The treatment methods for post‐stroke visual impairment: A systematic review

Abstract Aim To provide a systematic overview of interventions for stroke related visual impairments. Method A systematic review of the literature was conducted including randomized controlled trials, controlled trials, cohort studies, observational studies, systematic reviews, and retrospective medical note reviews. All languages were included and translation obtained. This review covers adult participants (aged 18 years or over) diagnosed with a visual impairment as a direct cause of a stroke. Studies which included mixed populations were included if over 50% of the participants had a diagnosis of stroke and were discussed separately. We searched scholarly online resources and hand searched articles and registers of published, unpublished, and ongoing trials. Search terms included a variety of MESH terms and alternatives in relation to stroke and visual conditions. Article selection was performed by two authors independently. Data were extracted by one author and verified by a second. The quality of the evidence and risk of bias was assessed using appropriate tools dependant on the type of article. Results Forty‐nine articles (4142 subjects) were included in the review, including an overview of four Cochrane systematic reviews. Interventions appraised included those for visual field loss, ocular motility deficits, reduced central vision, and visual perceptual deficits. Conclusion Further high quality randomized controlled trials are required to determine the effectiveness of interventions for treating post‐stroke visual impairments. For interventions which are used in practice but do not yet have an evidence base in the literature, it is imperative that these treatments be addressed and evaluated in future studies.


| INTRODUCTION
Visual impairments following stroke may include abnormalities of central and/or peripheral vision, eye movements and a variety of visual perception problems such as inattention and agnosia. The visual problems (types of visual impairment) can be complex including ocular as well as cortical damage (Jones & Shinton, 2006;Rowe et al., 2009a).
Visual impairments can have wide reaching implications on daily living, independence, and quality of life. Links with depression have also been documented in the literature (Granger, Cotter, Hamilton, & Fiedler, 1993;Nelles et al., 2001;Ramrattan et al., 2001;Tsai et al., 2003;West et al., 2002). The estimation of the overall prevalence of visual impairment is approximately 60% at the acute stage following stroke (Ali et al., 2013;Barrett et al., 2007;Clisby, 1995;Freeman & Rudge, 1987;Isaeff, Wallar, & Duncan, 1974;Rowe et al., 2009b;. A review of the individual prevalence figures and the recovery rates for each of the possible post-stroke visual impairments has been reported elsewhere in the literature (Hepworth et al., 2016).
In order to treat and manage visual impairments caused by stroke it is important to establish the range and effectiveness of the available treatment options. The aim of this literature review is to provide a comprehensive synthesis of the evidence relating to treatment of visual problems after stroke.

| METHODS
We planned an integrative review, aiming to bring together all evidence relating to intervention of stroke-related visual problems. A detailed protocol was developed prior to the review. This review was carried out as part of a larger synthesis of evidence relating to visual problems after stroke.

| Data extraction
A pre-designed data extraction form was designed. Data was extracted and documented by one author (LH) and verified by another (FR).

