fields in which more research might be necessary in the future.

Our study comprised the following steps: identifying the clinical question, searching for relevant reviews, appraising the quality of the reviews [with the Scottish Institute of Guidelines Network (SIGN) checklist], categorizing the topics of the included systematic reviews and consulting experts in the research fields [6,7].

We performed a systematic search for systematic reviews published in MEDLINE, EMBASE, CINAHL and PsychINFO in September 2011 and in the Cochrane Database of Systematic Reviews (CDSR) in May 2012. The search was based on a search strategy developed by Kroes et al. [8]. Terms for dementia were combined with names of dementia medications, and the search terms were retrieved with the program PubReMiner, a text-mining program that gives a frequency analysis of used terms in titles and abstracts of the identified reviews [9]. We used a sensitive systematic review filter of the SIGN [10] to further specify the search. The search strategy is available from the authors.

We included systematic reviews of randomized controlled trials (RCTs), clinical controlled trials or observational studies that investigated a pharmacological intervention for dementia. Eligible reviews had to include patients with dementia [AD, vascular dementia, Lewy body disease (DLB) or Parkinson dementia, frontotemporal dementia and a category named ‘unspecified dementia' that contained dementia in general or mixed forms]. In addition, the reviews had to be published or updated between 2008 and December 2011, written in English, French, German or Dutch and the full text article had to be accessible. For updated Cochrane reviews, the original publication dates were used. Reviews were excluded when (a) they assessed patients with mild cognitive impairment or dementia in Down syndrome or AIDS; (b) the medication was not registered by the European Medicines Agency (EMA) [11] or the Food and Drug Association (FDA) [12] (e.g. experimental medications or alternative drugs), and (c) the intervention only aimed at the outcomes of caregivers. One reviewer screened the titles and abstracts of the references for eligibility, two reviewers independently selected the full text articles, and a third reviewer was consulted when the first two reviewers were uncertain about the inclusion of a review article.

The methodology checklist for systematic reviews and meta-analyses of the SIGN (five items) was used to assess the risk of bias of the included reviews [10]. The items of this checklist were adapted to our topic and summarized in a total estimation of the risk of bias (table 1). A detailed description of the quality assessment is available from the authors. Two researchers independently performed the quality assessment, and when necessary, disagreements were resolved through discussion with a third reviewer.

A standardized data extraction form was used to systematically extract the data from reviews with a low risk of bias (++ or + according to the SIGN checklist; table 1).

To provide an overview of the subjects included in the systematic reviews, we developed a matrix (table 2a-d). Each part of table 2 describes one type of dementia [unspecified dementia (including mixed types), AD, Parkinson dementia and DLB as well as vascular dementia]. Then, we categorized the medications into mechanisms of action or drug classes and made categories of symptoms that were addressed in the reviews. We put all these data in the matrix; by doing this per intervention it became clear whether a systematic review was available for the outcome. The categories were approved by experts in the different fields. For the high-quality reviews, we quoted a sentence from the abstract that summarizes the conclusions of the articles (table 3).

An expert panel was involved to comment on the findings (i.e. whether they were consistent or conflicting with current practice) and to indicate if they missed relevant topics. In addition, we asked the experts which relevant practical developments they expect for the near future and for which topics they would advise future research.

Based on project time lines and cost considerations, we recruited several experts from different medical fields and geographical regions in the Netherlands. Our expert panel, composed of a neurologist, a psychiatrist, a specialist for internal medicine, a geriatrician and a caregiver of a dementia patient, commented on the steps in several phases of the review process by personal communication, e.g., via e-mail contact. These expert-based comments were added to the overview of systematic reviews.

The findings are presented as an overview of available systematic reviews and the fields in which more research is necessary. Some reviews are mentioned more than once because they cover more than one topic. More details of the reviews included in this scoping study can be obtained from the authors.

The search yielded 1,569 records. Based on title and abstract, we read the full texts of 119 articles. In the end, 62 articles, including 34 Cochrane reviews, fulfilled the predefined inclusion criteria (fig. 1). The most prevalent reason for excluding a review was that the reported intervention was not FDA or EMA registered. Out of the 62 reviews, 55 (90%) reviews were assessed as having a low risk of bias. To describe the results, only the reviews with a low risk of bias were used.

Out of 62 reviews, 7 were considered as having a high risk of bias due to multiple flaws in the methodology (fig. 1). Fifty-five reviews scored a low risk of bias (++ or + according to the SIGN checklist). Of these, 34 received ++ and 21 received +. However, 13 of the + reviews lacked an assessment of the risk of bias of included studies, which means that the conclusions of these reviews are less reliable. The 55 low-risk-of-bias reviews can be found in table 2.

