Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Review - Intervention

Interventions for visual field defects in people with stroke

Authors' declarations of interest

Version published: 23 May 2019 Version history

https://doi.org/10.1002/14651858.CD008388.pub3
Collapse all Expand all

Background

Visual field defects are estimated to affect 20% to 57% of people who have had a stroke. Visual field defects can affect functional ability in activities of daily living (commonly affecting mobility, reading and driving), quality of life, ability to participate in rehabilitation, and depression and anxiety following stroke. There are many interventions for visual field defects, which are proposed to work by restoring the visual field (restitution); compensating for the visual field defect by changing behaviour or activity (compensation); substituting for the visual field defect by using a device or extraneous modification (substitution); or ensuring appropriate diagnosis, referral and treatment prescription through standardised assessment or screening, or both.

Objectives

To determine the effects of interventions for people with visual field defects after stroke.

Search methods

We searched the Cochrane Stroke Group Trials Register, the Cochrane Eyes and Vision Group Trials Register, CENTRAL, MEDLINE, Embase, CINAHL, AMED, PsycINFO, and PDQT Databse, and clinical trials databases, including ClinicalTrials.gov and WHO Clinical Trials Registry, to May 2018. We also searched reference lists and trials registers, handsearched journals and conference proceedings, and contacted experts.

Selection criteria

Randomised trials in adults after stroke, where the intervention was specifically targeted at improving the visual field defect or improving the ability of the participant to cope with the visual field loss. The primary outcome was functional ability in activities of daily living and secondary outcomes included functional ability in extended activities of daily living, reading ability, visual field measures, balance, falls, depression and anxiety, discharge destination or residence after stroke, quality of life and social isolation, visual scanning, adverse events, and death.

Data collection and analysis

Two review authors independently screened abstracts, extracted data and appraised trials. We undertook an assessment of methodological quality for allocation concealment, blinding of outcome assessors, method of dealing with missing data, and other potential sources of bias. We assessed the quality of evidence for each outcome using the GRADE approach.

Main results

Twenty studies (732 randomised participants, with data for 547 participants with stroke) met the inclusion criteria for this review. However, only 10 of these studies compared the effect of an intervention with a placebo, control, or no treatment group, and eight had data which could be included in meta‐analyses. Only two of these eight studies presented data relating to our primary outcome of functional abilities in activities of daily living. One study reported evidence relating to adverse events.

Three studies (88 participants) compared a restitutive intervention with a control, but data were only available for one study (19 participants). There was very low‐quality evidence that visual restitution therapy had no effect on visual field outcomes, and a statistically significant effect on quality of life, but limitations with these data mean that there is insufficient evidence to draw any conclusions about the effectiveness of restitutive interventions as compared to control.

Four studies (193 participants) compared the effect of scanning (compensatory) training with a control or placebo intervention. There was low‐quality evidence that scanning training was more beneficial than control or placebo on quality of life, measured using the Visual Function Questionnaire (VFQ‐25) (two studies, 96 participants, mean difference (MD) 9.36, 95% confidence interval (CI) 3.10 to 15.62). However, there was low or very‐low quality evidence of no effect on measures of visual field, extended activities of daily living, reading, and scanning ability. There was low‐quality evidence of no significant increase in adverse events in people doing scanning training, as compared to no treatment.

Three studies (166 participants) compared a substitutive intervention (a type of prism) with a control. There was low or very‐low quality evidence that prisms did not have an effect on measures of activities of daily living, extended activities of daily living, reading, falls, or quality of life, and very low‐quality evidence that they may have an effect on scanning ability (one study, 39 participants, MD 9.80, 95% CI 1.91 to 17.69). There was low‐quality evidence of an increased odds of an adverse event (primarily headache) in people wearing prisms, as compared to no treatment.

One study (39 participants) compared the effect of assessment by an orthoptist to standard care (no assessment) and found very low‐quality evidence that there was no effect on measures of activities of daily living.

Due to the quality and quantity of evidence, we remain uncertain about the benefits of assessment interventions.

Authors' conclusions

There is a lack of evidence relating to the effect of interventions on our primary outcome of functional ability in activities of daily living. There is limited low‐quality evidence that compensatory scanning training may be more beneficial than placebo or control at improving quality of life, but not other outcomes. There is insufficient evidence to reach any generalised conclusions about the effect of restitutive interventions or substitutive interventions (prisms) as compared to placebo, control, or no treatment. There is low‐quality evidence that prisms may cause minor adverse events.

PICOs

Population
Intervention
Comparison
Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Interventions for visual field defects in people with stroke

Review question

Are there effective treatments for people who have visual field defects after stroke?

Background

Stroke can cause some people (20% to 57% of people with stroke) to lose the ability to see the entire space in front of them ‐ often one complete half of the normal field of vision is lost. These problems with seeing are called visual field defects. Visual field defects can make it difficult for people to function normally ‐ especially moving about freely, avoiding obstacles, reading, driving, and taking part in rehabilitation for other stroke‐related problems.

We wanted to see whether treatments for visual field defects could improve stroke survivors' abilities in activities of daily living (our primary outcome of interest), or other (secondary) outcomes. Secondary outcomes that we were interested in included the size of the visual field defect, independence (in functional abilities), quality of life, ability to scan/search for objects, reading ability, balance and falls, depression and anxiety, and adverse events.

Study characteristics

We included 20 studies (involving 547 stroke participants) that investigated the effect of treatments for visual field defects. However, only 10 of these studies compared the effect of a particular treatment with no treatment. Three of these studies investigated a type of eye movement training designed to improve the lost visual field (a 'restitutive' intervention). Four of the included studies investigated the effect of scanning training, which involves training people to 'scan' across the space in front of them and into the 'lost' visual field, in order to better cope with their lost vision (a 'compensatory' intervention). Three of the included studies investigated the effect of wearing a special prism on a pair of glasses, which increases the amount a person can see on their affected side (a 'substitutive' intervention). One of the studies investigated the effect of specialised assessment by an orthoptist (a hospital‐based vision specialist), compared to standard care.

Search date

We searched for studies up to May 2018.

Key results

Only two studies presented data relating to how treatment can improve stroke survivors' abilities in activities of daily living, and there was a lack of consistency across studies that limited our ability to draw clear conclusions. There was insufficient evidence to draw any conclusions about the effectiveness of restitutive interventions as compared to control. There was low or very low‐quality evidence that scanning training may help improve quality of life, but may have no effect on other outcomes (including adverse events). There was low or very‐low quality evidence that prisms may have an effect on ability to scan (look) for objects, but may cause a range of minor adverse events (particularly headache) and may have no effect on other outcomes. Limitations with the evidence meant that we could not draw any conclusions about the benefits of assessment interventions.

Quality of the evidence

The quality of the evidence was low to very low, and in general was insufficient to reach conclusions about the effects of interventions for people with visual field defects.

Authors' conclusions

Implications for practice

There is limited low‐quality evidence that compensatory scanning training may improve an important outcome (quality of life) in patients with visual field defects following stroke, but 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. There is insufficient evidence to reach a conclusion about the impact of compensatory scanning training on other outcomes.

There is insufficient evidence to reach generalised conclusions about the benefits of visual restitution therapy (restitutive intervention), prisms (substitutive intervention), or assessment or screening interventions for patients with visual field defects after stroke. Prisms may cause a range of adverse events, particularly headache.

Implications for research

Specific implications for research, based on the findings of this review, are outlined below.

Are further randomised controlled trials (RCTs) required?

RCTs are required to determine the effect of:

  • compensatory scanning training compared to no treatment, placebo, or usual care;

  • restitutive interventions compared to no treatment, control, or placebo;

  • substitutive interventions compared to no treatment, control, or placebo;

  • assessment or screening interventions compared to standard care.