| Quality assessment
Two reviewers (KH and LH) independently reviewed the quality of the studies included in this review using the following four checklists. For the evaluation of the quality of evidence in randomized control and control trials, an adapted version of the CONSORT (Consolidated Standards of Reporting Trials) statement was used.
The CONSORT statement covers 25 items within the following domains; title/abstract, introduction, methods, results, discussion, and other information (Moher et al., 2010). An adapted version of the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement was used to assess the quality of crosssectional, cohort, and control studies. The STROBE statement covers 22 items from introduction, methods, results, and discussion (Elm et al., 2007). An adapted version of the PRISMA (Preferred Reporting for Systematic reviews and Meta-Analyses) statement was used to assess quality of evidence in review articles, including the three Cochrane review papers used. This covers 27 items within title, abstract, introduction, methods, results, discussion, and funding (Moher, Liberati, Tetzlaff, & Altman, 2009). Finally, an adapted version of the GRACE (Good Research for Comparative Effectiveness) statement was used for observational studies with comparative effectiveness. This statement covers 11 items within the domains of data and methods. There is no formal scoring system used in this checklist, but it is suggested that if a paper addresses the majority of the checklist items, then it is deemed reliable (Dreyer, Velentgas, Westrich, & Dubois, 2014).
The adapted version of the STROBE statement used in this review included 18 items. Only the information pertinent to quality appraisal of the studies was included. The items excluded were not considered relevant information i.e. the title/abstract, background, setting, and funding. The adapted version of the CONSORT statement included 31 items of relevance.
All domains covered in these checklists are important factors to consider when evaluating the quality of evidence and risk of bias in the reported articles. These domains were graded 'high risk', 'low risk', or 'unclear risk'. If it was clear the domain was performed then this would be described as "reported" and would be recorded as having a low risk of bias. If the domain was not included this would be described as "not reported" and deemed a high risk of bias. Insufficient evidence would be labeled as an "unclear" risk. Figure 1 illustrates the results of the search. Forty-nine articles (3,613 participants and 529 health care professionals) were included. This number includes four Cochrane reviews relating to interventions available for visual problems following stroke. In view of the high standard and rigorous methods of Cochrane reviews, the findings of these four papers are summarized as an overview, followed by a review of trials and studies not included in the Cochrane reviews. Tables 2-6 display key characteristics of the included studies. The 49 included studies consisted of four Cochrane systematic reviews, seven randomized trials, one randomized crossover trial, two non-randomized controlled trials, 27 prospective observational studies, three retrospective analysis, four prospective surveys/ questionnaires and one prospective observational study with a questionnaire. One study only used a control group for the pre-treatment data and so was treated as a prospective observational study and not a controlled trial (Woodhead et al., 2013).

| RESULTS
The included articles reported on interventions for one or a combination of two or more visual impairments. Thirty-three studies (2,233 participants and 69 health care professionals) reported on interventions for visual field loss (Table 2). Nine reported on interventions for visual inattention/ neglect (227 participants and 732 health care professionals, Table 3). Seven of the studies (1,029 participants and 529 healthcare professionals) reported on intervention for ocular motility healthcare professionals) reported on intervention for reduction of central vision (Table 5) and two (187 participants) reported on interventions for visual perceptual defects (Table 6).

| Quality of the evidence
A total of 49 articles were included in this review paper and the quality of evidence was assessed for each (Tables S1-S4). Evidence was deemed to be of good quality if the article reported ≥75% of the items on the relevant assessment checklist. Overall, nine of the reported articles scored 100% in the quality of evidence assessment. Thirty-four out of the 49 articles included in this review reported between 75 and 99% of the checklist items assessed and were deemed to have good quality. Five reported between 50 and 74% of the items. The remaining one article failed to reach 50%, achieving 26% respectively (Zihl & von Cramon, 1979).

