Post-stroke Visual Impairment: A Systematic Literature Review of Types and Recovery of Visual Conditions

Structured summary 2 Provide a structured summary including, as applicable: background; objectives; data sources; study eligibility criteria, participants, and interventions; study appraisal and synthesis methods; results; limitations; conclusions and implications of key findings; systematic review registration number. 1


INTRODUCTION
Types of visual impairment following stroke can be complex including ocular as well as cortical damage [1][2][3][4][5][6]. Visual impairment can have a wide ranging impact on activities of daily living, independence and quality of life. Links with depression have also been found [7][8][9][10][11]. Many studies provide information on prevalence of various visual conditions from their sample based on cross section and case note observation studies [12][13][14][15][16][17]. Accurate estimates of prevalence or incidence of visual impairment for stroke survivors remains unknown. Determination of prevalence of visual impairment in a stroke unit is important in order to enable appropriate planning of efficacious referrals to an eye specialist for assessment, treatment and targeted advice [6,18,19].
The aim of this systematic literature review was to provide a comprehensive synthesis and exploration of reported evidence relating to visual problems after stroke with specific attention to incidence and prevalence.

Visual Impairment Definitions
Visual impairment is a deficit of visual function and includes abnormalities of peripheral vision, central vision, eye movements and a variety of perception problems [1,3,4,20].
Visual field loss is loss of a section of the field of vision and can either be central or peripheral. Following stroke visual field loss is frequently homonymous, with a loss in the same half of the visual field of both eyes. The types of visual field loss can include, hemianopia, quadrantanopia, constriction and scotomas [20,21]. It is also possible to have a loss of the central area of vision.
There are a wide range of ocular motility problems which can occur as a result of stroke including strabismus, cranial nerve palsies, gaze palsies, vergence abnormalities and nystagmus [22]. Strabismus is the misalignment of the eyes, which can be longstanding from childhood or occur as a result of an insult to the extra-ocular muscles or the cranial nerves supplying them. Eye movement palsies or pareses following stroke can include cranial nerve palsy, horizontal gaze palsy and/or vertical gaze palsy. Nystagmus is a continuous oscillatory movement of the eyes and is frequently associated in which both eyes move symmetrically. It may occur in every position of gaze or only be present in certain gaze positions. A further consideration is that patients commonly have multiple defects concurrently [23].
There are a number of different perceptual problems which can occur after stroke. The most recognised is visual inattention/neglect, in which the individual does not respond or attend to visual stimuli on the affected side. Other perceptual problems are also reported such as agnosia, visual hallucinations and image movement problems [24].

METHODS
We conducted an integrative review, aiming to bring together all evidence relating to incidence, prevalence and recovery from stroke-related visual problems. The review observed and is reported according to the PRISMA guidelines (Appendix 1). This review was not registered with PROSPERO [25].

Inclusion Criteria for Considering
Studies for This Review

Types of studies
The following types of studies were included: randomised controlled trials, controlled trials, prospective and retrospective cohort studies and observational studies. Case reports and casecontrolled studies were excluded, as they specifically look at selected cases and are therefore unable to report incidence or prevalence. All languages were included and translations obtained when necessary.

Types of participants
We included studies of adult participants (aged 18 years or over) diagnosed with a visual impairment as a direct result of a stroke. Studies which included mixed populations were included if over 50% of the participants had a diagnosis of stroke and data were available for this subgroup.

Types of outcome and data
We defined incidence as the number of new cases of any visual condition occurring during a certain period in a stroke survivor population. We defined prevalence as the number of cases of any visual condition present in a stroke survivor population at a certain time. We defined a measure of recovery as being present if prevalence figures were available at more than one time point post stroke. The visual impairments included are defined below.