Such RCTs must:

  • have adequate power (i.e. with an appropriate power calculation undertaken based on existing trial evidence);

  • have adequate allocation concealment, blinding of outcome assessor, and intention‐to‐treat analysis;

  • clearly define trial participants, with particular care relating to the diagnosis and inclusion of patients with visual field defects or visual neglect, or both;

  • consider the severity of the visual field defect and plan subgroup analyses, where appropriate;

  • include measures of functional ability in activities of daily living;

  • collect and report data relating to adverse events;

  • report clear and usable data.

We recommend that future RCTs concentrate on answering the specific question relating to the effectiveness of interventions compared to control, placebo, no treatment, or usual care rather than comparisons with variations of the same 'type' or category of intervention, or comparisons of different doses, adjuncts to treatment, or modes of delivery. We believe that until such time as the benefits of interventions for visual field defects compared to control, placebo, no treatment, or usual care have been established (or refuted), it is not beneficial to compare the relative effects of different interventions. We recommend that future RCTs should include outcomes that are of importance to people affected by this problem, such as quality of life, confidence, and participation in social activities and relationships. We recommend that consensus is reached relating to the important outcome measures which all future trials of interventions for visual field defects in people with stroke should assess (i.e. a core outcome set), and that stroke survivors, carers, health professionals, and researchers are involved in developing this core outcome set.

A number of RCTs are currently ongoing. Once they are completed it will be important to update this review and to re‐evaluate the need for further RCTs of interventions for visual field defects.

Are other primary research studies required?

We do not recommend other study designs aimed at investigating the effects of visual restitution therapy or prisms, although development or pilot studies may be justified to support the design of a full trial if such work has not previously been completed. Given the variation in compensatory interventions included in this review, we do recommend that other primary research studies are considered to explore an optimal scanning training intervention. Such studies may consider the evaluation of computerised as well as non‐computerised interventions, the effective dosage, and the role of reading‐specific training. However, this work ought to be done as a prequel to a well‐designed RCT of the scanning intervention. Studies evaluating different types of scanning training should provide clear details of the treatment parameters, including the eye movements which are being trained by the intervention.

There is a need for further research to identify optimal outcome measures for use within future RCTs in this area. Such research may consider the relationship between functional measures of activities of daily living (ADL), measures of vision‐related extended ADL (such as reading, driving, navigating within an environment), and laboratory measures of the visual field.

Are further systematic reviews required?

We do not recommend any further systematic reviews aimed at addressing the effectiveness of interventions for visual field defects after stroke. We do recommend that this review is updated when the ongoing RCTs are completed.

Summary of findings

  • The quality of evidence summarised in this review is judged to be low to very low.

  • Methodological quality of studies is, in general, poor or poorly reported, providing insufficient high‐quality evidence on which to reach generalisable conclusions.

  • Limited low‐quality evidence suggests compensatory interventions may improve an important outcome (quality of life) in patients with visual field defects following stroke, but 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. There is insufficient evidence to reach a conclusion about the impact of compensatory scanning training on functional activities of daily living, or other outcomes.

  • There is insufficient evidence to reach generalised conclusions about the benefits of vision restoration therapy for patients with visual field defects after stroke.

  • There is insufficient evidence to reach generalised conclusions about the benefits of prisms for patients with visual field defects after stroke; there is some low‐quality evidence that prisms may cause adverse events.

  • High‐quality RCTs are needed to compare compensatory, restitutive, substitutive, and assessment interventions with placebo, control, no treatment, or usual care.

Summary of findings

Open in table viewer
Summary of findings 1. Summary of findings: Restitutive interventions versus control

Restitutive interventions compared with control for visual field defects in patients with stroke

Patient or population: stroke survivors with visual field defects

Settings: any rehabilitation setting

Intervention: restitutive interventions

Comparison: control, placebo, or no intervention

Outcomes

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Functional ability in activities of daily living

(no data)

No studies

Insufficient evidence

Visual field

(TAP border position in degrees of visual angle from zero vertical meridian)

After intervention

MD 1.02 (‐1.37 to 3.41)

19
(1 study, Kasten 1998)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study had high ROB for 'other bias' (relating to potential financial interest in the intervention), study had uncertain ROB for allocation concealment and incomplete outcome data

  • Indirectness ‐ included participants with diagnoses other than stroke

  • Imprecision ‐ very small study population (n = 19)

Extended activities of daily living

(no data)

No studies

Insufficient evidence

Reading ability

(no data)

No studies

Insufficient evidence

Falls

(no data)

No studies

Insufficient evidence

Quality of life

(improved or not improved ‐ derived from percentage of those who reported subjective improvements of vision)

OR 13.00 (2.07 to 81.48)

30*

(1 study, Kasten 1998)

*The data used in this analysis were derived from 30 of the original 38 participants, which included data from an additional 19 participants with optic nerve injury who had also received the same interventions in a separate (but parallel) trial. Participants with optic nerve injury do not meet the inclusion criteria for this review.

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study had high ROB for 'other bias' (relating to potential financial interest in the intervention), study had uncertain ROB for allocation concealment and incomplete outcome data

  • Indirectness ‐ analysis contained data from a subset of participants from a separate trial, who were not relevant to this review

  • Indirectness ‐ included participants with diagnoses other than stroke

  • Imprecision ‐ very small study population (n = 19)

Scanning ‐ cancellation

(no data)

No studies

Insufficient evidence

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.

CI: confidence interval
MD: mean difference
n: number
OR: odds ratio
ROB: risk of bias
TAP: Tuebingen Automated Perimeter

Open in table viewer
Summary of findings 2. Summary of findings: Compensative interventions versus control

Compensative interventions compared with control for visual field defects in patients with stroke

Patient or population: stroke survivors with visual field defects

Settings: any rehabilitation setting

Intervention: compensative interventions

Comparison: control, placebo, or no intervention

Outcomes

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Functional ability in activities of daily living

(no data)

No studies

Insufficient evidence

Visual field

(Functional field score and relative change in visual field score, combined)

After intervention

SMD ‐0.11 (‐0.92 to 0.70

(no significant effect)

95

(2 studies, De Haan 2015; Rowe 2010)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ one study judged as high risk of bias for at least one domain

  • Inconsistency ‐ one study had baseline differences

  • Inconsistency ‐ I2 = 75%

  • Indirectness ‐ studies explored very different compensatory interventions

Extended activities of daily living

(Mobility questionnaire and change in Nottingham EADL, combined)

After intervention

SMD 0.49 (‐0.01 to 0.99)

(no significant effect)

97

(2 studies, De Haan 2015; Rowe 2010)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ one study judged as high risk of bias for at least one domain

  • Indirectness ‐ outcome measures were very different; for one study outcome was a mobility measure, rather than a general measure of EADL

  • Indirectness ‐ studies explored very different compensatory intervention

s

Reading ability

(Reading speed; various tests)

After intervention

SMD 0.26 (‐0.05 to 0.58)

(no significant effect)

162

(4 studies, Aimola 2011; De Haan 2015; Rowe 2010; Spitzyna 2007)

⊕⊕⊝⊝
low

Reasons for downgrades:

  • Risk of bias ‐ three studies judged as high risk of bias for at least one domain

  • Indirectness ‐ studies explored very different compensatory intervention

Falls

(no data)

No studies

Insufficient evidence

Quality of life

(National Eye Institute Visual Function Questionnaire (NEI ‐ VFQ‐25) total score)

After intervention

MD 9.36 (3.10 to 15.62)

(favours compensatory)

96

(2 studies, De Haan 2015; Rowe 2010)

⊕⊕⊝⊝
low

Reasons for downgrades:

  • Risk of bias ‐ two studies judged as high risk of bias for at least one domain

  • Indirectness ‐ studies explored very different compensatory interventions

Scanning ‐ cancellation

(cancellation tests ‐ time to complete)

After intervention

SMD ‐0.01 (‐0.40 to 0.39)

(no significant effect)

97

(2 studies, Aimola 2011; De Haan 2015)

⊕⊕⊝⊝
low

Reasons for downgrades:

  • Risk of bias ‐ two studies judged as high risk of bias for at least one domain

  • Indirectness ‐ studies explored very different compensatory interventions

Adverse events

(number of participants with reported events during intervention period)

OR 5.18 (0.24 to 112.57

(favours control)

108

(2 studies, De Haan 2015; Rowe 2010)

(NB. no events recorded in De Haan 2015, which did not explicitly report adverse events as an outcome measure)

⊕⊕⊝⊝
low

Reason for downgrades:

  • Inconsistency ‐ no events from one study, means pooled result was not estimable for that study; large confidence intervals

  • Indirectness ‐ studies explored very different compensatory interventions

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.