| Visual field loss
Visual field loss can affect the peripheral and/ or central field of vision following stroke although, less frequently, the central visual field may present as an isolated defect. Visual field defects can often present with visual perceptual disorders, such as visual inattention and / or agnosia, further complicating the treatment of the visual field loss. One Cochrane review relating to visual field loss following stroke focused on three types of interventions: restitutive, compensatory and substitutive (Pollock et al., 2012a). Functional ability in performing activities of daily living was used as a primary outcome measure. Thirteen trials were identified as meeting the inclusion criteria (Bainbridge & Reding, 1994;Carter, Howard, & O'Neil, 1983;Jobke, Kasten, & Sabel, 2009;Kasten, Bunzenthal, Muller-Oehring, Mueller, & Sabel, 2007;Kasten, Wurst, Behrens-Baumen, & Sabel, 1998;Plow et al., 2010;Poggel, Kasten, & Sabel, 2004;Rossi, Kheyfets, & Reding, 1990;Roth et al., 2009;Spitzyna et al., 2007;Szlyk, Seiple, Stelmack, & McMahon, 2005;Weinberg et al., 1977Weinberg et al., , 1979. Limited meta-analyses were possible and were only completed for compensatory interventions. A key finding was the limited evidence for all interventions related to visual field loss following stroke. It was not possible to comment on the effectiveness of restitutive or substitutive interventions. Pollock, Hazleton, & Brady (2011b) reported that at least half of Orthoptists in Scotland provided typoscopes, Peli prisms, reading aids and scanning therapy to stroke patients with field loss, with advice on head postures and general information being the most frequently reported strategy.
Concurrently, Rowe et al. (2013a) reported that advice and raising awareness of the field loss were the most common forms of treatment  (Mazer et al., 2003), there were no significant differences in improved driving performance between those undertaking the useful field of view attention retraining programme (UFOV) and those receiving general computer-based training. In the Cochrane review on interventions for visual field loss (Pollock et al., 2012a), a recommendation was reached that compensatory interventions were more favorable than a placebo or control at improving specific tasks but not at aiding recovery of the visual field.
Expansion of the field by 1-48 degrees has been reported (Zihl & von Cramon, 1985), however, expansion of the visual field due to natural recovery early after stroke onset cannot be ruled out. Specific improvements, however, relate more to speed and accuracy of eye . Nelles et al. (2010) reported that such training was associated with increased activity in the ipsilateral cortex to the insult after training with reports that training is task specific. Eye search training improves eye scanning into the affected side with little objective improvement in reading, whilst reading training improves reading ability with little objective improvement on visual search (Ong et al., 2012;Zihl, 1995).
In a recent trial, combined training resulted in an improvement in both eye search and reading (Aimola et al., 2014).
Other compensatory interventions listed in the literature are the use of typoscopes, rulers, and vertical reading. Vertical reading was initially mentioned in the literature as an anecdotal report by a patient describing this as helpful with their hemianopia (Wang, 2003). It has since been stated as a rehabilitation option in review articles but no empirical evidence has been published Schuett, 2009;Trauzettel-Klosinski, 2010).
An ongoing randomized controlled trial acknowledged in the above Cochrane review is currently comparing compensatory intervention (visual search training), substitutive intervention (Peli prisms) and standard care in the form of verbal and written advice, for the treatment of hemianopia following stroke (Rowe, Barton, et al., 2014). The results are yet to be reported but intend to provide a comparison of the above treatments with regard to effectiveness.

| Substitutive treatment
Peli prisms use one or two high strength prisms, placed above and/or below the pupil, with the prism base out on the spectacle lens to the side of visual field loss (Peli, 2000). These prisms create a shift of images on the side of the visual field loss so they move to overlay on the seeing field. This in turn acts as a cue for the patient to look toward the affected side.
In a study of Peli prisms, Giorgi, Woods, and Peli (2009)  period effect as the participants were aware they would switch to a second prism. As a result, only 12% reported that they would continue to wear the first prism until they had made a comparison with the second, rather than a comparison against no prisms. Forty-four percent continued wear after trialling the second prism (Bowers et al., 2014).

| Visual restoration treatment
Visual restoration therapy (VRT) involves presenting light stimuli at the border area of visual field loss (Pollock et al., 2012a). One key difference between reported studies is the amount of training prescribed. Some studies (n = 7) prescribed a set amount of training for the whole cohort and others had allowed a range in the amount of training completed by their participants (n = 6). Not one of the studies prescribed exactly the same amount of training, rendering it difficult to make direct comparisons.
Three studies prescribed specific session length and number per week but did not specify the total length of treatment (Schmielau & Wong, 2007;Zihl & von Cramon, 1979, 1982. Across these studies, the mean reported expansion of the visual field border ranged from 1 to 11.3 degrees. Eye movement recordings were not undertaken and thus improvement in the visual field due to eye movements could not be excluded.
The majority of studies (n = 7) prescribed variable session lengths and numbers. The length of session varied from 30 min to 1 hr for around six months of training (Gall & Sabel, 2012;Mueller, Mast, & Sabel, 2007;Poggel et al., 2007;Reinhard et al., 2005

