Ginkgo biloba for cognitive impairment and dementia

Objectives

This is a protocol for a Cochrane Review (intervention). The objectives are as follows:

To assess the efficacy and safety of Ginkgo for the treatment of people with diagnoses of cognitive impairment or dementia.

Background

Description of the condition

By 2050 the world’s population aged 60 and older is expected to total 2 billion (Alzheimer's Disease International 2015). While this represents a triumph for population health it carries an inherent set of challenges – adapting the current healthcare system to the needs of this burgeoning cohort. A key area of focus for policy, research and care services is addressing prevention and management of dementia and cognitive decline associated with aging.

Alzheimer’s disease has recently advanced from the seventh to the sixth leading cause of death in the USA (Johnson 2014). The estimated USA prevalence is 5.5 million, a number that is expected to triple to 16 million by 2050 (Dawson 2017). The disease accounted for USD (US dollar) 259 billion of USA expenditures on health care in 2016 with more than USD 109 billion of direct costs, which is greater than heart failure and cancer combined (WHO 2018). More than 47 million people have the disease worldwide, with an expected prevalence of 131.5 million people by 2050, most of whom will be living in the developing world (Alzheimer's Disease International 2015).

Dementia is not a single entity. It encompasses a variety of clinical syndromes which are most commonly due to progressive, neurodegenerative diseases. In the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM‐5), the term 'dementia' has been replaced by 'major neurocognitive disorder'. Diagnosis requires evidence of cognitive decline that is not due to another mental disorder and is severe enough to interfere with independent functioning (APA 2013). The commonest subtype of dementia is that due to Alzheimer's disease. Other common subtypes are vascular dementia, dementia with Lewy bodies, dementia in Parkinson's disease, and frontotemporal dementia. Around 50% of cases are mixed type (Newman 2012). There are some reversible dementia syndromes, for example due to thyroid problems or vitamin deficiencies; none of the common neurodegenerative dementias have cures, however. Dementia is often preceded by a period of milder cognitive dysfunction. Mild cognitive impairment (MCI) is a condition intermediate between normal cognitive changes that may occur with age and more serious symptoms that indicate dementia. Someone with MCI has detectable cognitive decline, more than expected from age alone, but retains functional independence. MCI is a risk factor for eventually developing dementia due to Alzheimer’s disease or another type of dementia, although it should be noted that mild cognitive impairment does not always progress to dementia (NIH 2019).

The diagnosis of dementia is based on clinical assessment, including a history from an informant who knows the person well, neuropsychological tests and functional assessment. It is important to note that some cognitive changes are part of the normal aging process (Newman 2012). These changes primarily involve relatively subtle decreases in speed of processing, attention and memory (Harada 2013). Differentiating between these normal changes and MCI can be challenging and often focuses on greater than usual declines in memory or executive functioning (Grundman 2004; Moreira 2019). By definition MCI does not impair everyday functioning while dementia impacts an individual’s ability to function independently. The later stages of dementia are associated with increased healthcare costs and eventually result in severe disability and death (Livingston 2017).

Given the different clinical and pathological variants of dementia, it is evident that no single causal factor is sufficient and necessary to explain progression from normal age‐related cognitive decline to dementia. Although the etiology for dementia due to Alzheimer's disease remains uncertain, there are genetic factors – with genes for amyloid precursor protein (APP), presenilin 1 (PSEN1), and presenilin 2 (PSEN2) contributing to early‐onset Alzheimer's disease; and apolipoprotein E (APOE) E4 allele associated with an increased risk for late‐onset Alzheimer’s disease (Elahi 2017). A review by the Lancet Commission also identified nine health and lifestyle risk factors that, in combination, contribute to 35% of the risk for dementia (Livingston 2017). It is clear that the mechanisms responsible for dementia are complex and comprize numerous interactions between biological, psychological and social factors that exert different effects at different stages of the disease process. However, there is currently no known effective treatment for most progressive forms of dementia.

