Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Protocol - Intervention

Citicoline for treating people with acute ischemic stroke

This is not the most recent version

Collapse all Expand all

Abstract

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

To assess the clinical benefits and harms of citicoline compared with placebo or any other control for treating people with acute ischemic stroke.

Background

Description of the condition

Stroke is one of the leading causes of long‐lasting disability and mortality (Martynov 2015; Yan 2017). The global burden of stroke has increased in the past two decades (Hankey 2017). In 2015, ischemic heart disease and stroke were the two leading causes of premature mortality worldwide, and the leading causes of years of life lost (YLLs) in 119 countries and territories (Wang 2016). Finding a safe, inexpensive therapy to prevent and treat stroke is becoming an area of interest in international public health.

Stroke is defined as an acute (sudden) episode characterized by clinical signs of focal or global disturbance of cerebral (brain) function caused by infarction (blockage in blood flow) or hemorrhage (bleeding) in the relevant part of the brain, retina, or spinal cord, lasting more than 24 hours, or of any duration if imaging (computed tomography [CT]; or magnetic resonance imaging [MRI]) or autopsy show focal infarction or hemorrhage relevant to the symptoms. By comparison, a transient ischemic attack (TIA) is defined as a focal dysfunction of less than 24 hours' duration and with no imaging evidence of infarction (Hankey 2017; Portegies 2016). Stroke is classified into ischemic stroke (obstruction of cerebral blood), intracerebral hemorrhage (a focal collection of blood within the brain parenchyma or ventricular system that is not caused by trauma), and subarachnoid hemorrhage (bleeding into the subarachnoid space). Ischemic stroke is the most prevalent type and accounts for 75% to 85% of all strokes (Cassella 2017; Martynov 2015). This Cochrane Review will focus on acute ischemic stroke.

Different strategies have been proposed for the recovery from, and treatment of, ischemic stroke. During the first few hours of onset of an ischemic stroke, blood flow can be restored through mechanical thrombus extraction (thrombectomy) or by thrombolysis ('clot‐busting') with the administration of recombinant tissue plasminogen activators (rt‐PAs) that break down blood clots. However, only 5% to 20% of people with acute ischemic stroke are treated with thrombolysis or thrombus extraction, or both (Martynov 2015; Wardlaw 2014).

Three parenchymal vascular states are seen in people with acute ischemic stroke, namely: the ischemic core, the zone of penumbra adjacent to it, and the region of benign oligemia adjacent to that (Manning 2014). The ischemic penumbra region is defined as the "zone of nonfunctioning but still viable tissue that may recover its function if blood flow can be restored for example, by therapeutic intervention" (Astrup 1981; Astrup 1982). The penumbra region represents that portion of the ischemic territory that is still potentially salvageable, and so it is the target of all acute therapies (Hillis 2015; Manning 2014; Martynov 2015). Therefore, the zone of penumbra must be protected; this protection is called neuroprotection. In this scenario, neuroprotection has been defined as "administering therapy as rapidly as possible following the onset of symptoms in an effort to minimize cerebral infarction while the ischemic brain is awaiting reperfusion" (Patel 2017). There are pharmacological and non‐pharmacological forms of neuroprotection (Patel 2017; Wang 2017), and this Cochrane Review will focus on one type of pharmacological neuroprotector: citicoline.

Several studies have suggested that citicoline is effective in treatment of central nervous system disorders (CNS), including acute and chronic cerebral ischemia, intracerebral hemorrhage, global cerebral hypoxia, and neurodegenerative diseases (Martynov 2015). These studies claim that citicoline treatment reduces size of infarct (region of dead tissue), decreases free fatty acid concentration, decreases neurologic deficits, restores animal learning performance, reduces glutamate‐mediated injury, preserves phosphatidylcholine levels, and improves neuronal survival (Clark 2009).

Description of the intervention

Citicoline is the generic name for cytidine 5'‐diphosphocholine (CDP‐choline). Citicoline is the combination of two molecules: cytidine and choline. These molecules cross the blood‐brain barrier separately; after reaching the brain cells they combine to generate CDP‐choline (Martynov 2015). Thus, it is a natural endogenous compound and a precursor for the synthesis of phosphatidylcholine, one of the components of cell membranes. During cerebral ischemia, phosphatidylcholine is broken down into free fatty acids and free radicals that increase the likelihood of ischemic injury (Clark 2009).