T A B L E 3 (Continued)
Kruse, Wolf, & Guenther, 2013). The shorter sessions were repeated more than once per day, adding up to a possible maximum per day commitment of 70 min. The frequency of training varied between six times per week and daily.
A number of studies reported expansion of the visual field following treatment (Bergsma, Elshout, van der Wildt, & van den Berg, 2012;Mueller et al., 2007;Romano et al., 2008). However, for studies in which fixation was controlled and assessed using the scanning laser ophthalmoscope, little or no change in the visual field area was noted (Marshall, Chmayssani, O'Brien, Handy, & Greenstein, 2010;Reinhard et al., 2005;Sabel et al., 2004). Despite little or no improvement in the visual field area, patients reported an improvement in quality of life and ADL, such as mobility and reading (Bergsma et al., 2012;Gall & Sabel, 2012;Mueller et al., 2007;Plow, Obretenova, Fregni, Pascual-Leone, & Merabet, 2012;Sabel et al., 2004). Although not statistically significant, reports of visual hallucination or less dense areas of visual field loss were also more likely to show improvement (Poggel et al., 2007;Sabel et al., 2013). The majority of studies recruited patients with chronic homonymous hemianopia (longer than six months post onset). Recruitment within three to six months post-stroke could not rule out natural recovery . Thus, subjective improvements noted by patients are more likely to represent adaptation to the visual field defect.

| Strabismus and ocular motility
Strabismus pertains to misalignment of the two eyes such that one eye does not point in the same direction as the fellow eye. Ocular

| Pharmacology treatment
A Cochrane review relating to eye movement defects following stroke focused solely on pharmacologic interventions for nystagmus, as no trials relating to restitutive, compensatory or substitutive treatments were found specifically for stroke populations with other ocular motility disorders (Pollock et al., 2011). Functional ability in performing activities of daily living was used as a primary outcome measure. Two trials were identified as meeting the inclusion criteria, which included a limited number of stroke patients (n = 5) (Leigh, Burnstine, Ruff, & Kasmer, 1991;Strupp et al., 2003). In view of the limited number of trials identified and the limited number of stroke patients included, the authors recommended a wider review of interventions in acquired brain injury (ABI) populations. This Cochrane review is now on-going (Rowe, Noonan, et al., 2014).
Further temporary intervention for ocular misalignment is botulinum toxin (BT) which has been reported widely in the literature for its use with strabismus (Rowe & Noonan, 2012). Its effects are reported to last for around three months. BT can also be helpful when planning a more permanent intervention such as ocular muscle surgery.

| Substitutive treatment
Prisms are commonly used in clinical practice for the treatment and amelioration of the symptom of diplopia. Prisms may take the form of a temporary Fresnel prism or with a permanent prism ground into a spectacle lens. The theory of prisms is that the image of the object is shifted by a magnitude proportional to the strength of the prism, thus compensating for the eye misalignment (Firth & Whittle, 1994). The images are moved such that they overlap and allow the brain to fuse the images back to one image, in cases where the patient has potential for binocular single vision. Alternatively, the images are moved so they are separated to place the second image into a pre-existing visual suppression area or, separated to an extent so that the second image can be ignored and/or is less troublesome for the patient.

| Compensatory treatment
There are occasions when the use of prisms is not suitable, such as the deviation being too large and the presence of torsion or variable deviations (Firth & Whittle, 1994). In these circumstances occlusion can be used, which is frequently in the form of an opaque patch to eradi- Drug dosage increased by 1 tablet per week until patient is taking 4 tablets per day. ignored (Hadid, Wride, Griffiths, Strong, & Clarke, 2008). It is also possible to provide partial sector occlusion for patients where diplopia is only bothersome in one direction of gaze (Routt, 2011). Furthermore, advice on compensatory strategies include adaptive head postures, reading options and the use of appropriate task lighting to optimize visual function (Rowe, et al., 2013b).