The 55 reviews with a low risk of bias mainly addressed the fields of Alzheimer (n = 22) and unspecified dementia (including mixed forms, n = 29; table 2a, b). Surprisingly, treatment for Parkinson dementia/DLB and vascular dementia was assessed in only 3 and 6 reviews, respectively, and frontotemporal dementia in none (table 2c, d). The numbers increased to over 56 because some reviews addressed more than one type of dementia [13,14,15]. Most included reviews reported cognitive decline (n = 33), behavioral symptoms (n = 21) and adverse events (n = 26) as primary outcomes.

AD and Parkinson Dementia/DLB

For cognitive decline in AD, 17 reviews were available. The reviews by Birks et al. [16,17,18] showed that cholinesterase inhibitors are efficacious for mild to moderate AD as well as for Parkinson dementia [19,20], and they are cost-effective [17,21]. With regard to the effects, the different drugs are comparable, but with respect to the side effects, donepezil might be the better choice [22].

Memantine proved to be effective for patients with moderate to severe AD, although not for other forms of dementia [23]. For many interventions such as aspirin, steroids and NSAIDs [24,25,26,27], nicotine [28], hormone replacement [29] and thiamine [30], the evidence for an effect on cognitive decline is limited because of the lack of studies of high methodological quality and sufficient power. For Ginkgo biloba, the results are conflicting [13,31], although it appears to be safe [32].

Vascular Dementia

For vascular dementia, the efficacy of galantamine has been tested in 2 studies; 1 found some evidence of benefit, whereas the other did not find clear advantage over the placebo [33]. For rivastigmine, no proper RCTs were available [34]. Therefore, cholinesterase inhibitors are not recommended for vascular dementia.

Unspecified Dementia

Nine reviews about cognitive decline in unspecified dementia were identified. For antihypertensives, there are indications that they are beneficial [14,35]. No evidence was found to support the use of piracetam [36], melatonin [37], vitamins [38,39,40], statins [41] and antidepressive agents [42].

For this topic, no distinction between the different types of dementia was made. Twenty-seven systematic reviews described interventions for behavioral problems such as agitation, aggression and behavioral and psychological symptoms of dementia (BPSD). According to McShane et al. [43], the use of memantine resulted in a consistent, small reduction in the incidence of agitation in demented patients. However, there was no available evidence addressing the question as to whether prevalent agitation can be treated with memantine. No studies or only studies of low methodological quality were available for pain medication [44,45], and further studies are needed to establish the efficacy of cyproterone [46]. Melatonin has an effect in sundowning in demented patients [47], but trazodone [48], valproate [49] and haloperidol [50] have not shown any treatment effect for agitation among demented patients, whereas they have potential side effects [51,52,53].

For this topic, no distinction between the different types of dementia was made. Eight reviews described neuropsychiatric symptoms such as hallucinations and delusions. No evidence was found for an effect of anticonvulsant mood stabilizers [54] and second-generation antipsychotics (SGAs) [53]. Moreover, SGAs increase the risk of falling in nursing home residents [51] and of cerebrovascular events [52]. Two reviews found that antidepressants can be effective in the treatment of BPSD [55,56], whereas Martinón et al. [48 ]found no effect of trazodone.

The number of reviews about depression in demented patients is small. Only 1 review showed weak support for antidepressants being effective in patients with depression and dementia (type not specified) [42].

Activities of daily living, institutionalization and quality of life, although important from the patient's point of view, were only investigated as secondary outcomes. Table 2a-d shows the reviews describing these outcomes.

The expert panel commented on the matrix (table 2a-d) and on the conclusions of the reviews (table 3). Furthermore, we asked the experts about promising interventions and implications for further research regarding the pharmacological treatment of dementia.

There are sufficient systematic reviews about the efficacy of cholinesterase inhibitors for cognitive impairment, mild to moderate stadia of AD and Parkinson dementia as these are most frequently used in current practice. In addition, various systematic reviews about the efficacy of memantine on cognition are available. For several other interventions such as NSAIDs, selegiline and hormone replacement, there is sufficient evidence to show that these are not effective.

Furthermore, the panel was asked about the topics they missed in the current literature on dementia. Little is known about the combination of cholinesterase inhibitor therapy with memantine, although recently a primary study has been published on this topic [57]. Neither is it known whether cholinesterase inhibitors are effective in combination with other drug types such as antihypertensive drugs. Furthermore, which cholinesterase inhibitor is preferable in the light of efficacy and side effects is still a topic of discussion, and little evidence is available regarding the continuation or discontinuation of these medications when the disease progresses [57]. Therefore, our expert team suggested that comparative effectiveness research should be performed instead of more placebo-controlled trials on these types of drug therapy.

It is striking that often drugs are prescribed for symptoms of demented patients, while aggregated evidence is lacking. For example, few reviews were published on the treatment of depression in dementia although depression is highly prevalent in dementia [58]. In addition, although epidemiological studies show that the number of patients with AD affected by delirium varies from 22 to 89% in the community and hospitalized populations [59], it is another striking example of a syndrome that is hardly studied in these patients. Likewise, few reviews address the ability of patients to live independently or to perform their activities of daily living without help, although these outcomes are likely to be meaningful for patients as well as their caregivers [60].