Citicoline is synthesized in oral and enteral formulations for clinical practice. Citicoline is a water‐soluble compound, and pharmacokinetic studies on healthy adults have shown an optimal absorption of both oral and intravenous doses. Once absorbed, citicoline is converted to choline and cytidine, which circulate in the body, enter systemic circulation and cross the blood–brain barrier for resynthesis into citicoline in the brain (Clark 2009). Citicoline has been administered intravenously at doses ranging between 500 mg and 2000 mg respectively in randomized controlled trials (Avarez‐Sabín 2013; Clark 1997).

How the intervention might work

The cytidine‐5'‐diphosphocholine (CDP‐choline or citicoline) pathway begins with the uptake of exogenous choline into the cell, followed by different enzymatic reactions that lead to the synthesis of phosphatidylcholine. The hypothesized mechanism of action assumes that citicoline undergoes hydrolysis and dephosphorylation to yield cytidine and choline products. These breakdown products then enter the brain separately and act as substrates for the resynthesis of CDP‐choline. This mechanism is believed to slow down phospholipid breakdown and accelerate phospholipid resynthesis, which are necessary for membrane repair (Grieb 2014). Citicoline is a free radical scavenger that reduces the availability of free radicals after acute ischemic episodes (Rajah 2017), and so may slow the damage. It may facilitate recovery by activating neurogenesis, synaptogenesis, and angiogenesis (creation of brain cells, gaps between them, and blood supply), and enhances neurotransmitter metabolism. In an animal model, administration of citicoline showed an up‐regulation of synaptophysin in the penumbra region that could indicate an increase of synaptic (brain) activity (Martynov 2015).

The administration of citicoline increases brain choline, and accelerates the synthesis of phosphatidylcholine, decreases levels of free fatty acids, and reduces the generation of free radicals. Overall, the administration of citicoline may protect cell membranes by reducing phosphatidylcholine breakdown (Martynov 2015). Recent research suggests that citicoline may enhance endogenous brain plasticity and repair even when it is administered several hours after the ischemic event (Clark 2009; Overgaard 2014). An animal study reported significantly improved motor and functional recovery in the citicoline‐treated group at the end of a 28‐day intervention period, when citicoline was given 24 hours after middle cerebral artery stroke, and taken for 28 days (Clark 2009).

Why it is important to do this review

This Cochrane Review is important because controversy exists regarding the clinical effectiveness of citicoline for treating people with acute stroke. Some randomized placebo, or head‐to‐head, controlled clinical trials and meta‐analyses suggest limited or no clinical benefits (Clark 1999; Clark 2001; Dávalos 2012; Mitta 2012; Secades 2016; Shi 2016), while other randomized, placebo‐controlled clinical trials report beneficial effects (Alviarez 2007; Clark 1997; Corso 1982; Tazaki 1988). Furthermore, Dávalos 2002, Secades 2016, and Shi 2016 have several methodological limitations such as lack of reporting summary of finding tables, use of odds ratio (OR) instead of risk ratio (RR), and use of cumulative meta‐analysis (Sterne 2001), instead of Trial Sequential Analysis (TSA) (Wetterslev 2017). In addition, drug companies have funded many trials (Alviarez 2007; Clark 1997; Clark 1999; Clark 2001; Dávalos 2012), and it is necessary to assess the effectiveness of the drug independently (Lundh 2017). Although, on 17 April 2009 citicoline was banned in the USA and Canada (Chen 2016), it is still used in countries such as Argentina, Austria, Chile, Indonesia, Mexico, Portugal, Thailand, and Venezuela (Chen 2016). Therefore, conducting a systematic review according to Cochrane methodology is fundamental to identify the clinical benefits and harms of citicoline compared with placebo or any other control for treating people with acute ischemic stroke.

Objectives

To assess the clinical benefits and harms of citicoline compared with placebo or any other control for treating people with acute ischemic stroke.