| Restitutive treatment
Conservative treatment options for specific ocular motility problems, such as convergence insufficiency, include vergence exercises (Rowe et al., 2011b). Improving ocular convergence with orthoptic vergence exercises can eliminate the symptom of diplopia and asthenopia in the near position (Adler, 2002). Rowe et al. (2009a) found reduced convergence of <10 cm was present in one third of stroke survivors which frequently contributed to reading difficulty.
Once recovery has ceased and if a deviation persists, a more permanent intervention may be considered, such as ocular muscle surgery. There are a variety of procedures for the many types of ocular motility conditions, which are detailed in the literature but are not specific to stroke populations. For example, one trial (Carruthers, Kennedy, & Bagaric, 1990) reported surgical success in 92.7% of adult participants receiving surgery for horizontal strabismus compared to 50.6% of those receiving BT after 6 months.
For cases of acquired nystagmus, relatively few ophthalmologists reported the use of surgical management (Choudhuri et al., 2007). For an overview of management options for nystagmus, including pharmacological, optical, surgical, and botulinum toxin, see Thurtell and Leigh (2010).
Although many of the treatment options for eye movement disorders have not been established within a stroke population specifically, the benefit would be much the same as with other cohorts.
Furthermore, the lack of high quality clinical trials to determine the efficacy of treatments such as prisms and occlusion may not necessarily be required. It is well-established that these treatments will alleviate the symptoms of diplopia without the need of clinical trials to prove so.

| Central vision
Impaired central vision includes reduced visual acuity and contrast sensitivity. Pollock et al. (2012b) completed a Cochrane review investigating whether interventions used to treat other visual problems which are age related, also improved the functional outcome following stroke.
In addition to stroke related visual problems, the authors also included patients with cataracts, glaucoma, age-related macular degeneration, or diabetic retinopathy. They used functional ability as the primary outcome measure. Twenty four potential trials were found. However, it was not clear if these trials included stroke as a sub-group. In view of this, the authors took the decision to exclude these trials as age-related visual problems are already well-covered by other Cochrane systematic reviews: age-related macular degeneration (Casparis, Lindsley, Kuo, Sikder, & Bressler, 2012;Eandi, Giansanti, & Virgili, 2008;Evans, 2013 to their stroke and it is important that they have access to their glasses, or receive a retest for glasses after their stroke (Lotery et al., 2000). For those patients who still have reduced central vision even with glasses correction, low visual aids (LVAs) such as magnifiers may be helpful (Rowe et al., 2011b). LVAs have been shown to be effective amongst patients suffering visual impairment for a variety of reasons, such as cataracts and macular degeneration. Information on reading aids such as electronic and non-electronic optical aids, magnifiers and colored filters is available (Beasley & Davies, 2013;Virgili, Acosta, Grover, Bentley, & Giacomelli, 2013). A further systematic review addresses the use of low vision services, such as standard hospital-based services, multidisciplinary services and services with an emphasis on the psychological needs of the patient (Binns et al., 2012). Further modifications to light and environment to aid visually impaired people at home include the use of color and contrast, avoiding clutter and using accessible appliances (Cooper, 2013;Joule, Levenson, & Brown, 2014). However, these have yet to be validated in the literature for their use in a stroke population.
Overall, advice and visual aids may be of benefit to stroke survivors with central visual impairment, however, these have not yet been evaluated within a specific stroke population. Further research is required to determine the benefit of these therapies following stroke.
Meta-analyses showed no significant persistent effect either on standardized assessments or for functional ability. A subsequent trial (Machner et al., 2012) examined the effect of hemifield eye patching and optokinetic stimulation (OKS). This treatment was described as a "forced-use" therapy comprising of sector occlusion over the non-neglecting side of plano lenses and removed when completing the OKS. The results showed that both the control group and those receiving treatment had an equal improvement in neglect-related functional disability over time.