CI: confidence intervals
EADL: extended activities of daily living
MD: mean difference
NEI‐VFQ‐25: National Eye Institute Visual Function Questionnaire
OR: odds ratio
SMD: standardised mean difference

Open in table viewer
Summary of findings 3. Summary of findings: Substitutive interventions versus control

Substitutive interventions compared with control for visual field defects in patients with stroke

Patient or population: stroke survivors with visual field defects

Settings: any rehabilitation setting

Intervention: compensative interventions

Comparison: control, placebo, or no intervention

Outcomes

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Functional ability in activities of daily living

(Barthel Index)

After 4 weeks of treatment

Wearing prisms

MD ‐4.00 (‐17.86 to 9.86)

(no significant effect)

39

(1 study, Rossi 1990)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study judged as high risk of bias for at least one domain

  • Indirectness ‐ included data from participants with neglect

  • Imprecision ‐ small study population (n = 39)

Visual field

(change in visual field area & change in error scores, from baseline)

After intervention

Not wearing prisms

SMD 0.12 (‐0.46 to 0.70)

Wearing prisms

SMD 1.12 (0.44 to 1.80)

85

(2 studies, Rossi 1990; Rowe 2010)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ one study judged as high risk of bias for at least one domain

  • Indirectness ‐ included data from participants with neglect

  • Indirectness ‐ studies cannot be combined due to differences in testing (wearing/not wearing prisms)

Extended activities of daily living

(Change in EADL from baseline; mobility improvement scores, in Logits)

After intervention

Not wearing prisms

SMD 0.20 (‐0.44 to 0.85)

Wearing prisms

SMD 0.24 (‐0.26 to 0.75)

99

(2 studies, Bowers 2014; Rowe 2010)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ one study judged as high risk of bias for at least one domain

  • Indirectness ‐ one study outcome was a mobility measure, rather than a general measure of EADL

  • Indirectness ‐ included participants with diagnoses other than stroke

  • Indirectness ‐ studies cannot be combined due to differences in testing (wearing/not wearing prisms)

Reading ability

Not wearing prisms

MD 2.80 (‐7.13 to 12.73)

(no significant effect)

45

(1 study, Rowe 2010)

⊕⊕⊝⊝
Low

Reasons for downgrades:

  • Imprecision ‐ small study population (n = 45)

  • Imprecision ‐ wide confidence intervals

Falls

(number of falls)

After intervention

Wearing prisms

OR 1.21, (0.26 to 5.76)

(no significant difference)

39

(1 study, Rossi 1990)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study judged as high risk of bias for at least one domain

  • Indirectness ‐ included data from participants with neglect

  • Imprecision ‐ small study population (n = 39)

Quality of life

(Visual Function Questionnaire (VFQ‐25))

After intervention

Not wearing prisms

MD 8.40 (‐4.18 to 20.98)

(no significant effect)

43

(1 study, Rowe 2010)

⊕⊕⊝⊝
Low

Reasons for downgrades:

  • Imprecision ‐ small study population (n = 43)

  • Imprecision ‐ wide confidence intervals

Scanning ‐ cancellation

(line cancellation errors)

After intervention

Wearing prisms

MD 9.80 (1.91 to 17.69)

(favours substitutive)

39

(1 study, Rossi 1990)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study judged as high risk of bias for at least one domain

  • Indirectness ‐ included data from participants with neglect

  • Imprecision ‐ small study population (n = 39)

  • Imprecision ‐ wide confidence intervals

Adverse events

(number of participants with reported events during intervention period)

OR 87.32 (4.87 to 1564.66)

(favours control)

59

(1 study, Rowe 2010)

⊕⊕⊝⊝
Low

Reason for downgrades:

  • Inconsistency ‐ large confidence intervals

  • Imprecision ‐ data from only one study

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.

EADL: extended activities of daily living
MD: mean difference
OR: odds ratio
SMD: standardised mean difference
VFQ‐25: Visual function questionnaire

Open in table viewer
Summary of findings 4. Summary of findings: Assessment/screening interventions versus control

Assessment/screening interventions compared with control for visual field defects in patients with stroke

Patient or population: stroke survivors with visual field defects

Settings: any rehabilitation setting

Intervention: assessment/screening interventions

Comparison: control, placebo, or no intervention

Outcomes

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Functional ability in activities of daily living

(FIM)

After intervention

MD ‐6.97 (‐23.78 to 9.84)

(no significant effect)

37

(1 study, Jarvis 2012)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study judged as high risk of bias for at least one domain

  • Imprecision ‐ small study population (n = 37)

  • Imprecision ‐ wide confidence intervals

Visual field

(no data)

No studies

Insufficient evidence

Extended activities of daily living

(no data)

No studies

Insufficient evidence

Reading ability

(no data)

No studies

Insufficient evidence

Falls

(no data)

No studies

Insufficient evidence

Quality of life

(no data)

No studies

Insufficient evidence

Scanning ‐ cancellation

(no data)

No studies

Insufficient evidence

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.

FIM: Functional Independence Measure
MD: mean difference

Background

Description of the condition

Following stroke, a common visual problem is loss of one‐half of the visual field in both eyes; this is called hemianopia or hemianopsia. As it affects the same side in both eyes, it is termed a homonymous hemianopia. For example, left hemisphere stroke causes the loss of the nasal field of the left eye and temporal (outer field) of the right eye, resulting in an inability to see to the right of the centre of the field of view. Visual field defects are common following stroke; the prevalence has been reported as being between 20% and 57% of people (Ali 2013; Rowe 2007; Rowe 2009). The extent of the loss within the visual field may vary, from the loss of the entire half of the visual field to the loss of only a portion of the affected half. It has been reported that 70% of those with visual field loss will maintain a small area of central vision (macular sparing) (Kerkhoff 1999).

The association between visual impairment and disability in activities of daily living has been well‐established (Wolter 2006). Visual field defects can affect functional ability and quality of life following stroke (Dombovy 1986; Jongbloed 1986). Studies have demonstrated that people with visual field defects have an increased risk of falling (Ramrattan 2001), and that visual field loss is a predictor of poor functional status at discharge from a stroke unit (Kaplan 1982). People report walking into objects, tripping and falling, feeling unsafe, getting lost, and experiencing panic when in crowded or unfamiliar areas (Windsor 2008). Stroke survivors may struggle with reading, and with accomplishing everyday tasks such as shopping and handling their finances (Warren 2009), and they report severe difficulty returning to work, and marked loss of self‐confidence (Rowe 2017).

Visual field loss may also impact on a person's ability to participate in rehabilitation, to live in their own home, and on depression, anxiety, social isolation, and quality of life following stroke (Hepworth 2016; Jones 2006). Visual field defects often co‐exist with visual neglect or other perceptual problems. Differentiation between visual field defects and visual neglect can be difficult (Jones 2006).

Description of the intervention

There are many different treatment and management approaches available for people with visual field defects. This review considered any intervention that is specifically targeted at improving the visual field defect or improving the ability of the person to cope with the visual field loss.