Methods

Criteria for considering studies for this review

Types of studies

We will include randomized controlled trials (RCTs) irrespective of publication status. We will not apply any limitation by language, country, or duration of follow‐up. We will only include parallel‐design trials.

Types of participants

People (children or adults) with acute ischemic stroke, irrespective of etiology. We will use clinical diagnosis with imaging as an eligibility criterion.

Types of interventions

We will include trials that compare citicoline with placebo, usual care, or other interventions.

Citicoline administered at any dose, by any route, and for any duration of treatment, versus no intervention, placebo, or other interventions. Since acute ischemic stroke requires a variety of medical treatments (that is, primary interventions), we will consider citicoline as a supplementary intervention. Thus, for the purpose of this review, eligible randomized controlled trials will be those that compare the same primary interventions with and without citicoline supplementation.

Types of outcome measures

Primary outcomes

Secondary outcomes

  • Adverse events during the first 30 days, assessed according to recommendations from Lineberry 2016.

  • Functional recovery: assessed with the Barthel Index (at 90 days) (Harrison 2013; Mahoney 1965). See Appendix 2 for details.

  • Neurological function assessed with the National Institutes of Health Stroke Scale (NIHSS) (Harrison 2013). We will assess it at the first 24 hours (acute phase), at 72 hours, and at discharge. See Appendix 3 for details.

  • Quality of life (at 90 days): assessed using the 36‐item Short Form Survey (SF‐36), EuroQol, the Stroke Specific Quality of Life scale (SS‐QoL) (Harrison 2013).

Search methods for identification of studies

We will search for trials in all languages and arrange for translation of relevant articles where necessary.

Electronic searches

We will search in the Cochrane Stroke Group Trials Register and the following electronic databases:

  • Cochrane Central Register of Controlled Trials (CENTRAL; latest issue) in the Cochrane Library;

  • MEDLINE Ovid (from 1948 to present) (Appendix 4);

  • Embase Ovid (from 1974 to present);

  • LILACS (Latin American and Caribbean Health Science Information database; 1982 to present).

With assistance from the Cochrane Stroke Group's Information Specialist, we will model the subject strategies for the database searches on the search strategy designed for MEDLINE (Appendix 4). We will combine all search strategies deployed with subject strategy adaptations of the highly sensitive search strategy designed by Cochrane for identifying randomized controlled trials and controlled clinical trials (as described in the Cochrane Handbook for Systematic Reviews of Interventions Chapter 6: Lefebvre 2011).

Searching other resources

We will also search the following ongoing trials registries:

In order to identify unpublished information submitted for the marketing approval of citicoline, we will also search the following sites:

We will screen the reference lists of relevant studies and use Cited Reference Search within Web of Science to identify further studies for potential inclusion in the review, and we will contact trialists and companies for further information.

Data collection and analysis

We will conduct data collection and analysis of data according to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a; Higgins 2011b).

Selection of studies

Two review authors (AMC, CV) will independently screen titles and abstracts of the references obtained as a result of our searching activities, and will exclude obviously irrelevant reports. We will retrieve the full‐text articles for the remaining references and, independently, two or more review authors (AMC, CV, IS) will screen the full‐text articles and identify studies for inclusion. They will also identify and record reasons for exclusion of the ineligible studies. We will resolve any disagreements through discussion or, if required, we will consult a third review author (JMF). We will collate multiple reports of the same study so that each study, not each reference, is the unit of interest in the review. We will record the selection process and complete a PRISMA flow diagram.

Data extraction and management

Independently, two review authors (AMC, CV) will extract data from included studies. We will develop an Excel spreadsheet based on the 'Data extraction template for included studies' from the Consumers and Communication Group resources for authors. We plan to describe the details of the intervention following recommendations from Hoffmann 2014 and Hoffmann 2017.

Assessment of risk of bias in included studies

Independently, two review authors (AMC, CV) will assess risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011c). We will resolve any disagreements by discussion or by involving another review author (JMF, XB). We will assess the risk of bias according to the following domains.

  • Random sequence generation

  • Allocation concealment

  • Blinding of participants and personnel

  • Blinding of outcome assessment

  • Incomplete outcome data

  • Selective outcome reporting

  • Other bias

We will grade the risk of bias for each domain as high, low, or unclear and provide information from the study report together with a justification for our judgment in the 'Risk of bias' tables. We will include company funding under 'Other bias'.