New research should focus on the symptomatic treatment of the earliest and most salient complaints in AD and on disease-modifying interventions acting at the level of the amyloid cascade. Promising nutritional interventions, immunotherapy acting at the level of the amyloid cascade, drugs that target tau, progranulin or TDP-43 are missing in this overview because these interventions are not yet registered for clinical use or are still in the development phase [61]. Consequently, systematic reviews on these topics have not been performed yet.

We performed a scoping review that aimed to give an overview of the subjects and methodological quality of the currently available systematic reviews on the pharmacological treatment of the most prevalent forms of dementia. In addition, it identifies gaps in the existing literature and facilitates dissemination.

Our scoping review is the first systematic overview of recent systematic reviews regarding the pharmacological treatment of dementia. It shows that in the field of dementia, most research is designed to determine the effects of cholinesterase inhibitors, NMDA receptor antagonists and antipsychotics (table 2). Furthermore, our scoping review also shows that for many indications, no systematic review is available. Without aggregated evidence, it is difficult and time-consuming to provide evidence-based care to patients.

Several topics, such as the treatment of cognitive impairment in AD with cholinesterase inhibitors, have been thoroughly studied over the past few years. Starting up new projects on these topics thus is of little value. Head-to-head comparison trials to find out which intervention is best and when to discontinue such interventions are of greater interest for future research. Such research might better be performed as coordinated international prospective cohort studies in daily practice, as RCTs are expensive and time-consuming.

Many reviews about the treatment of behavioral problems have been published. From the available systematic reviews, it appeared that these kinds of symptoms are very difficult to treat with medication; moreover, currently used therapies such as antipsychotics have major side effects. Perhaps the management of these problems should rather be nonpharmacological, with interventions such as environmental modification, task simplification and appropriate activities [62,63].

Our scoping review has a number of strengths. First, it provided a broad overview of both the recent systematic reviews available and the gaps in the literature concerning the pharmacological treatment of dementia. Second, we performed an extensive search in several databases; thereby, the chance of missing relevant publications was reduced. Furthermore, 2 independent assessors selected the reviews and determined the risk of bias, which reduced the chance of errors.

However, there are also some limitations worth noting. Although in general the quality of the systematic reviews in the field of pharmacological treatment options for dementia was high, the quality assessment of the primary studies was lacking in many cases. Whether the results of a systematic review can be relied on for clinical practice depends on two factors: its own risk of bias and the risk of bias of the included studies. It has been shown that even systematic reviews published in leading journals often have important methodological limitations potentially leading to biased results and different answers to the same question [64]. However, we have chosen to extract the data from these reviews as well because apart from such biased results, the reviews had been performed well and no alternative evidence with a smaller risk of bias was available.

For identifying the gaps in the literature, we relied on the expert panel. Although this expert panel consisted of prominent professionals from different branches involved in the treatment and care of demented patients, expert opinions are also dependent on personal interests and knowledge. Besides, all experts originated from the Netherlands.

We have summarized the results in a narrative way and thereby relied on the reports of the authors. Investigating the extent of the drug effects and assessing their clinical relevance was beyond the scope of this overview. We checked for the presence of results regarding efficacy by looking at evidence from placebo-controlled trials and evidence regarding the difference in effect between various drugs. We also looked for evidence concerning side effects, but other factors such as pharmacoeconomics and implementation research are not described. Furthermore, because the aim of this scoping review was to give an overview of available reviews, new and unregistered medication is not reported. However, we have covered this topic in the expert opinion part.

This scoping review is a comprehensive overview of the currently available systematic reviews on the pharmacological treatment of dementia and thus a starting point for clinicians and guideline developers to find systematic reviews regarding various pharmacological treatments for dementia. It shows the gaps in the existing literature and indicates where future reviews and/or primary research might be necessary. When used as a starting point for guideline development, not only treatment effects are necessary. It is also important to know if the evidence identified is applicable to the patients to whom the guideline is directed (e.g. by looking at the results of comparative effectiveness research) and to consider other factors such as the patient perspective, available resources and local circumstances. Finally, this scoping review may guide the development of quality indicators, which are used more and more in implementing guidelines and assessing the value of treatment components.

We would like to thank Miranda Langendam, Pauline Heus and Roy Elbers from the Dutch Cochrane Centre for their contribution to the design and methods of this study. We would also like to thank René Spijker for his help with the development of the search strategy, and Prof. Dr. Raymond Koopmans for his clinical input as a member of the expert panel.

None of the authors has personal, financial or potential conflicts of interest.

This scoping review was performed for the Healthcare Insurance Board (www.cvz.nl), but the organization was not involved in the research process and the writing of the manuscript.