Treatments for visual field defects can be described as restitution, compensation or substitution (Hämäläinen 2004; Kerkhoff 2000). In addition to these types of treatments, this review also considered assessment and screening interventions that are specifically targeted at people with visual field defects.

These interventions may include, but are not limited to, the following.

  • Restitutive interventions: visual field training, contrast sensitivity training, fusional (binocular vision) training.

  • Compensatory interventions: saccadic (fast, simultaneous) eye movement training, training in visual search strategies, training eye movements for reading, use of eye blinks or colour cues, training in activities of daily living.

  • Substitutive interventions: prisms (Peli 2000; Rossi 1990), eye patches, adapted lighting, magnification, environmental modification.

  • Assessment and screening interventions: standardised visual assessment, screening and referral for visual assessment and intervention.

These are all complex interventions and, therefore, there can be substantial variations, even within interventions of the same type. For example, there can be differences between interventions in relation to the mode of delivery (e.g. therapist‐delivered, self‐directed, or computer‐based), personnel involved in delivery (e.g. vision specialists, such as orthoptists; stroke‐care rehabilitation professionals, such as occupational therapists), and in the dose of the intervention (amount of training per day, or per week, and length of intervention period).

How the intervention might work

Interventions for visual field defects are proposed to work by either restoring the visual field (restitution); compensating (adapting) for the visual field defect by changing behaviour or activity (compensation); substituting for the visual field defect by using a device or extraneous modification (substitution); or ensuring appropriate diagnosis, referral and treatment prescription through standardised assessment or screening, or both.

Restitution

This includes the biochemical events that help restore functional neural (nervous system) tissue, for example, the reduction of oedema, absorption of blood, restoration of normal neuronal physiology and restoration of axon (part of a nerve cell) transport. In the past, researchers thought that restitutive approaches would have limited effect in visual rehabilitation (Kerkhoff 2000). However, in the last decade, reports have suggested that expansion of the visual field can be achieved with specific interventions after brain or optic nerve injury (Romano 2008; Sabel 2000; Sahraie 2006). Restitutive interventions include those where there is direct training of the impaired function or repetitive stimulation of the visual field. Commercially available treatments, including NovaVision® Visual Restorative Therapy, give people repeated exposure to stimuli targeting a vision deficit in a direct attempt to help activate the brain to restore vision (NovaVision 2009).

Compensation

Compensation aims to improve the mismatch between the skills of those affected and the demands placed on them by their environment by teaching them to compensate using a spared or intact function (Kerkhoff 1999; Kerkhoff 2000; Peli 2000). Interventions include teaching people compensatory visual search or scanning techniques, varying from simple training strategies to interventions using computerised scanning schedules and specially‐developed commercially available tools (such as NVT 2009). When describing interventions for visual field defects, the term compensation may be used synonymously with the term adaptation, but we use the term compensation throughout this review.

Substitution

Substitution involves adaptation of visual components that have been lost or disrupted through the use of optical devices or environmental modifications (Kerkhoff 1999; Kerkhoff 2000; Peli 2000). Optical devices can include prisms, which shift the image received into an area that can be perceived, and typoscopes, which provide a guide for eye movements when reading.

Assessment and screening interventions

These may work by ensuring that the visual field defect is appropriately diagnosed, enabling other interventions to be prescribed. In those who have co‐existing visual field defects and visual neglect; determining the action of an effective intervention can be difficult. The co‐existence of visual neglect could prevent interventions aimed at the visual field defect from working effectively. For example, teaching scanning techniques to people with visual field defects may help them learn to compensate by scanning to the affected field but may not be as effective in people with co‐existing visual neglect.

Why it is important to do this review

The services available to people with visual problems following stroke are presently inconsistent. There are various treatment and management approaches that are available for people with visual field defects. However, these are used to varying degrees in clinical practice (Rowe 2014), and clinicians lack research‐based guidance on the choice of treatment (Hanna 2017). There is a recognised need to identify the evidence base for treatments for visual problems following stroke.There is considerable controversy and debate about the effectiveness of restitutive interventions (Horton 2005a; Horton 2005b; Reinhard 2005; Sabel 2006). There are a number of published reviews of the literature relating to interventions for visual problems following stroke. Barrett 2009 and Riggs 2007 provided reviews of visual problems, which included a small section relating to visual field problems after stroke; both of these reviews were limited in their scope and methodological quality. Bouwmeester 2007 completed a systematic review of the effect of one intervention (visual training) on visual field defects in people with brain damage, including stroke. Lane 2008 provided a narrative review of evidence for interventions for visual field loss. Das 2010 provided a narrative review with an emphasis on restitutive interventions, and primarily discussed a range of cohort studies that used localised, repetitive perceptual training aimed at reversing visual field loss induced by cortical damage. Hanna 2017 provided a narrative review of evidence for visual impairments, including visual field defect, summarising evidence from earlier versions of this review and evidence from non‐randomised studies. There is agreement amongst these reviews that there is a need for high‐quality studies of the effectiveness of interventions for visual field defects. In order to determine the current evidence for the effectiveness of any treatment or management approaches for stroke patients with visual field defects, and to enable appropriate planning and prioritisation of future primary research, it is essential that there is an up‐to‐date high‐quality systematic review of the existing evidence base.

Objectives

The key objective of this review was to determine the effects of interventions for people with visual field defects after stroke.

Research questions

  • Do interventions for visual field defects improve functional ability following stroke?

  • Are interventions for visual field defects more effective at improving functional ability in people with a visual field defect only than in those both with a co‐existing visual field defect and visual perceptual problems?

Specific objectives

  • To determine if in 1) all participants with visual field defects following stroke (with or without visual perceptual problems), 2) those with visual field defects and no visual perceptual problems, and 3) those with co‐existing visual field defects and visual perceptual problems:

    • restitutive interventions are more effective than control, placebo, or no intervention at improving functional ability in activities of daily living;

    • compensative interventions are more effective than control, placebo, or no intervention at improving functional ability in activities of daily living;

    • substitutive interventions are more effective than control, placebo, or no intervention at improving functional ability in activities of daily living;

    • assessment and screening interventions are more effective than control, placebo, or no intervention at improving functional ability in activities of daily living;

    • any one active intervention is more effective than any other active intervention at improving functional ability in activities of daily living.

  • To determine if in 1) all participants with visual field defects following stroke (with or without visual perceptual problems), 2) those with visual field defects and no visual perceptual problems, and 3) those with co‐existing visual field defects and visual perceptual problems:

    • restitutive interventions are more effective than control, placebo, or no intervention at improving secondary outcomes;

    • compensatory interventions are more effective than control, placebo, or no intervention at improving secondary outcomes;

    • substitutive interventions are more effective than control, placebo, or no intervention at improving secondary outcomes;

    • assessment and screening interventions are more effective than control, placebo, or no intervention at improving secondary outcomes;

    • any one active intervention is more effective than any other active intervention at improving secondary outcomes.

  • To explore the relationship between participant characteristics and the effect of interventions aimed at improving functional abilities in activities of daily living using subgroup analysis.

  • To make specific recommendations for future research into the effectiveness of interventions for visual field defects based on a knowledge of the existing evidence base.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs) and randomised controlled cross‐over trials (the first phase analysed as a parallel‐group trial).

Types of participants

Adult participants (over 18 years of age) after stroke (using the World Health Organization (WHO) definition of stroke, or a clinical definition if not specifically stated; that is, signs and symptoms persisting longer than 24 hours) who have been diagnosed as having a visual field defect. 

Where studies included participants with visual field defects due to reasons other than stroke (e.g. traumatic brain injury), in addition to participants with visual field defects due to stroke, we included these studies. We documented the number of participants with each clinical diagnosis, and planned to use this information when exploring heterogeneity.