See Appendix 5 for details of domains.

Measures of treatment effect

For binary outcomes, such as all‐cause mortality and adverse events, we plan to calculate the relative risk (RR) with 95% confidence interval (CI).

For continuous outcomes, such as functional outcome, degree of disability or dependence in daily activities, and neurological behavioral and cognitive function, we plan to calculate the mean difference (MD) with 95% CI. If ordinal data are reported, we will use a proportional odds model as a measure of treatment effect with Stata statistical software (STATA) (Bath 2012: Deeks 2017; Scott 1997). If different scales are used for measuring the same outcome, for example quality of life, we plan to use the standardized mean difference (SMD) with 95% CI. We will also estimate ratio of means (RoM) with 95% CI from mean difference (Friedrich 2011). Due to practitioners understanding and preferring dichotomous presentations of continuous outcomes, which they perceive to be the most useful (Johnston 2016), we will estimate odds ratios (OR) with 95% CI and the number needed to treat for an additional beneficial outcome (NNTB) from SMD with Furukawa's method (Furukawa 1999; Furukawa 2011).

As recommended in the Cochrane Handbook for Systematic Reviews of Interventions, if necessary we will multiply the mean values from one set of studies by ‐1 to ensure that all the scales point in the same direction (Deeks 2017).

If statistical information is missing (such as standard deviations), we will try to extract them from other relevant information in the paper, such as P values and CIs.

We will calculate the NNTB if the RR was significant (P value < 0.05). NNTB is a measure of assessment of clinical useful of the consequences of treatment (Laupacis 1988). We will estimate NNTB with GraphPad software and with the Cochrane Stroke Group NNT calculator. If ordinal data are reported, we will estimate NNTB according to Bath 2011.

Unit of analysis issues

The unit of analysis will be participants. We will exclude cluster randomized controlled trials.

As recommended in the Cochrane Handbook for Systematic Reviews of Interventions, we will conduct the following plan to assess the outcomes with multiple observations.

  1. For primary outcomes (all‐cause mortality and degree of disability or dependence in daily activities according to the modified Rankin scale) and secondary outcomes (adverse events, functional recovery and quality of life), we will select a single time point and will analyze only data at this time for trials in which it will be presented.

  2. Neurological function assessed with the National Institutes of Health Stroke Scale (secondary outcome), we plan to define different periods of follow‐up, and perform separate analyses. These will reflect short‐term, medium‐term, and long‐term follow‐up (Deeks 2017).

Dealing with missing data

We will assess the percentage of dropouts for each included trial, and for each intervention group, and will evaluate whether an intention‐to‐treat (ITT) analysis had been performed or could have been performed from the available published information. We will contact study authors to resolve any questions arising from this issue.

In order to undertake an ITT analysis, we will seek data from the trial authors about the number of participants in treatment groups, irrespective of their compliance and whether or not they were later thought to be ineligible, otherwise excluded from treatment, or lost to follow‐up. If this information is not forthcoming, we will perform a 'per protocol' analysis of those who completed the study, being aware that it may be biased.

We will include participants with incomplete or missing data in sensitivity analyses by imputing them according to the following scenarios (Hollis 1999).

  • Extreme case analysis favoring the experimental intervention ('best‐worse' case scenario): none of the drop‐outs/participants lost from the experimental arm, but all of the drop‐outs/participants lost from the control arm experienced the outcome, including all randomized participants in the denominator.

  • Extreme case analysis favoring the control ('worst‐best' case scenario): all drop‐outs/participants lost from the experimental arm, but none from the control arm experienced the outcome, including all randomized participants in the denominator.

  • Gamble‐Hollis analysis, which takes account of the uncertainty and generates uncertainty intervals for a trial incorporating both sampling error and the potential impact of missing data (Gamble 2005). This method increases the uncertainty of the trials using the results from the best‐case and worst‐case analyses (Chaimani 2014).