Description of the intervention

Ginkgo biloba (Ginkgo) is the only surviving species of the plant division Ginkgophyta and is an endangered nonflowering tree that can be traced over 200 million years (Camfield 2016; Isah 2015; Singh 2008). Although native to China, Ginkgo is also cultivated elsewhere, and the fruits and leaves have been used since early human history as both a food and a traditional medicine (Camfield 2016; McKenna 2001). Ginkgo leaves and other aerial parts have been part of traditional East Asian medicine practices for at least the last 2000 years (Benzie 2011; Yang 1998). In addition to ongoing empirical traditional use of the plant, numerous commercially‐prepared products containing Ginkgo extractions (i.e. extracts) are currently available. Various parts of the Ginkgo tree possess bioactive compounds, including ginkgolides and bilobalides (i.e. terpene trilactones unique to the plant), as well as flavonoids (flavones, biflavones, flavonols, tannins, and associated glycosides) (Isah 2015). In recent decades, extracts of the dried green Ginkgo leaves have been investigated for treating several medical conditions, particularly those related to diseases of the peripheral and cerebral circulation (Isah 2015; McKenna 2001). Ginkgo‐derived extract EGb 761 has been used in many existing studies (Camfield 2016); the makeup of this product includes flavone glycosides (22% to 27%), terpene lactones including ginkgolides A, B and C (2.8% to 3.4%) and bilobalide (2.6% to 3.2%), and less than 5 ppm ginkgolic acids, constituents of known allergic and cytotoxic potency (Baron‐Ruppert 2001; Zuo 2017). Because Ginkgo natural products encompass a variety of preparations and doses of active compounds, a comprehensive investigation of various Ginkgo natural products is warranted to determine which, if any, may be beneficial for the treatment of cognitive impairment and dementia.

How the intervention might work

For centuries the raw material from Ginkgo has been used for its broad medicinal efficacy in various parts of Asia. Currently, the most commonly used and studied extract is EGb 761 and its major purified components, which include ginkgolides A, B, C, and J as well as bilobalide. EGb 761 and its components have been studied in pre‐clinical and clinical settings for their beneficial effects on a wide range of pathological states, from vascular problems (Tian 2017; Wu 2019) and ischemic stroke (Feng 2019; Tulsulkar 2016) to diabetes‐related health complications (Shi 2019; Zhao 2016) to neurodegenerative conditions leading to cognitive decline and dementia, including Alzheimer’s disease (Gauthier 2014; Rapp 2018; Savaskan 2018). Alzheimer’s disease is characterized by the extracellular accumulation of amyloid‐beta (Aß) and hyperphosphorylated tau protein‐based neurofibrillary tangles. Many of these pathologies are of reasonably well‐known etiologies involving oxidative stress, inflammation, and cell death. EGb 761 reportedly acts as an anti‐oxidant (Drieu 2000; Tunali‐Akbay 2007), anti‐inflammatory (Jiang 2014; Gargouri 2018), stress‐reducing (Amri 1996; Amri 2002), and neuroprotective agent (Montes 2015; Nabavi 2015; Singh 2019). The molecular mechanisms and pathways of Alzheimer's disease are not completely understood, however, with the result that the identification of therapeutic targets and the design of effective therapies remain very challenging.