Visual Impairment Definitions
Visual impairment is a deficit of visual function and includes abnormalities of peripheral vision, central vision, eye movements and a variety of perception problems [1,3,4,20].
Visual field loss is loss of a section of the field of vision and can either be central or peripheral. Following stroke visual field loss is frequently homonymous, with a loss in the same half of the visual field of both eyes. The types of visual field loss can include, hemianopia, quadrantanopia, constriction and scotomas [20,21]. It is also possible to have a loss of the central area of vision.
There are a wide range of ocular motility problems which can occur as a result of stroke including strabismus, cranial nerve palsies, gaze palsies, vergence abnormalities and nystagmus [22]. Strabismus is the misalignment of the eyes, which can be longstanding from childhood or occur as a result of an insult to the extra-ocular muscles or the cranial nerves supplying them. Eye movement palsies or paresis following stroke can include cranial nerve palsy, horizontal gaze palsy and/or vertical gaze palsy. Nystagmus is a continuous oscillatory movement of the eyes and is frequently associated in which both eyes move symmetrically. It may occur in every position of gaze or only be present in certain gaze positions. A further consideration is that patients commonly have multiple defects concurrently [23].
There are a number of different perceptual problems which can occur after stroke. The most recognised is visual inattention/neglect, in which the individual does not respond or attend to visual stimuli on the affected side. Other perceptual problems are also reported such as agnosia, visual hallucinations and image movement problems [24].

Search Methods for Identification of Studies
We used systematic strategies to search key electronic databases and contacted known individuals conducting research in stroke and visual impairment. We searched Cochrane registers and electronic bibliographic databases (Appendix 2). In an effort to identify further published, unpublished and ongoing trials, we searched registers of ongoing trials, handsearched journals and conference transactions, performed citation tracking using Web of Science Cited Reference Search for all included studies, searched the reference lists of included trials and review articles about vision after acquired brain injury and contacted experts in the field (including authors of included trials, and excluded studies identified as possible preliminary or pilot work). Search terms included a comprehensive range of MeSH terms and alternatives in relation to stroke and visual conditions (Appendix 2).

Selection of Studies
The titles and abstracts identified from the search were independently screened by two authors (FR, LH) using the pre-stated inclusion criteria. The full papers of any studies considered potentially relevant were then considered and the selection criteria applied independently by two reviewers (FR, LH). In the case of disagreement for inclusion of studies, an option was available to obtain a third author opinion (CN).

Data Extraction
A pre-designed data extraction form was used which gathered information on sample size, study design, assessments undertaken, visual conditions reported, timing of assessment and population type. Data was extracted and documented by one researcher (LH) and verified by another (FR).

Data Analysis
Due to the heterogeneous nature of the studies, a narrative analysis was undertaken. The exception to this was a calculation to estimate the prevalence of overall visual impairment following stroke. Strict criteria of only studies using consecutive recruitment from a stroke population and reporting an overall prevalence for visual impairment were used for the mean prevalence calculation.

Quality Assessment
To assess the quality of the studies included in this review, two checklists were considered relevant to the study designs in our inclusion criteria: the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) checklist [26,27]. The checklist was adapted as the original was designed to assess the quality of reporting rather than the potential for bias within a study. There is currently no 'gold standard' quality assessment tool for observational studies [28]. The STROBE Statement covers 22 items covering the whole of the articles from introduction, method, results and discussion, which are important to consider when assessing the quality of observation studies (including cohort, case-control and crosssectional studies). The adapted version used in this review included 18 items; only the information which is pertinent to quality appraisal of the studies was included. Using Boyle's recommendations for the evaluation of prevalence studies, the items exclude which were not considered relevant information, such as the title, abstract, background, setting and funding [29].

RESULTS OF THE SEARCH
The search results are outlined in Appendix 3. Sixty-four articles (26,321 participants) were included. Of the 64 included studies, none of which were RCTs, 52 were prospective observational studies and 12 were retrospective analyses. Consequently quality of study was assessed using the STROBE checklist. Although none of the studies were RCTs, one study was a retrospective analysis of data from an RCT archive [30]. Studies excluded from this review are outlined in a Appendix 4. Quality appraisal using the adapted STROBE checklist is outlined in a Appendix 5.
Seven of the studies (14,573 participants) reported on overall visual impairment. Nineteen of the studies (17,924 participants) reported on visual field defects; 22 of the studies (4330 participants) reported on ocular alignment and motility defects; nine of the studies (2097 participants) reported on central vision problems; and 13 of the studies (2885 participants) reported on types of perceptual visual deficits following stroke (including visual neglect/inattention, visual hallucinations, agnosia and reduced stereopsis). Several studies reported on two or more of these categories.
None of the studies included had a specific primary aim to calculate either prevalence or incidence of visual impairment following stroke. Fifty five studies were studies specifically investigated visual impairment following stroke, this included studies looking at specific visual problems such as visual inattention. The remaining 16 studies investigated symptoms and signs of stroke, which included reported visual impairment.