Assessment of heterogeneity

We will quantify statistical heterogeneity using the I² statistic, which describes the percentage of total variation across trials that is due to heterogeneity rather than sampling error (Higgins 2003). We will consider statistical heterogeneity to be present if I² is greater than 60% (Deeks 2017). We will quantify 95% CI or an uncertainty interval of I² with Stata statistical software (STATA) (Kontopantelis 2010). If there were simultaneously statistical heterogeneity and three or more RCTs, we will determine the 95% prediction interval (PI), which takes into account the whole distribution of the effects (Riley 2011). Prediction intervals in meta‐analysis show the expected range of true effects in similar studies (IntHout 2016). We will estimate the 95% prediction interval with Stata statistical software (STATA). If there are 10 or more randomized controlled trials and I² is greater than 60%, we will conduct a meta‐regression with Stata statistical software (STATA).

Assessment of reporting biases

If there are 10 or more randomized controlled trials, we will use the contour‐enhanced funnel plot to differentiate asymmetry that is due to publication bias from that due to other factors (Peters 2008). We will assess likelihood of publication bias with Harbord and Peters tests (Sterne 2011). We will use Stata statistical software to produce conventional and contour funnel plots (STATA).

Data synthesis

We will perform meta‐analyses with 95% CI using either a fixed‐effect or random‐effects model. In case of statistical heterogeneity (I² > 60%), we will report data using the random‐effects model and prediction interval (Deeks 2017; IntHout 2016; Riley 2011). We will conduct meta‐analysis with Review Manager 5.3 (RevMan 2014).

Trial sequential analysis (TSA)

We will apply trial sequential analysis (TSA), as cumulative meta‐analyses are at risk of producing random errors due to sparse data and repetitive testing of the accumulating data (Brok 2008; Brok 2009; Imberger 2015; Imberger 2016; Thorlund 2010; Wetterslev 2008; Wetterslev 2009; Wetterslev 2017). To minimize random errors, we will calculate the required information size (i.e. the number of participants needed in a meta‐analysis to detect or reject a certain plausible intervention effect) (Wetterslev 2008; Wetterslev 2017). The required information size calculation should also account for the heterogeneity or diversity present in the meta‐analysis (Wetterslev 2008; Wetterslev 2009; Wetterslev 2017). We will use the event proportion in the control group; assumption of a plausible RR reduction of 20% or the RR reduction observed in the included trials with low risk of bias; a risk of type I error of 5%; a risk of type II error of 10%; and the empirical diversity of the meta‐analysis for estimating the diversity‐adjusted required information size (Wetterslev 2009; Wetterslev 2017). We will add the trials according to the year of publication, and, if more than one trial has been published in a year, we will add trials alphabetically according to the last name of the first author. On the basis of the required information size, we will construct trial sequential monitoring boundaries (Lan 1983; Thorlund 2011; Wetterslev 2008). These boundaries determine the statistical inference one may draw regarding the cumulative meta‐analysis that has not reached the required information size; if the trial sequential monitoring boundary for benefit or harm is crossed before the required information size is reached, and further trials may turn out to be superfluous. On the other hand, if the boundary is not surpassed, it is probably necessary to continue conducting trials in order to detect or reject a certain intervention effect. This can be determined by assessing whether the cumulative Z‐curve crosses the trial sequential boundaries for futility. If futility boundaries are crossed, then further trials may be unnecessary (CTU 2011). We will conduct TSA using software from the Copenhagen Trial Unit (CTU 2011; Thorlund 2011).

GRADE and 'Summary of findings' table

We will create a 'Summary of findings' table using the following outcomes: all‐cause mortality; degree of disability or dependence in daily activities on the modified Rankin scale; adverse events; functional recovery (Barthel Index); neurological function (National Institutes of Health Stroke Scale: NIHSS); and quality of life (Table 1). We will use the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness and publication bias) to assess the quality of a body of evidence as it relates to the studies that contribute data to the meta‐analyses for the prespecified outcomes (Atkins 2004). We will use methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011), and GRADEproGDT software (GRADEpro GDT 2015). We will justify all decisions to downgrade the quality of the evidence using footnotes, and we will make comments to aid the reader's understanding of the review where necessary.