The complexity of the pathophysiology combined with the challenge of using an herbal extract has, not surprisingly, led to a multitude of potential pathways through which EGb 761 might act. Studies of EGb 761 have been conducted using established cell lines, primary cell cultures, and animal models. Gargouri 2018 showed the anti‐neuro‐inflammatory effects of EGb 761 in LPS‐activated primary microglial cells by acting on the COX/PGE₂ pathway. The effects of EGb 761 on neurite growth have also been a subject of great interest as neuroplasticity is significantly reduced in aging and in neurodegenerative diseases. Lejri 2019a used SH‐SY5Y neuroblastoma cell line in a 2D and 3D surface culture and found that EGb 761 stimulated neurite outgrowth by acting on the Akt/mTOR pathway. In a recent study (Usuki 2020), the neurite outgrowth effects in PC12 cell line was attributed to the lactone moieties of bilobalide. Kuo 2019 conducted ex vivo mouse model studies comparing ginkgolide A to memantine, a low‐affinity antagonist of the N‐methyl‐D‐aspartate (NMDA) glutamate receptor and a drug clinically used to treat the symptoms of dementia due to Alzheimer's disease. They found that ginkgolide A inhibited NMDA receptors and prevented the neuronal Aß‐induced increase in c‐Jun N‐terminal kinase phosphorylation. Furthermore, behavioral changes, locomotor activity, and memory retention were assessed using the Y and Morris Water Maze and the Open Field test. The results showed that the treatment improved the memory of the wild‐type mice. The authors suggested that ginkgolide A could be a phytomedicinal‐based drug acting similarly to memantine (Kuo 2019), but possibly with fewer side effects.

Rats with homocysteinemia‐induced amnesia that were treated with EGb 761 exhibited an improvement of their cognitive impairment and a reduction in the AD‐like Tau hyperphosphorylation, attributable to the anti‐oxidative activity of Ginkgo biloba (Zeng 2018). Long‐term feeding (2 or 5 months) of Human P301S tau mutant‐transgenic mice with EGb 761 resulted in a switch of the microglia from pro‐ to anti‐inflammatory status, a decrease in the tau protein content, recovery of the CREB phosphorylation, inhibition of p38‐MAPK and GSK‐3ß, and an amelioration of the decline in cognitive function. EGb 761 and its purified compounds ginkgolide A and bilobalide but not ginkgolides B or C amplified autophagy and tau protein degradation in the neuronal lysosomes. This study demonstrated the molecular specificity of EGb 761 and its purified compounds targeting various pathways and the importance of the molecular synergy in herbal products (Qin 2018).

Using systems medicine and computer modeling approaches, Li and colleagues have recently reported the top 30 putative Alzheimer's disease target proteins, which were seen to be associated with multiple pathways, including anti‐oxidation, anti‐inflammation, anti‐apoptosis, mitochondrial protection, ion homeostasis, hormone synthesis, and neuroprotection via effects on amyloidogenesis and tau protein phosphorylation, all of which may be mechanisms responsible for beneficial effects of Ginkgo (Li 2018). Interestingly, the authors identified six putative Alzheimer's‐disease‐related proteins for which research is currently lacking (Li 2018 for details). Those newly discovered molecules represent additional potential therapeutic targets for neurodegenerative diseases and could potentially be added to the panoply of existing conventional cholinergic and anti‐glutamatergic agents as a treatment option.

Despite being the most studied phytomedicinal extract, the molecular mechanisms underlying the potential positive health outcomes of EGb 761 remain elusive. Furthermore, although standardized EGb 761 is widely considered the 'gold standard' of Ginkgo preparations there are a wide number of other Ginkgo supplements or Ginkgo‐based medications on the market that involve the whole herb or other extracts (e.g. LI 137), and these preparations likely differ somewhat from EGb 761 in their pharmacological properties and mechanisms (Bader 2018). The complex synergistic effects of the various Ginkgo compounds and the multifaceted disease processes in dementia make isolation of specific mechanisms difficult. Nevertheless, Ginkgo compounds are the topic of ongoing research as potentially effective therapeutic agents offering dementia treatment with fewer or less severe side effects than alternative medications such as cholinesterase inhibitors, and research continues into potential mechanisms.