We defined a visual field defect as a homonymous loss of vision contralateral to the side of the lesion. We accepted a clinical diagnosis of visual field defect. We documented the method of diagnosing a visual field defect. 

We excluded participants with monocular visual field defects due to retinal stroke.

Types of interventions

We included any intervention that was specifically targeted at improving the visual field defect or improving the ability of the participant to cope with the visual field loss. We classified interventions as either restitution, compensation, substitution, or assessment and screening (see Description of the intervention).

We compared interventions with no treatment, placebo, and control, within four specific preplanned comparisons:

  • restitutive interventions versus no treatment, placebo, or control;

  • compensatory interventions versus no treatment, placebo, or control;

  • substitutive interventions versus no treatment, placebo, or control;

  • assessment and screening interventions versus standard care.

We considered studies which compared one active intervention with another active intervention within a narrative synthesis. We did not plan to conduct any meta‐analyses comparing one active intervention with another active intervention as we anticipated that there would be substantial variation in the interventions, and that it would not make sense to combine the results.

Two review authors (CH, AP) independently classified the types of interventions in each included trial as either restitution, compensation, substitution, or assessment and screening. We anticipated that we might experience some difficulties in the classification of some interventions, in particular, the classification of interventions as either restitutory or compensatory, and had planned to reach consensus through discussion, involving a third review author when necessary. If there was uncertainty about the action of a particular intervention, we planned to carry out sensitivity analyses to explore the effect of removing and including the relevant trial(s). However, the two independent review authors agreed on all classifications and did not require further discussion with a third review author.

Types of outcome measures

Where possible, we assessed the outcome at the end of the intervention period and at a follow‐up point (ideally six months after the intervention had finished, but we accepted any follow‐up point after the intervention period had finished, documenting the time point).

Primary outcomes
Functional ability in activities of daily living (ADL)

We included the following validated scales: Barthel Activities of Daily Living Index (Mahoney 1965), Functional Independence Measure (FIM) (Smith 1990), modified Rankin Scale (mRS) (Wilson 2002), Katz Index of Activities of Daily Living (Katz 1963), and Rehabilitation Activities Profile (Van Bennekom 1995). If more than one of these functional ability scales was reported, we used the scale appearing earliest in our list.

Secondary outcomes

We included the following secondary outcomes. We prestated outcome measurement tools/scales which we anticipated, and planned that if more than one of the scales or measures was reported, we would use the scale appearing earliest in our list. If additional tools/scales were reported, but none from our prestated list, we included these.

  • Functional ability in extended activities of daily living (EADL): Nottingham Extended Activities of Daily Living scale (Nouri 1987), Lawton Instrumental Activities of Daily Living (Lawton 1969), Frenchay Activities Index (Holbrook 1983), Rivermead Activities of Daily Living (ADL) score (Lincoln 1990).

  • Reading ‐ reading ability: reading speed (text reading time), reading accuracy (Wide Range Achievement Test (WRAT)(Wilkinson 2006), Gray Oral Reading test (Bryant 2011).

  • Visual field: visual field outcomes subdivided into 1) gross visual screening: confrontation tests, Harrington Flocks Visual Screener; 2) kinetic perimetry: Goldmann perimetry, Tangent Screen measures; 3) static perimetry: Humphrey Automated Perimetry, Tubinger Automated Perimetry (TAP), High resolution perimetry (HRP). For perimetry outcomes: when more than one measure had been taken with the same instrument we reported border position for the intact visual field and used it for analysis in preference to hit or detection rate.

  • Balance: Berg Balance Scale (Berg 1989), Functional Reach (Duncan 1990), Get‐Up and Go test (Mathias 1986), Standing Balance test, Step Test, or other standardised balance measure. We did not include measures of weight distribution or postural sway during standing as the relationship between ability to maintain balance and these outcomes is not established.

  • Falls: number of reported falls, Falls Efficacy Scale (Tinetti 1990).

  • Depression and anxiety: Hospital Anxiety and Depression scale (Zigmond 1983), Beck Depressive Inventory (Beck 1987), General Health Questionnaire (Goldberg 1979), Geriatric Depression Scale (Cinnamon 2011).

  • Discharge destination or residence after stroke: dichotomous variable ‐ discharged to previous place of residence (i.e. place of residence prior to stroke) or discharged to alternative destination.

  • Quality of life and social isolation: EQ5D (Rabin 2001), Health‐related quality of life scale (Williams 1999), Quality of Well Being scale (Kaplan 1993), SF36 (Garrett 1993).

  • Visual scanning: cancellation techniques.

  • Adverse events: any reported adverse events, excluding falls, death.

  • Death.

Search methods for identification of studies

See the methods for the Cochrane Stroke Group Specialised register. We searched for trials in all languages and arranged for the translation of trials where necessary.

Electronic searches

We searched the Cochrane Stroke Group Trials Register (May 2018), the Cochrane Eyes and Vision Group Trials Register (May 2018) and the following electronic bibliographic databases:

  • Cochrane Central Register of Controlled Trials (CENTRAL; 2015, Issue 1) in the Cochrane Library (accessed May 2018) (Appendix 1);

  • MEDLINE Ovid (1950 to 16 May 2018) (Appendix 2);

  • Embase Ovid (1980 to 16 May 2018) (Appendix 3);

  • CINAHL EBSCO (Cumulative Index to Nursing and Allied Health Literature; 1982 to 16 May 2018) (Appendix 4);

  • AMED Ovid (Allied and Complementary Medicine; 1985 to 16 May 2018) (Appendix 5);

  • PsycINFO (1967 to 16 May 2018) (Appendix 6);

  • ProQuest Dissertations & Theses (PQDT) database (1861 to 22 March 2015) (Appendix 7).

Searching other resources

In an effort to identify further published, unpublished and ongoing trials we:

  • searched the following registers of ongoing trials:

    • US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov; May 2018) (Appendix 8);

    • World Health Organization International Clinical Trials Registry Platform (apps.who.int/trialsearch; May 2018) (Appendix 8);

    • ISRCTN Registry (www.isrctn.org) (formerly known as the Current Controlled Trials; www.controlled‐trials.com) (March 2015);

    • Health Service Research Projects in Progress (wwwcf.nlm.nih.gov/hsr_project/home_proj.cfm) (March 2015);

    • National Eye Institute Clinical Studies Database (clinicalstudies.info.nih.gov/cgi/protinstitute.cgi?NEI.0.html) (March 2015);

    • Stroke Trials Registry (www.strokecenter.org/trials/) (March 2015);

  • handsearched the following journals and conference proceedings:

    • Australian Orthoptic Journal (1959 to August 2018);

    • British Orthoptic Journal (1939 to August 2003);

    • British and Irish Orthoptic Journal (2004 to August 2018);

    • International Orthoptic Association (IOA) (www.liverpool.ac.uk/orthoptics/research/search.htm) (1967 to August 2018);

    • International Strabismological Association (ISA) (1966 to August 2018);

    • Proceedings of the European Strabismological Association (ESA) (1969 to August 2018).

We searched the reference lists of included trials and review articles about vision after stroke and contacted experts in the field.

Data collection and analysis

One review author (CH or PC) ran all the electronic searches, downloaded references into bibliographic software, and removed duplicates. One review author excluded any titles which were obviously not related to stroke and vision (one of CH, PC, or AP). We obtained the abstracts for remaining references and two review authors (two of CH, AP, PC, SJ, AK) independently considered each of these abstracts, excluded any studies that were clearly not RCTs or cross‐over trials, and excluded any studies where the intervention was not specifically aimed at improving the visual field defect or the participant's ability to cope with the visual field defect. The review authors resolved any disagreements through discussion, involving a third review author where necessary. We obtained the full papers for any studies included at this stage.