QUALITY OF THE EVIDENCE
Three paper reported 100% of the items requested by the adapted STROBE checklist [31]. Sixteen papers reported 90% or more of the requested items, 51 papers reported 75% or more. Sixty-one reported 50% or more and three papers failed to reach 50%, achieving 17%, 33% and 39% [32][33][34]. Only 36% of papers reported limitations of their studies. Results from all papers were reported and the individual results for each paper are outlined in a Appendix 5.

Visual Impairment
Our search of the literature did not reveal any studies that specifically aimed to assess the incidence of visual impairment following stroke. We identified a number of studies that report an overall figure of prevalence for visual impairment. All these studies, however, were judged to have limitations relating to the methods of recruitment or assessment. Thus a calculation of incidence was not possible and estimates are calculated for prevalence.
Three prospective studies of stroke populations (n=709) report an average prevalence of visual impairment post stroke of 65% ranging from 62-71% (Table 1) [32,33,35]. These studies evaluated a general stroke population including medical and orthoptic assessments undertaken during the acute stroke phase within one week of onset to three months post stroke onset. Further to these three studies of general stroke populations, one prospective study (n=915) recruited a sub population of stroke survivors with suspected visual impairment who received full orthoptic assessment, typically within three weeks of stroke onset [6]. They reported a prevalence of 92% visual impairment. It is unknown what was missed from the general stroke population as not all individuals can report visual symptoms and referrals were evaluated to be more accurate when visual symptoms were taken into consideration in addition to ocular signs in comparison to ocular signs alone [36]. Ali et al., analysed results from a database for stroke survivors recruited to a variety of strokerelated clinical trials and reported a baseline prevalence of 60% visual impairment [30]. This cohort would typically include those who are able and willing to participate in a clinical trial and are therefore, not representative of the whole population, for example individuals with cognitive impairment and aphasia are less likely to be recruited [37].
Three studies (n=13,541) used a stroke assessment tool (NIHSS ± status questionnaire) which only partly assesses visual function [30,31,38]. The National Institute of Health Stroke Scale (NIHSS) is an assessment tool that only assesses for the presence of visual field loss and horizontal gaze problems [39]. Thus it is not a full assessment of the possible visual problems which can manifest as a result of stroke. It can therefore be argued that the numbers presented by these studies are not a true measure of overall incidence of visual impairment following stroke. In addition to the NIHSS, the Questionnaire for Verifying Strokefree Status (QVSFS) was used. However this questionnaire only asks the patient about painless complete or partial vision loss [40]. The range of overall incidence of visual problems was 19-25.9% from these studies which was considerably less than studies with more comprehensive vision assessment methods.

Visual Field Loss
The reported prevalence of visual field loss after stroke varies considerably in the literature from 5.5% to 57% (Table 2) and most probably due to its dependence on the type and affected area of a stroke, inclusion criteria and the timing of assessments and the method of testing used [41][42][43][44].
Seven studies (n=1210) recruited stroke patients consecutively either as they were admitted to hospital acute stroke units or rehabilitation wards. Assessment of visual fields by confrontation and/or perimetry on admission after stroke onset detected visual field loss in up to 57% [32,33,41,45-48]. The mean prevalence of visual field loss after stroke was calculated as 31% [32,33,41,45-48]. These studies typically assessed patients in the acute phase with homonymous hemianopia or quadrantanopia defects most frequently detected.
In addition to the above studies, seven prospective studies (n=15,388) of stroke subpopulations report prevalence of visual field loss [21,30,43,49-51]. These sub-populations typically include only stroke survivors with hemianopic or quadrantanopic field loss or with suspected visual impairment of any type, or do not recruit consecutively. Thus reported prevalence is not representative of the full stroke population.
Prevalence of visual field loss has been described based on symptom reporting by patients in four studies (n=1362) ranging from 14.6 to 22.7% [42,52-54]. These reports are considerably lower and likely reflecting the poor reliability of detection by patient reported symptoms. In addition to those formally diagnosed with visual field loss following stroke, it is important to consider how many patients are unaware of their visual loss. Celesia et al. conducted a prospective observation study (n=32) to investigate the presence of hemianopic anosognosia [54]. From a sample of thirty two patients with homonymous visual field loss, 62% were unaware of their visual deficit. In a recent paper it was reported that only 45% of participants with visual field loss reported symptoms of the visual field loss [36]. It is important to note that not all patients had isolated visual field loss. Multiple visual impairments caused by stroke were reported such as visual acuity loss, eye movement abnormalities and perceptual difficulties. This discrepancy between those who do not complain of symptoms and have a diagnosis of visual field loss may highlight an under estimation in the incidence in this and other studies.
For studies whose population samples have solely included patients with visual field loss post stroke, it is not possible to establish prevalence. However, several of these studies have shown almost equal numbers suffering right or left defects [34,44,55,56].