Why it is important to do this review

As described above, dementias such as those due to Alzheimer's disease are an evolving public health issue and there are no established cures or effective treatments to slow or reverse disease progression. Despite the absence of clear evidence from early randomized controlled trials for the effect of Gingko in preventing or treating dementia it remains a popular herbal remedy for the mitigation of cognitive decline and neuroprotective effects are biologically plausible ‒ it has been shown to reduce Aβ aggregation and toxicity (Luo 2002; Wu 2006); and may also modulate vascular endothelial repair through activation of heme oxygenase‐1 (Wu 2019). The last Cochrane Review on this topic was in 2009 and did not find evidence for improvement in cognitive function compared with placebo (Birks 2009). However there have since been studies showing a modest improvement in cognitive performance, ability to perform activities of daily living (ADLs) and global cognitive function (Savaskan 2018; Tan 2015; Yang 2016; Yuan 2017). The Asian Clinical Expert Panel on Neurocognitive Disorders reached a consensus on the importance of EGb 761 use alone or as adjunctive therapy for mild cognition decline and dementia, especially in patients who show limited benefit from conventional drugs (Kandiah 2019). In the World Federation of Societies of Biological Psychiatry Guidelines, Gingko has been assessed for the level of evidence as category B and grade 3, reflecting the growing confidence in the efficacy of this herbal extract (Lejri 2019b). Given our evolving understanding of the complex disease process of dementia, a comprehensive review of the literature and a systematic assessment of all available studies for methodological rigor might allow clearer conclusions to be drawn about the potential effectiveness of a relatively cheap over‐the‐counter remedy. Given its ubiquity, ease of access and relatively benign side‐effect profile, even small improvements in cognitive function might result in improved quality of life for patients and carers and widespread cost savings if implemented at a large scale.

Objectives

To assess the efficacy and safety of Ginkgo for the treatment of people with diagnoses of cognitive impairment or dementia.

Methods

Criteria for considering studies for this review

Types of studies

We will include randomized controlled trials (RCTs). We will exclude quasi‐randomized trials, which we define as trials in which methods that are not strictly random (e.g. alternation, medical record number) are used to allocate participants to treatment groups. If the study report does not describe the method of randomization we will contact the authors to confirm that the trial was truly randomized. We will not restrict study eligibility by language or publication status.

Types of participants

We will include trials in adults with existing cognitive impairment or dementia. To be inclusive of populations with cognitive impairment, we will accept both trials including participants with diagnoses of mild cognitive impairment according to specific diagnostic criteria (e.g. the criteria for mild cognitive impairment proposed by the First Key Symposium in 2004 (Winblad 2004)) and trials with participants characterized as cognitively impaired according to more general criteria (e.g. age‐related memory impairment, subjective memory impairment). To be inclusive of populations with dementia, we will accept both trials including participants meeting formal diagnostic criteria for dementia (e.g. the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM‐5) diagnosis of Major Neurocognitive Disorder) and trials with participants identified with probable dementia based only on cognitive test scores or other clinical criteria.

We will include studies which focus on both healthy participants and participants with cognitive impairment or dementia only if they separately report the outcome data for participants with cognitive impairment or dementia.

Types of interventions

We will include studies that evaluate Ginkgo as an intervention for cognitive impairment or dementia, including any preparation or extract, at any dose and by any route of administration (e.g. oral or intravenous).

The comparisons of interest will be:

• Ginkgo versus placebo;

• Ginkgo versus no treatment (e.g. wait list, or no other specified active intervention);

• Ginkgo versus other active intervention;

• Ginkgo + other active intervention versus placebo + the same active intervention; and

• Ginkgo + other active intervention versus other active intervention alone.

For all comparisons involving active interventions, we will group similar active interventions together, and use subgroup analyses to evaluate similar but different interventions within that category (e.g. different types of medications or natural products (or both), with additional consideration of preparation method).

We will allow concomitant treatments, as long as they are given to both intervention groups.

Types of outcome measures

Because dementia is a chronic and progressive condition, we will only include studies that follow participants for a minimum of three months. We will extract outcomes assessed at short‐term (3 to 4 months, and closest to 3 months), intermediate‐term (between 4 and 11 months, and closest to 6 months), and long‐term (at least 12 months and closest to 1 year) time points. We will consider the intermediate‐term outcomes (closest to 6 months) to be the outcomes of primary interest.