Selection of studies

Two review authors independently applied the selection criteria by considering and documenting the type of studies, type of participants, intervention, comparison intervention, and the outcome measures (two of AP, CH, SJ, AK). Each review author classified studies as 'include' or 'exclude'. If there was disagreement between these two review authors, they reached consensus through discussions involving a third review author.

We listed any excluded studies that included participants with visual field defects in the Characteristics of excluded studies table and provided the reason for exclusion. We did not list studies that were excluded because they included participants who did not have visual field defects (i.e. visual neglect, eye‐movement disorders, age‐related visual problems) in the Characteristics of excluded studies table unless the two review authors agreed that there was a clear reason to do so.

Data extraction and management

We used a pre‐designed data extraction form to extract data from the included studies. Two review authors (two of AP, CH, SJ) independently documented the following.

  • Methods: study design, method of randomisation.

  • Participants: number of participants, inclusion criteria, time since stroke, type, nature and location of lesion. We documented the method of diagnosing the visual field defect and the type and extent of the visual field loss; the presence or absence of visual perceptual problems, and the method of diagnosis; and the country of origin of participants. We documented whether the included participants had visual field defects only (no visual neglect), co‐existing visual field defects and visual neglect, or whether the participants were a mixed group (some with and some without visual neglect). If there was a mixed group of participants, we documented whether data were available for the visual field defect‐only group and the group with co‐existent visual field and visual neglect. Where information was available, we documented the presence or absence of eye movement disorders or low vision, accepting a clinical diagnosis of these.

  • Interventions: description of interventions given to each treatment group including, if relevant, the duration, intensity, frequency and dose. We classified the type of intervention as restitution, compensation, substitution, or assessment and screening; and the type of control as no treatment, placebo, control, or standard care. We documented the professional background of the person providing the intervention (e.g. occupational therapist, orthoptist).

  • Outcomes: we documented the primary and secondary outcomes relevant to this review. If a study used a number of different methods of measuring the same outcome, we noted the outcome to be used for any subsequent analysis.

  • Notes: we noted any important confounding variables. If more than two intervention groups were included in the study, we noted the method of including these groups in any subsequent analysis.

In addition, the review authors independently documented, if data allowed, the following demographics of the included participants: age, gender, place of residence, type of stroke, side of stroke, time since stroke, initial visual field defect, and initial functional ability. The review authors resolved any data extraction discrepancies through discussion.

Assessment of risk of bias in included studies

Two independent authors (two of AP, CH, SJ) assessed risk of bias by grading the following domains as 'low risk', 'high risk' or 'unclear risk' of bias for each included study. We documented this within the 'Risk of bias' tables.

Allocation concealment

Studies with adequate concealment included those that used central randomisation at a site remote from the study, computerised allocation in which records were in a locked readable file that could be assessed only after entering participant details, or the drawing of opaque envelopes. Studies with inadequate concealment included those using an open list or table of random numbers, open computer systems, or drawing of non‐opaque envelopes. Studies with unclear concealment included those with no or inadequate information in the report.

Blinding

Adequate concealment included studies which stated that a masked outcome assessor was used, and that had masking of participants and key study personnel and did not identify any 'unmasking'. Inadequate concealment included studies that did not use masking of the outcome assessor, personnel, or participants, where there was incomplete masking, or where the report clearly identified that 'unmasking' occurred during the study. We documented concealment as unclear if a study did not state, or if there was insufficient information to judge, whether or not personnel, participants, and outcome assessors were masked. We acknowledged that for some (but not all) interventions for visual field defects, masking of personnel or participants, or both, is not possible, and considered the potential for any lack of blinding of personnel or participants, or both, to introduce bias.

Incomplete outcome data

Studies adequately addressing incomplete outcome data either had: no missing outcome data; missing outcome data that were unlikely to be related to true outcome; missing outcome data that were balanced in numbers across intervention groups, with similar reasons for missing data across groups; a reported effect size (difference in means or standardised difference in means) among missing outcomes that were not enough to have a clinically relevant impact on observed effect size; or missing data that had been imputed using appropriate methods. Studies inadequately addressing incomplete outcome data either had: missing outcome data that were likely to be related to true outcome, with either imbalance in numbers or reasons for missing data across intervention groups; a reported effect size (difference in means or standardised difference in means) among missing outcomes enough to induce clinically relevant bias in the observed effect size; as‐treated analysis done with substantial departure of the intervention received from that assigned at randomisation. We documented addressing of incomplete outcome data as unclear if there was insufficient reporting to allow this to be assessed, or if this was not addressed in the report.

Other bias

We assessed a study not to be free of other bias if it was assessed to have at least one important risk of bias, such as: a potential source of bias related to the specific study design used, an extreme baseline imbalance, a claim to have been fraudulent, financial association with the intervention, or some other problem. If there was insufficient information, or the information provided was unclear, we documented the risk of other bias as unclear.

We produced a 'Risk of bias' summary figure to illustrate the potential biases within each of the included studies.

Measures of treatment effect

We used the Review Manager software RevMan 5.3 (RevMan 2014) to carry out statistical analyses to determine the treatment effect of:

  • restitutive interventions (compared to no treatment, control, placebo, or standard care);

  • compensatory interventions (compared to no treatment, control, placebo, or standard care);

  • substitutive interventions (compared to no treatment, control, placebo, or standard care);

  • assessment and screening interventions (compared to standard care).

We used a random‐effects model for all statistical analyses. For dichotomous, variables we calculated and reported Peto odds ratios (ORs) with 95% confidence intervals (CIs). For continuous data, we calculated the treatment effect using standardised mean differences (SMDs) and 95% CIs where studies used different scales for the assessment of the same outcome, and using mean differences (MDs) and 95% CIs where all studies used the same method of measuring an outcome.

The primary outcome of functional ability in activities of daily living, and secondary outcomes of functional ability in extended activities of daily living, visual field data, balance, depression and anxiety, and quality of life and social isolation comprise either ordinal data from measurement scales, or continuous data. We analysed these as continuous variables.

Where reported outcomes had a measurement scale where a lower value is indicative of a better outcome (e.g. depression and anxiety scales) we multiplied the reported values by ‐1 so that in all analyses a higher value was indicative of a better outcome.

If studies reported change values and the baseline value was available, we calculated the value at follow‐up (change value ‐ baseline value). If studies reported change values and the baseline value was not available, we used these data in meta‐analyses but planned sensitivity analyses to investigate the effect of including the data.

We planned to analyse falls, discharge destination, adverse events, and deaths as dichotomous variables.

Unit of analysis issues

We anticipated that the majority of trials would have a parallel‐group design in which each individual participant was randomised to one of two, or more, treatment groups. Where studies had two or more active intervention groups eligible for inclusion within the same comparison (against a control, placebo, or no treatment group), we intended to 'share' the control group data between the multiple pair‐wise comparisons in order to avoid double counting of participants within an analysis.

If studies used a randomised controlled cross‐over design, we planned to analyse data from the first phase only. We did not anticipate that any studies would use a cluster‐randomised design.

Dealing with missing data

If an included study did not report a particular outcome, we did not include that study in the analysis of that outcome.

If an included study had missing data (e.g. reported means but not standard deviations for the follow‐up data), we took logical steps to enter an assumed value. Such steps included estimating a standard deviation (SD) based on a reported standard error and estimating a follow‐up SD based on a baseline value. We performed calculations of SDs from standard errors and P values using methods described in section 7.7.3.3 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We planned to do sensitivity analyses to investigate the effect of entering assumed values. We also contacted authors in an attempt to obtain missing data.

Assessment of heterogeneity

We visually assessed heterogeneity by looking at the extent of overlap of the CIs on the forest plots. We considered the P value, considering that with P < 0.1 there was likely to be heterogeneity. We considered the I2 statistic. We considered I2 > 50% as substantial heterogeneity. If I2 > 50%, we explored the individual trial characteristics to identify potential sources of heterogeneity.