Ocular Motility/Strabismus
Three prospective studies (n=1262) reported an average prevalence of all ocular motility problems as 33% (Table 3) with a range from 22% to 54%, [18,35,57]. Assessments were usually within the acute period and two studies used detailed orthoptic evaluation of eye movements and binocular vision [18,35]. Methods of ocular motility assessment are important to the accuracy of identification of eye movement abnormalities to ensure full detection of deficits in various gaze positions.

Eye alignment
Strabismus may occur as an isolated finding or in association with ocular motility problems and is reported in 16.5% to 52% of stroke survivors recruited to three prospective observation studies (n=626), with an average prevalence of 38% [32,35,58]. These studies used validated orthoptic assessments to detect presence of strabismus, increasing their accuracy of detection. In a sub-population prospective multicentre observational study, 19% of the sample were identified with strabismus [23]. Pre-existing strabismus was acknowledged in 2.5%, thus 16.5% were considered to be a direct result of stroke. The cause of the strabismus in 70% of cases was an ocular motility defect. Only 36% were symptomatic with diplopia, which highlights an issue in relying purely on symptoms alone. This study has a risk of under-estimating the prevalence, as the sample is not representative of the whole stroke population.
Diplopia is reported as a symptom in many papers which is a result of a misalignment of the eyes and a disruption of binocular vision. Other studies have highlighted the discrepancy between patients who do or do not report diplopia in the presence of strabismus or ocular motility defects. There is a risk that a proportion is not captured, if the symptom of diplopia is relied upon to identify ocular motility defects. The majority of studies reporting the incidence of diplopia limit recruitment to include strokes affecting specific areas of the brain [43, 59,60], are retrospective [42,53] or required informed consent [61]. These studies cannot be generalised to the whole stroke population and also carry a risk of under estimating the true prevalence of strabismus.

Eye movement palsy
Seven studies (n=2783) report figures for gaze palsies including horizontal and/or vertical gaze positions and have a mean prevalence following stroke of 26% (range 18-44%) [22,32,35,43,57,62,63]. These defects may occur in isolation or in conjunction with other visual problems, and are the most common of all ocular motility abnormalities [22,57]. Horizontal gaze palsies are more prevalent than vertical and complete palsies more prevalence than partial [22,32,35,63].
Cranial nerve palsies affecting the ocular motor muscles include third, fourth and sixth nerves with a mean post-stroke prevalence of 16% (range 3 to 39%) from three studies (n=2329) [18,32,43,57]. Third nerve and sixth nerve palsies are reported as being more prevalent than fourth nerve palsies in these stroke populations [18,32,64]. Where ocular movement assessment only tests horizontal gaze (such as with the NIHSS screening tool) the identification of all ocular cranial nerve palsies is limited. It is likely that more subtle nerve palsies and those involving the vertical muscles may be missed.

Nystagmus
Following stroke, nystagmus is reported in an average of 11% (range 4 to 48%) in three studies (n=438) [35,62,65]. In most prospective and retrospective studies reporting nystagmus, the specific types of nystagmus are not reported. This, in addition to lack of information regarding the method of assessment, makes it difficult to assess if the more subtle types, or nystagmus not present in primary position, have been missed. These factors increase the risk of an underestimation of prevalence. When reported, common types of acquired nystagmus are gaze evoked, multi-vector and upbeat [66]. The studies described to date, frequently report when the stroke has affected the posterior circulation, including the cerebellum [42, 60,67,68]. No studies have reported the prevalence of nystagmus in anterior circulation strokes in isolation. It is, therefore not possible to estimate the proportion of cases which are potentially missed by restricting populations to posterior circulation strokes only.