Primary outcomes

• Global clinical status (e.g. as measured by tools such as the Clinician's Interview‐Based Impression of Change (CIBIC), the Clinical Dementia Rating (CDR), or the Global Deterioration Scale (GDS))

• Global cognitive function (e.g. as measured by tools such as the Mini Mental State Examination (MMSE) or the cognitive subscale of the Alzheimer Disease Assessment Scale (ADAS‐Cog), among others)

• Ability to perform activities of daily living (ADL) (e.g. as measured by tools such as the Bristol Activities of Daily Living Scale (Bristol ADL) or the Disability Assessment for Dementia (DAD), among others)

• Adverse events (both serious and not serious, and including ginkgo‐drug interactions when reported)

Secondary outcomes

• Memory‐specific cognitive measures (e.g. Logical Memory Test)

• Executive‐function‐specific cognitive measures (e.g. Verbal Fluency Test)

• Quality of life (e.g. as measured by disease‐specific tools such as the Cornell‐Brown Scale for Quality of Life in Dementia (CBS), the Dementia Quality of Life Instrument (DQoL), or the Psychological Well‐Being in Cognitively Impaired Persons (PWB‐CIP) tool, among others)

Search methods for identification of studies

Electronic searches

We will search ALOIS (www.medicine.ox.ac.uk/alois), the Cochrane Dementia and Cognitive Improvement Group (CDCIG) specialized register. ALOIS is maintained by the Information Specialists for the Cochrane Dementia and Cognitive Improvement Group, and contains studies that fall within the areas of dementia prevention, dementia treatment and management, and cognitive enhancement in healthy older people. The studies are identified through a monthly search of the following.

• Databases: Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library; MEDLINE; Embase; CINAHL; PsycINFO; LILACS; and Web of Science;

• Trials registers: ClincialTrials.gov and the World Health Organization (WHO) ICTRP Portal (www.who.int/ictrp/en) which covers ClinicalTrials.gov; ISRCTN; Australian and New Zealand Clinical Trial Registry; Chinese Clinical Trial Register; India Clinical Trials Registry; German Clinical Trials Register; Iranian Registry of Clinical trials; Sri Lanka Clinical trials Registry; the Netherlands National Trial Register.

Additional information about ALOIS can be found on the ALOIS website: alois.medsci.ox.ac.uk/about-alois.

We will run separate searches in CENTRAL, MEDLINE, Embase, CINAHL, and PsycINFO to ensure comprehensive and up‐to‐date searches for this review. We will also search IndMED (medind.nic.in/imvw; inception to present); China Network Knowledge Infrastructure (CNKI) (www.cnki.net) (1915 to present); Chinese Scientific Journals Database (VIP) (www.cqvip.com) (1989 to present); Wan Fang data (www.wanfangdata.com.cn/index.html) (1985 to present); SinoMed (www.sinomed.ac.cn) (1978 to present); Korean Medicine Information System (Oriental Medicine Advanced Searching Integrated System (OASIS)) (1963 to present); KoreaMed (Korean Association of Medical Journal Editors) (www.koreamed.org) (1958 to present); Korean Medical Database (KMbase) (kmbase.medric.or.kr) (1958 to present); Research Information Service System (RISS) (www.riss.kr) (1958 to present); The Town Society of Science Technology (TSSN) (society.kisti.re.kr) (1963 to present); Korean Studies Information Service System (KISS) (kisseng.kstudy.com) (1954 to present); and Korean Traditional Knowledge Portal (KTKP) (www.koreantk.com/ktkp2014) (1963 to present). These searches will ensure our inclusion of relevant South Asian and East Asian trials. The search strategy we will use to retrieve reports of trials from MEDLINE Ovid can be seen in Appendix 1.

Searching other resources

We will screen references of included trials and relevant systematic reviews and practice guidelines identified during the screening process.

Data collection and analysis

Selection of studies

Two authors (LSW, TMF) will independently screen the titles and abstracts of all retrieved citations, and assess the full text of all potentially eligible references. We will record reasons for exclusion at the full‐text stage in the 'Characteristics of excluded studies' table. We will resolve disagreements by consensus, or by involving a third author (NKP or HA).