Assessment of reporting biases

We attempted to avoid reporting biases by using a comprehensive search strategy that included searching for unpublished studies and searching trials registers. We planned to carry out sensitivity analyses to explore the effect of publication type.

Data synthesis

Two review authors (AP, CH, or SJ) independently extracted data from the included trials. One review author (AP) entered the data into RevMan 5.3 (RevMan 2014), and the other review author checked the entries. They resolved any disagreements through discussion, with reference to the original report.

Subgroup analysis and investigation of heterogeneity

We intended to explore heterogeneity by subgroup analyses to investigate the effect of:

  • time after stroke (zero to three months, three to six months, more than six months);

  • type of visual field defect (homonymous hemianopia, other) (We anticipated that the majority of the participants would have homonymous hemianopia. However, we documented the type of visual field defect and planned subgroup analyses to investigate the effect of including participants with types other than homonymous hemianopia);

  • extent of visual field loss if homonymous (complete hemianopia, partial hemianopia, quadrantanopia);

  • presence or absence of visual neglect (no visual neglect, all participants with co‐existing visual field defects and visual neglect, mixed group of participants some with and some without visual neglect);

  • macular sparing, macular splitting field loss;

  • type of treatment (e.g. for compensatory interventions: saccadic eye movement, activities of daily living training; for substitutive interventions: prisms, patches, environmental modifications; for assessment and screening: by orthoptist, occupational therapist, doctor).

We planned to use an established method for subgroup analyses (Deeks 2001). We planned to carry out these subgroup analyses when there were six or more studies included in a single analysis, all with sufficient information to determine the subgroups.

Sensitivity analysis

We planned to carry out sensitivity analysis to explore the effect of the following methodological features.

  • Allocation concealment: we planned to re‐analyse data, excluding trials with inadequate or unclear allocation concealment.

  • Masking of outcome assessor: we planned to re‐analyse data, excluding trials without or with unclear masking of outcome assessor.

  • Missing outcome data: we planned to re‐analyse the data, excluding trials with inadequate or unclear methods of dealing with missing outcome data.

  • Other bias: we planned to re‐analyse the data, excluding trials assessed to have other bias, or unclear as to whether they had other bias.

  • Type of intervention: we planned to re‐analyse data, excluding trials where the classification of the type of intervention was uncertain.

  • Publication type (peer‐reviewed journal, conference abstract or proceedings, doctoral dissertation): we planned to re‐analyse data including only those trials from peer‐reviewed journals.

We planned to carry out these planned sensitivity analyses when there are six or more studies included in a single analysis.

GRADE assessment and 'Summary of findings' tables

We presented the results of the main preplanned comparisons of the review in 'Summary of findings' tables;

Within each 'Summary of findings' table, we summarised data for the primary outcome of interest (functional ability in activities of daily living), the six secondary outcomes for which we had identified the greatest volume of evidence in previous versions of this review (visual field, extended activities of daily living, reading ability, falls, quality of life, scanning ‐ cancellation), and any data related to adverse events.

For each of the preplanned comparisons, we assessed quality of the evidence using the GRADE approach (Guyatt 2011a), considering each of the following criteria.

  • Risk of bias due to flawed design or conduct of studies (Guyatt 2011b).

  • Imprecision (e.g. when confidence intervals for treatment effect are wide) (Guyatt 2011d).

  • Inconsistency (e.g. when point estimates vary widely, I² is large) (Guyatt 2011e).

  • Indirectness (e.g. variations in participants, interventions, comparisons, and outcomes) (Guyatt 2011f).

  • Publication bias (may be explored with the use of funnel plots and classed as not suspected, suspected, strongly suspected or very strongly suspected) (Guyatt 2011c).

We documented identified concerns relating to any of the above criteria, and downgraded the level of evidence accordingly (one downgrade for each concern, and a maximum of two downgrades for each of the listed criteria). If there were no downgrades the level of evidence was high quality, if there was one downgrade the level of evidence was moderate quality, if there were two downgrades the level of evidence was low quality, and if there were more than two downgrades the level of evidence was very low quality. We used the following definitions of evidence.

  • High quality: when further research is very unlikely to change our confidence in the estimate of effect.

  • Moderate quality: when further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.

  • Low quality: when 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: when we are very uncertain about the estimate.

One review author (AP) assessed quality of evidence, reported identified concerns, and applied downgrades. Other review authors checked agreement with these judgements and resolved any disagreements through discussion.

Results

Description of studies

See: Characteristics of included studies; Characteristics of excluded studies; Characteristics of ongoing studies, Characteristics of studies awaiting classification.

Results of the search

Results of the search are shown in Figure 1. Our search strategy identified 17,224 records from the main electronic databases, and we identified a further eight studies through our wider search. One review author (AP, CH or PC) eliminated 15,658 titles that were clearly irrelevant, and two review authors (AP, CH) applied the inclusion criteria to the remaining 1574 abstracts, identifying 178 to be considered at full text. Of these, we identified a total of 20 studies for inclusion. In addition, we identified seven ongoing studies (see Characteristics of ongoing studies), and two studies that require further assessment (see Characteristics of studies awaiting classification).

Figure 1
Study flow diagram.

Study flow diagram.

Included studies

We included 20 studies (732 randomised participants, with data available for 638, of whom 547 (85%) had a diagnosis of ischaemic or haemorrhagic stroke) in this updated review (Aimola 2011; Bainbridge 1994; Bowers 2014; De Haan 2015; Elshout 2016; Gall 2013; Jarvis 2012; Jobke 2009; Kasten 1998; Kasten 2007; Keller 2010; Modden 2012; Plow 2010; Poggel 2004; Rossi 1990; Roth 2009; Rowe 2010; Schuett 2012; Spitzyna 2007; Szlyk 2005).

The previous version of this review included 13 studies (344 randomised participants, 285 of whom were participants with stroke) (Bainbridge 1994; Carter 1983; Jobke 2009; Kasten 1998; Kasten 2007; Plow 2010; Poggel 2004; Rossi 1990; Roth 2009; Spitzyna 2007; Szlyk 2005; Weinberg 1977; Weinberg 1979). However, we have excluded three of these studies from this update (Carter 1983, Weinberg 1977; Weinberg 1979: see Excluded studies for reason for these exclusions).

There were only abstracts available for Bainbridge 1994 and Gall 2013. In the previous version, Plow 2010 was based on data from a conference abstract and ongoing trials register data only; however, for this update a full paper was available.

We present a brief overview of the studies below. Full descriptions of the included studies can be found in the Characteristics of included studies table and in Table 1 (Demographics of included studies: settings of included studies), Table 2 (Demographics of included studies: demographics of included participants), Table 3 (Demographics of included studies: visual problems of included participants), and Table 4 (Outcome measures within included studies).

Open in table viewer
Table 1. Demographics of included studies: settings of included studies

Study

Country

Number of centres

Setting for intervention

Trial registration

Aimola 2011

UK

Multicentre ("from local hospitals or as self‐referrals")

Community

(participants' own homes)

UK Clinical Research Network Portfolio (UKCRN, ID 7144)

Bainbridge 1994

USA

Single

NS

NS

Bowers 2014

UK, USA

Multicentre (13 study sites)

University, hospital, private practice for fitting of prisms

Then use at home (participants' own homes)

clinicaltrials.gov

NCT00494676

De Haan 2015

Netherlands

2 ("Royal Dutch Visio and Bartiméus, the two centers of expertise for blind and partially sighted people in the Netherlands")

Training ... "was provided in Dutch at nine locations of Royal Dutch Visio and one location of Bartiméus in the Netherlands".

Participants were also given homework assignments.