Vergence
Clisby (n=140) reported 55% of patients to have reduced convergence and/or stereopsis [32]. Rowe et al. (n=243) reported reduced convergence from the initial ten month data set of the Vision in Stroke (VIS) study [69]. Using the gold standard 'normal' attainment for convergence of 6cm, 54% were judged to have reduced convergence. However, they also reported that 26% had convergence reduced less than 10cm, which could be judged to be a more appropriate standard for an older group of patients. Siong et al. reported 21% of the recruited population to have convergence reduced less than 15 cm [61].

Visual Acuity and Central Vision Deficit
Clinical assessment of visual acuity has been used to identify those with reduced vision and up to 70% of stroke survivors (  64,70]. These were frequently reported as missing, or the glasses present were dirty, broken or the wrong prescription. An important component of central visual function is contrast sensitivity, the reduction of which can deform image perception. Contrast sensitivity function has been reported to be abnormal in 62% of stroke patients (n=16) [71]. Different areas of the spectrum are impaired depending on the lesion site. For example, participants with parietal and temporal lesions have been reported to have reduced detection of low spatial frequencies whereas those with occipital and occipito-temporal lesions had difficulty with medium to high spatial frequencies [71]. Furthermore, reduced contrast sensitivity in stroke survivors, particularly those with severe functional difficulties, has been found to be associated with reduced activities of daily living [72].
Central vision is key to activities such as reading. However, reading difficulties may be caused by a wide range of visual impairments in addition to reduced visual acuity. Rowe

Visual Perception Abnormalities
The commonest form of visual perception disorder following stroke is visual neglect or inattention. The literature reporting the prevalence of visual neglect/inattention can be difficult to interpret. Often the different types of inattention (e.g. auditory, visual, and spatial) are not separated, so it is not always possible to isolate visual inattention.

RECOVERY OF VISUAL FUNCTION
Our literature search identified just one study that appears to report the recovery of overall visual problems following stroke ( Table 6). The majority that report recovery do so for visual field loss ( Table 7). Ali et al. had the largest sample for tracking recovery of multiple visual problems following stroke [30]. However, not all visual problems were included due to the use of the NIHSS which limits assessment to visual field loss and horizontal gaze paresis. There was a variable sample size at the three time points used (baseline, 30 days and 90 days post stroke). The authors reported a reduction of visual problems to 28.2% at 30 days and a further reduction to 20.5% at 90 days, compared to the initial 60.5% at baseline. The sample size considerably decreased between baseline (n=11,900) to 30 days post stroke (n=4,965).

Visual Field Loss
Recovery of visual field loss is reported by a number of studies but across variable time periods (Table 7). The percentage of patients recovering from visual field loss ranges from 0% to 44% for complete recovery and up to 72. achieved complete recovery at four weeks [46]. The majority of 42.1% had some central recovery and the remainder had quadrantic recovery. For a patient with complete homonymous hemianopia the recovery of the macula area can appear to be only a small recovery. However, this can have a considerable functional impact such as with reading ability. They were also able to demonstrate the reduced sensitivity of the confrontation method at detecting areas of recovery. Variances in reports related to whether the baseline visual field loss was complete or partial and/or congruous versus incongruous loss along with stroke-specific or mixed populations.

Ocular Motility Abnormalities and Strabismus
Less has been reported on the recovery of ocular alignment and motility problems following a stroke (

Visual Acuity and Central Vision Deficit
Little is reported on the recovery of vision following stroke (Table 9). We found one study (n=247) that outlined the recovery of reduced vision following stroke [

Visual Perception abnormalities 6.4.1 Visual inattention
Four studies (n=5286) have reported recovery of visual neglect/inattention [30,35,79,88]. The percentage of recovery reported in the literature ranges from 29% to 78% (Table 10). In contrast to other visual impairments, patients suffering with visual neglect were more likely to require a longer stay in hospital and have a poorer prognosis for recovering function [73]. Recovery is mostly seen within 3 months post onset [30,35,79] with approximately 10% full recovery within the first 2 weeks [90].