Data extraction and management

One author (LSW) will extract the characteristics of each study using a pre‐piloted data extraction form to collect the following information.

• Methods: study design, study setting, recruitment method, inclusion and exclusion criteria.

• Participants: number of participants randomized and analyzed, age, gender, diagnostic criteria, baseline cognitive status.

• Interventions: detailed description of active interventions, including product preparation, standardization, dose, and total length of treatment period.

• Comparators: description of control conditions and presence of co‐interventions.

• Outcomes: listing of all outcomes reported in the study.

• Other: trial sponsorship, conflicts of interest, details of assessment of adverse effects (active or passive surveillance, whether and how authors defined serious adverse events).

A second author (TMF or EL) will check the extraction of the characteristics of studies.

Two authors (LSW, EL) will independently extract the outcome data for our prespecified primary and secondary outcomes from each primary study, using a standard pre‐piloted extraction form. They will resolve any disagreements by consensus, or by consulting a third author (NKP).

When both endpoint and change data are available in the study report, we will use endpoint data in our primary analysis. In cases where study participants are lost to follow‐up, and both available case analyses and intention‐to‐treat analyses using imputation are presented, we will use the imputed data for our primary analysis.

In cases where both unadjusted and adjusted data are presented, we will use the adjusted data for our primary analysis. We will use sensitivity analyses to observe whether our results are robust to these decisions (Sensitivity analysis).

Assessment of risk of bias in included studies

Two authors (LSW, EL) will independently assess the risk of bias for each included study, using the criteria outlined in Chapter 8 of Version 5.2.0 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). We will extract information on risk of bias, including relevant quotations from the study and an explanation of the risk of bias decision, into a standard pre‐piloted form, and the authors will check understanding and agreement for at least two studies before proceeding to evaluate the remaining studies. The authors will resolve disagreements on risk of bias by consensus, or by involving a third author (NKP).

We will assess the risk of bias according to the following domains.

• Selection bias (random sequence generation, allocation concealment)

• Performance bias (blinding of participants, blinding of personnel or care providers)

• Detection bias (blinding of outcome assessors)

• Attrition bias (incomplete outcome data)

• Reporting bias (selective reporting of outcomes)

• Other bias (other sources of bias related to a particular trial design or specific circumstances, such as conflicts of interest

We will classify the risk of bias as low, high, or unclear in each domain (Higgins 2017). We will consider trials that are at low risk of bias in each domain to be at overall low risk of bias, and carry out sensitivity analyses in which we omit trials at overall higher risk of bias from analyses (Sensitivity analysis).

Measures of treatment effect

We will calculate risk ratio (RR) and 95% confidence intervals (CI) for dichotomous data and the mean difference (MD) or the standardized mean difference (SMD) and 95% CI for continuous data.

Unit of analysis issues

If we find relevant cluster‐randomized trials, we will follow the guidance in chapter 23.1 of the Cochrane Handbook for Systematic Reviews of Interventions to assess their suitability, and include them in the analysis if appropriate (Higgins 2019). If we find relevant cross‐over trials, we will follow the guidance in chapter 23.2 of the Cochrane Handbook for Systematic Reviews of Interventions and include the data from the first period of the trial in the analysis if appropriate (Higgins 2019).

Dealing with missing data

If primary outcome data or important study characteristics are missing from the study report, we will e‐mail the contact author or the primary investigator to obtain the missing information. We will note this contact and the results in the 'Characteristics of studies' table for the study.

Assessment of heterogeneity

We will assess clinical heterogeneity between studies qualitatively. For studies that we consider sufficiently clinically homogenous with regard to participants, interventions, comparators, and outcomes to combine in a meta‐analysis, we will assess the presence of statistical heterogeneity using the Chi² test. We will consider P less than 0.10 as evidence of statistically significant heterogeneity.

The I² statistic describes the percentage of variability in the overall effect that is due to heterogeneity rather than to chance (Deeks 2019). The suggested thresholds for interpretation of I² percentages are as follows.