Open in table viewer
Table 1. Template for the 'Summary of findings' table

Citicoline for treating people with acute ischemic stroke

Patient or population: people with acute ischemic stroke
Settings: inpatients
Intervention: citicoline

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Control

Citicoline

All‐cause mortality (at any time of trial)

Study population

Degree of disability or dependence in daily activities
Modified Rankin scale (from 0 to 6): lower score indicates better outcome

Study population

Adverse events assessed

Study population

Functional recovery
Barthel Index

Score 0 to 100 (higher score indicates greater independence)

Study population

Neurological function
National Institutes of Health Stroke Scale

(0 (no impairment) to a maximum of 42)

Study population

Quality of life (at the time of follow‐up)
SF‐36, Euro‐Qol, SS‐Qo

Study population

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: Confidence interval; RR: Risk ratio;

GRADE Working Group grades of evidence
High quality: further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: we are very uncertain about the estimate.

Subgroup analysis and investigation of heterogeneity

If we identify enough trials (five or more), we will conduct the following subgroup analysis:

  • Participants with diabetes mellitus versus participants without diabetes mellitus.

  • Participants with high blood pressure versus participants without high blood pressure.

  • Trials supported by pharmaceutical companies versus trials without support by pharmaceutical companies.

  • Trials with low risk of bias versus trials with high risk of bias.

  • Trials with small sample size (≤ 200 participants) versus trials with large sample size (> 200 participants).

We will only conduct subgroup analysis for primary outcomes.

Sensitivity analysis

We will perform the following sensitivity analysis in order to explore the influence of particular factors on the intervention effect size: 'best‐worst case' scenario versus 'worst‐best case' scenario and Gamble‐Hollis analysis (Gamble 2005). We will conduct sensitivity analysis with Stata statistical software (STATA).

We will only conduct sensitivity analysis for primary outcomes.

Fragility index

We will calculate the fragility index (FI) when the RR is significant (P value = < 0.05). FI is a measure used to identify the number of events required to change statistically significant results to non‐significant results (Walsh 2014). We will apply the FI only to RCTs that allocate in a 1:1 ratio, and to binary data (e.g. all‐cause mortality). We will estimate the FI with the Fragility Index Calculator.

Bayes factors

We will estimate the threshold for clinical relevance through use of Bayes factors (Jakobsen 2014). The Bayes factor is a likelihood ratio that indicates the relative strength of evidence for two theories (Dienes 2014; Dienes 2018Goodman 1999; Goodman 2005). The Bayes factor is a comparison of how well two hypotheses (the null hypothesis ‐ H0 ‐ and the alternative hypothesis ‐ H1) predict the data (Goodman 1999). The Bayes factor provides a continuous measure of evidence for H1 over H0. When the Bayes factor is 1, evidence is insensitive, the data are equally well predicted by both models and the evidence does not favor either model over the other (1 means the data are as well predicted by H1 as H0, so it should not be interpreted as favoring H0; rather the evidence does not point either way). As the Bayes factor increases above 1 (towards infinity) the evidence favors H1 over H0. As the Bayes factor decreases below 1 (towards 0) the evidence favors H0 over H1 (Dienes 2008; Dienes 2014; Dienes 2018). We will estimate the Bayes factor for primary outcomes.

Despite use of Bayes factors, we will base the conclusions of this Cochrane Review on the RevMan analysis.

Table 1. Template for the 'Summary of findings' table

Citicoline for treating people with acute ischemic stroke

Patient or population: people with acute ischemic stroke
Settings: inpatients
Intervention: citicoline

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Control

Citicoline

All‐cause mortality (at any time of trial)

Study population

Degree of disability or dependence in daily activities
Modified Rankin scale (from 0 to 6): lower score indicates better outcome

Study population

Adverse events assessed

Study population

Functional recovery
Barthel Index

Score 0 to 100 (higher score indicates greater independence)

Study population

Neurological function
National Institutes of Health Stroke Scale

(0 (no impairment) to a maximum of 42)

Study population

Quality of life (at the time of follow‐up)
SF‐36, Euro‐Qol, SS‐Qo

Study population

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: Confidence interval; RR: Risk ratio;

GRADE Working Group grades of evidence
High quality: further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: we are very uncertain about the estimate.

Figures and Tables -
Table 1. Template for the 'Summary of findings' table