Other perceptual deficits
One study (n=140) was found to report the recovery of visual hallucinations [89]. The authors reported that visual hallucinations (Charles Bonnet syndrome) persisted for several days or weeks after the onset of stroke before gradually subsiding. The median duration of visual hallucinations was 28 days and they stated that the first 90 days is when spontaneous recovery is most likely to occur.

LIMITATIONS AND RECOMMENDA-TIONS FOR FUTURE INCIDENCE, PREVALENCE AND RECOVERY STUDIES
None of the studies provided information about stroke survivors who were not admitted to a stroke unit/ward/rehabilitation unit. It is acknowledged that a proportion of stroke survivors have visual impairment only (usually occipital infarcts) but the numbers of these remain unknown.
The time of visual examination post stroke has a direct effect on the estimate of prevalence of visual problems that occur due to stroke. As recovery of visual conditions can occur rapidly in some cases during the first weeks post stroke, studies that assess visual function later than this early two week period are likely to detect those with persistent visual impairment. The extent of visual impairment for those with persistent visual conditions may also be misrepresented as these individuals may have had substantial improvement with only partial deficits remaining. Thus there is considerable potential for an underestimation of stroke related visual impairment.      Accuracy of non-specialist vision assessments and accuracy of screening tools and scores is likely to impact on reported prevalence figures. Where basic screening is undertaken, it is possible to miss subtle visual problems whose ocular signs are not included in the screening assessment. Thus there is the potential for underdiagnoses when the assessment is performed by the stroke team rather than an eye team specialist or where screening tools are used which only measure specific features of vision, e.g. detection of hemianopia or horizontal gaze defects only as with the NIHSS, or reliance on basic confrontation assessment rather than detailed confrontation or perimetry assessment.
Studies that report sub populations of stroke survivors are also prone to reporting bias for visual problems. Despite large sample sizes in studies that have included sub populations of stroke survivors, such as the VIS study of those already suspected of having visual impairment or studies of clinical trial databases, these studies are unlikely to be representative of the general stroke population [6, 30]. These estimates are potential under-or over-representations of the true prevalence of visual problems across all stroke survivors.
The time of the baseline assessment is crucial for studies tracking the recovery of visual impairment. If the baseline assessment is delayed, complete or partial recovery may have already taken place. Furthermore, it has not yet been accurately established at what time point recovery of each visual problem following stroke can be expected. If a study only has short period of follow-up, recovery could continue after the participant has completed the study. Both factors result in under-estimation of recovery of strokerelated visual impairment.
Future studies are required to establish the incidence for post-stroke visual impairment in the early acute period within the first week of onset. Such studies should involve a full stroke cohort with no exclusions so that visual impairment rates are comprehensively evaluated. These patients require follow-up at regular time intervals to plot change in visual impairment over the first week, first month and longer term after stroke onset to provide information on trajectory of improvement, if any, and rates for full, partial or no recovery. At baseline and follow-up visits, full specialist assessment is required such that subtle visual deficits that can cause visual impairment are not missed.

CONCLUSIONS
The literature currently available for review does not include any studies whose primary aim was to determine incidence or prevalence of visual impairment post stroke. Thus, this review can only provide estimates of prevalence for individual stroke related visual problems. The estimation of the overall prevalence of visual impairment was approximately 65% at baseline assessment. A reduction to approximately 20% is seen by three month post stroke, due to factors such as recovery, adaptation and death. The figures reported cover a wide range of prevalence for each visual problem. A variety of factors may be the cause of this wide range of figures including; the different study aims, research methods used, baseline assessments being conducted at different time points and different methods assessment. The prevalence is reported as being highest for eye movement defects, visual field loss and visual inattention. The existing literature regarding the recovery of visual problems following stroke is scarce for both individual deficits and overall visual recovery. Further prospective studies are required to establish the incidence of post-stroke visual impairment, the prevalence at various time periods post stroke and trajectory of improvement.

ETHICAL APPROVAL
It is not applicable. State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, included in the meta-analysis).

2-3
Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators.

4
Data items 11 List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made.

3-4
Risk of bias in individual studies 12 Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis.

Summary of evidence
24 Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (e.g., healthcare providers, users, and policy makers).

5-24
Limitations 25 Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level (e.g., incomplete retrieval of identified research, reporting bias).

24-25
Conclusions 26 Provide a general interpretation of the results in the context of other evidence, and implications for future research.

Funding
Funding 27 Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the systematic review.