• 0% to 40%: might not be important;

• 30% to 60%: may represent moderate heterogeneity;

• 50% to 90%: may represent substantial heterogeneity;

• 75% to 100%: considerable heterogeneity.

We will assess the importance of the value of the I² statistic in the context of the range and direction of effects as observed from the forest plots, and the strength of the evidence for heterogeneity based upon the Chi² test (Deeks 2019). We will also use subgroup analyses to explore potential sources of clinical heterogeneity (see Subgroup analysis and investigation of heterogeneity).

Assessment of reporting biases

We will use the Egger's test and a funnel plot to assess the potential for small‐study bias in meta‐analyses in which at least 10 studies are included (Egger 1997; Sterne 2011). We will also evaluate the possibility of selective outcome reporting for each included study as part of the assessment of risk of bias (Assessment of risk of bias in included studies).

Data synthesis

We will use meta‐analysis to combine the outcome data across trials when there are available data and sufficient clinical similarity of trials in population, interventions, comparators, outcomes, and timing of assessment.

We will carry out two main analyses, one including participants with dementia and one including participants with less severe cognitive impairments (e.g. MCI, subjective cognitive decline). We will then carry out subgroup analyses by condition, if the data permit (see Subgroup analysis and investigation of heterogeneity). We will use a random‐effects model for all analyses, because we expect some between‐study variation and we wish to generalize the study results beyond the specific studies included in the analyses. When it is not possible to carry out a meta‐analysis, we will provide a qualitative description of the results from clinically similar trials.

We will summarize dichotomous data using the RR and 95% CI. We will summarize continuous data using the MD and 95% CI for measures using the same scale, or the SMD and 95% CI when different scales were used to measure the same outcome (e.g. quality of life).

We will assess the overall certainty of the evidence for each outcome, whether or not there is available quantitative data. We will use the GRADE criteria outlined in Chapter 14 of the Cochrane Handbook for Systematic Reviews of Interventions to determine whether the evidence for each estimate is high, moderate, low or very low certainty (Schünemann 2019). The GRADE approach begins by considering evidence based upon RCTs to be high certainty. However, the certainty of evidence from RCTs may be downgraded one or more levels for serious concerns in any of the following areas.

• Study limitations (risk of bias)

• Inconsistency (heterogeneity)

• Indirectness of evidence

• Imprecision

• Publication bias

One review author (LSW) will apply the GRADE approach to each outcome, and one review author (EL) will check the assessments. If there is disagreement we will reach consensus through discussion or consultation with a third author (NKP or HA).

'Summary of findings' table

We will create a 'Summary of findings' (SoF) table for our main comparison, Ginkgo versus placebo. We will present the estimates and certainty of the evidence for the primary outcomes of global functioning, cognitive function, and ADLs at the intermediate‐term time point (closest to six months). We will also present the estimate and certainty of evidence for adverse effects in the SoF table.

Subgroup analysis and investigation of heterogeneity

If there are available data, we will carry out subgroup analyses based upon factors that may cause the effect of the intervention to vary. This will include the following factors.

• Type of cognitive impairment (e.g. mild cognitive impairment versus subjective cognitive decline, mild dementia versus severe dementia, dementia due to Alzheimer's disease versus vascular dementia), as different degrees or types of impairment may respond differently to Ginkgo.

• Type of Ginkgo intervention (e.g. standardized extract EGb 761 versus other standardized extracts versus non‐standardized Ginkgo extracts versus traditional preparation types (i.e. decoction or tea, tincture)), as different types and preparations of Ginkgo natural product may produce different effects.

• Different dosages of Ginkgo, as some recent research suggests that doses of standardized extract over 200 mg/day are associated with greater efficacy (Yuan 2017).