| Compensatory treatment
A survey of Occupational Therapists (Pollock, Hazelton, & Brady, 2011a) reported a high proportion delivered treatment for visual neglect (89%) and visual field defects (69%), most commonly non-computerized scanning training, activities of daily living training and provision of aids and modifications. Other compensatory methods of rehabilitation of visual neglect / inattention include occlusion and prism adaptation (Beis, André, Baumgarten, & Challier, 1999;Datié et al., 2006).
A Cochrane review meta-analysis initially showed cognitive rehabilitation to have a significant immediate effect on standardized assessments (Bowen et al., 2013). The analysis was repeated with only high quality trials included. This significant effect was not maintained.
In addition, trials which compared cognitive rehabilitation with visual scanning therapies were too heterogeneous to enable the authors to draw conclusions. In view of these findings the authors could not support or refute the interventions covered by the review. The recommendations were that clinicians should continue to follow national guidelines until further high quality evidence is available.
A further trial aimed to investigate whether or not smooth pursuit therapy is superior to standard scanning therapy (Kerkhoff et al., 2013). The authors reported more improvement following smooth pursuit training in both auditory and visual outcomes. These improvements were also seen for both mild and severe degrees of neglect with stability of improvement up to two weeks following training.
A variety of treatments have been described for visual neglect/ inattention after stroke, with compensatory scanning therapies appearing most favorable. However, due to lack of high quality evidence, these treatments cannot be recommended in clinical guidelines at present.

| Visual perceptual deficits
Visual neglect/inattention is the most frequently occurring visual perceptual disorder following stroke (Hepworth et al., 2016). Additional deficits include visual hallucinations, object agnosia, color detection problems, and difficulty judging depth (Rowe et al., 2009b).
Spontaneous recovery may occur for perceptual deficits. However, patients reported a benefit from verbal advice and coping strategies, as well as the relief associated with diagnosis and recognition of the impairment which can cause significant distress to the patient.
A Cochrane review reported on the interventions for perceptual disorders following stroke (Bowen, Knapp, Gillespie, & Nicolson, 2011), however, the relevant papers for this review have been extracted and discussed elsewhere (Edmans et al., 2000;Mazer et al., 2003).
Interventions for perceptual deficits are often reported as case studies or small retrospective cohorts. One prospective observational study used cross-modal word recognition training with a group of patients with pure alexia, which involved single words presented visually and via audio simultaneously. The group of patients were reported to read words from the training program quicker than untrained words, especially for the longer words. There was no transfer following training to letter or sentence reading. The improvement seen with words in the training program was not maintained at the follow-up visit at two to four weeks after training had finished (Woodhead et al., 2013).
A range of visual perceptual disorders can occur following stroke however, very few treatments for these have been discussed in the current literature. It is possible that a number of treatments including advice are being used in practice with no clear evidence base and as such, further research is required to establish these treatments.

| CONCLUSION
Overall, the findings from this review highlight implications for further research. There is a strong requirement for further high quality randomized controlled trials to determine the effectiveness of interventions when treating post-stroke visual impairments. Furthermore, the majority of studies included in this review used a small number of patients in their study populations. Future research must address these issues and should consider the impact of interventions.
It is important to note that some interventions have been tested on broader populations and not an isolated stroke survivor population.
However, in many visual conditions, the evidence can be applied to stroke survivors; for example, prisms have been shown to be effective in a general diplopia population and are an accepted and effective treatment.
The focus of future research should be relevant to activities of

CONFLICTS OF INTEREST
The authors have no conflicts of interest to declare.

ACKNOWLEDGMENTS
This paper presents independent research funded by the National institute for Health Research (NIHR) and the NIHR North West Coast CLAHRC.

DISCLAIMER
The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the department of health.