For each subgroup analysis in which each subgroup has at least two studies, we will use the I² statistic to assess the percentage of the variability in effect estimates that is due to genuine subgroup differences rather than to chance (Deeks 2019). For subgroup comparisons where the I² statistic is not appropriate, we will provide a narrative description of any differences. In both cases, we will be cautious about drawing conclusions about causality based on subgroup differences, as between‐studies comparisons are observational evidence.

Sensitivity analysis

If there are available data, we will carry out sensitivity analyses to assess the robustness of results to the following choices.

• Including all trials versus only trials at low risk of bias (as defined in Assessment of risk of bias in included studies).

• Using endpoint outcome data versus change outcome data.

• Using imputed outcome data versus available case outcome data.

• Using adjusted outcome data versus unadjusted outcome data.

History

Protocol first published: Issue 7, 2020

Appendices

Appendix 1. Search strategy for MEDLINE Ovid

1. exp Dementia/

2. exp *Cognition Disorders/

3. Neurocognitive Disorders/

4. alzheimer*.ti,ab.

5. dement*.ti,ab.

6. ((lewy* adj2 bod*) or (DLB or LBD or FTLD or FTD or frontotemporal or PDD)).ti,ab.

7. "major neurocognitive disorder".ti,ab.

8. "mild neurocognitive disorder*".ti,ab.

9. ((cognit* or memory or cerebr* or mental*) adj3 (declin* or impair* or los* or deteriorat* or degenerat* or complain* or disturb* or disorder* or insufficient* or chronic)).ti,ab.

10. MCI.ti,ab.

11. (aMCI or MCIa or nMCI or mMCI).ti,ab.

12. (ACMI or ARCD or SMC or CIND or BSF or AAMI or AACD or MNCD).ti,ab.

13. "Benign senescent forgetfulness".ti,ab.

14. ((CDR adj2 "0.5") or ("clinical dementia rating" adj3 "0.5")).ab.

15. ("organic brain disease" or "organic brain syndrome").ti,ab.

16. "preclinical AD".ti,ab.

17. "pre‐clinical AD".ti,ab.

18. "preclinical alzheimer*".ti,ab.

19. "pre‐clinical alzheimer*".ti,ab.

20. or/1‐19

21. exp Ginkgo biloba/

22. ginkgo.ti,ab.

23. "maidenhair extract*".ti,ab.

24. "maidenhair tree*".ti,ab.

25. GbE.ti,ab.

26. EGb.ti,ab.

27. EGb761.ti,ab.

28. Tanakan.ti,ab.

29. Tebonin.ti,ab.

30. Rokan.ti,ab.

31. Ginkoba.ti,ab.

32. "Pterophyllus salisburiensis leaf".ti,ab.

33. "Salisburia adiantifolia leaf".ti,ab.

34. "Salisburia biloba leaf".ti,ab.

35. "Salisburia macrophylla leaf".ti,ab.

36. Yinxingye.ti,ab.

37. gingko.ti,ab.

38. or/21‐37

39. randomized controlled trial.pt.

40. controlled clinical trial.pt.

41. randomized.ab.

42. placebo.ab.

43. drug therapy.fs.

44. randomly.ab.

45. trial.ab.

46. groups.ab.

47. or/39‐46

48. exp animals/ not humans.sh.

49. 47 not 48

50. 49 and 38 and 20

Contributions of authors

LSW, TMF, EL, NKP and HA drafted the protocol.

Sources of support

Internal sources

• No sources of support supplied

External sources

• National Institutes of Health (NIH), USA

  Support was received from the National Center for Complementary and Integrative Health of the NIH under award number R24AT001293. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

• NIHR, UK

  This protocol was supported by the National Institute for Health Research (NIHR), via Cochrane Infrastructure funding to the Cochrane Dementia and Cognitive Improvement group. The views and opinions expressed herein are those of the authors and do not necessarily reflect those of the Systematic Reviews Programme, NIHR, National Health Service or the Department of Health

Declarations of interest

L. Susan Wieland: none known

Termeh M. Feinberg: none known

Emilie Ludeman: none known

Nikhil K. Prasad: none known

Hakima Amri: none known

New