ISRCTN Registry

ISRCTN16833414

Elshout 2016

Netherlands

Unclear ("Patients throughout the Netherlands could sign up for our study voluntarily by filling in a form on our website")

Community

(participants' own homes)

NS

Gall 2013

Not clear

NS

NS

NS

Jarvis 2012

UK

Single

Stroke unit, Warring and Halton Hospitals,

NHS Foundation Trust

NS

Jobke 2009

Germany

NS

NS

NS

Kasten 1998

Germany

NS

Community

(participants' own homes)

NS

Kasten 2007

Germany

NS

Community

(participants' own homes)

NS

Keller 2010

Germany

Single

Neurological clinic

NS

Modden 2012

Germany

Single

Rehabilitation centre (inpatients)

NS

Plow 2010

USA

Single

Outpatient

(University clinic)

clinicaltrials.gov NCT00921427

Poggel 2004

Germany

Single

Community

(participants' own homes)

NS

Rossi 1990

USA

Single

Rehabilitation

(inpatient)

NS

Roth 2009

Germany

NS

Community

(participants' own homes)

NS

Rowe 2010

UK

Multicentre ("from stroke units based in 15 United Kingdom (UK) National Health Service (NHS) trusts")

Any (hospital, community)

Current Controlled Trials ISRCTN05956042.

Schuett 2012

Unclear. Authors from Austria, UK and Germany. "All participants were native German speakers."

NS

NS

NS

Spitzyna 2007

UK

NS

Community

(participants' own homes)

NS

Szlyk 2005

USA

Single; university

Outpatient clinic

NS

NHS: National Health Service
NS: not stated
UK: United Kingdom
USA: United States of America

Open in table viewer
Table 2. Demographics of included studies: demographics of included participants

Study

Number of participants

Age

Gender

Time since stroke/lesion

Initial functional ability

Type of stroke/lesion

Side of stroke/lesion

Aimola 2011

70 participants recruited, 52 participants included in analyses

Group 1

Mean 61.4 years, SD 10.3

Group 2

Mean 63.0 years, SD 10.9

Group 1

9 F

19 M

Group 2

7 F

17 M

NS

NS

Group 1

19 ischaemic stroke

4 haemorrhagic,

4 traumatic brain injury

1 tumour

Group 2

20 ischaemic stroke

2 haemorrhagic

2 traumatic brain injury

0 tumour

Side of field defect

Group 1

L 15/R 13

Group 2

L11/R 13

Bainbridge 1994

18

NS

NS

NS

NS

NS

NS

Bowers 2014

73 randomised; 67 completed first phase (before cross‐over); 61 completed second phase (after cross‐over)

For 61 participants included after the cross‐over:

median 58 years (range 18 to 89)

For 61 participants included after the cross‐over:

M 40

F 21

For 61 participants included after the cross‐over:

median 18 months (range 3 to 396)

Overall baseline mobility difficulty, for 61 participants included after the cross‐over:

mean ‐0.17 (SD 2.31) logits for n = 31 using oblique prisms

mean ‐0.06 (SD 1.89) logits for n = 30 using horizontal prisms

For 61 participants included after the cross‐over:

hemianopia was caused by stroke for 47 (77%)

For 61 participants included after the cross‐over:

L hemianopia 39 (64%)

De Haan 2015

54 randomised; data from 49 analysed (training group n = 26, control group n = 23)

Training group 55 ± 10.1 years

Control group 57 ± 13.0 years

M 32

F 17

Training group 18 ± 22.5 months

Control group 22 ± 24.6 months

NS

Ischaemic CVA 36

Haemorrhagic CVA 5

Traumatic brain injury 3

Penetrating head trauma 1

AVM extirpation 1

combined 3

L hemianopia 33

R hemianopia 16

Elshout 2016

40 recruited; data presented from first 3 cohorts of 10 only (n = 30); data from 27 analysed

Mean age 51.2 years (range 29 to 74)

M 22

F 5

Mean 26.3 months

(range 11 to 111)

NS

5 haemorrhagic stroke

22 ischaemic stroke

L‐sided field defect 14

R‐sided field defect 13

Gall 2013

39 (alternating current stimulation n = 15, sham n = 14)

NS

NS

NS

NS

NS ("patients with post‐chiasmatic visual pathway lesions")

NS

Jarvis 2012

64 randomised (experimental group n = 33, control n = 31)

Experimental: mean 70.4 years (SD 10.8)

Control: mean 69.4 years (SD 14.5)

M 40

F 24

NS

NS

Ischaemic 56

Haemorrhage 7

Combined 1

R‐sided stroke 41

L‐sided stroke 19

Bilateral 4

Jobke 2009

21

Group 1

Mean 51.5 years, SD 14.8

Group 2

Mean 47.3 years, SD 13.4

Group 1

M 7

F 1

Group 2

M 6

F 4

Group 1

Mean 89.0 months, SD 59.9

(range 67 to 225 months)

Group 2

Mean 89.4, SD 57.6

(range 40 to 236 months)

NS

Group 1

5 stroke/ischaemia

1 brain injury

1 tumour

1 surgery

Group 2

5 stroke/ischaemia

1 meningitis;

1 injury

3 surgery

NS

Kasten 1998

19

(plus 19 with pre‐chiasmal damage)

Data are presented for full group of 38

Group 1

? Mean 47.7 years, ? SD 12.9

Group 2

? Mean 55.3 years, ? SD 16.2

It is assumed the data presented are mean and SD, but this was not stated

Group 1

M 11

F 8

Group 2

M 13

F 6

Group 1

? Mean 6.8 months, ? SD11.4

Group 2

? Mean 7.2 months, ? SD 6.3

It is assumed the data presented are mean and SD, but this was not stated

NS

19 participants with post chiasmal injury; 10 were due to stroke, 4 due to trauma and 5 due to other reasons

NS

Kasten 2007

23

Group 1

Mean 41.1 years, SD 16.9

Group 2

Mean 39.3 years, SD 10.9

Group 3

Mean 44.3 years, SD 9.1

Group 1

M 5

F 2

Group 2

M 6

F 1

Group 3

M6

F3

Group 1

10 to 83 months, Mean 34.2*, SD 30.1*

Group 2

13 to 477 months, Mean 92.7*, SD 170.6*

Group 3

10 to 143 months, Mean 47.6*, SD 54.4*

NS

Group 1

4 stroke

1 trauma

1 cerebral aneurysmal bleeding

1 hypoxia

Group 2

3 stroke

3 trauma

1 surgery

Group 3

3 stroke

2 trauma

1 surgery

1 hypoxia

2 other

NS

Keller 2010

20

Group 1

Mean 54.7 years. SD 20.4

Group 2

Mean 63.6 years, SD 13.8

Group 1

M 6

F 4

Group 2

M 6

F 4

Group 1

Mean 8.5 weeks, SD 6.7

Group 2

Mean 4.2 weeks, SD 2.1

NS

Group 1

9 vascular

1 tumour

Group 2

9 vascular

1 traumatic

Group 1

4 left hemianopia

3 right hemianopia

1 UL quandrantanopia

1 LL quandrantanopia

1 UR quandrantanopia

Group 2

3 left hemianopia

3 right hemianopia

3 UL quandrantanopia

1 LL quandrantanopia

Modden 2012

45

RT Group: Mean 58.3 ± 11.4 years

CT group: Mean 57.1 ± 8.3 years

OT group: Mean 59.0 ± 11.1 years

RT group:

M 10

F 5

CT group:

M 9

F 6

OT group:

M 7

F 8

RT group:

Mean 4.7 weeks

CT group:

Mean 4.9 weeks

OT group:

Mean 4.3 weeks

"Patients were recruited on average about 4 weeks after their stroke."

NS

RT Group:

occipital 7

temporo‐occipital 2

temporomedial 5

parahippocampal 1

CT Group:

occipital 6

temporo‐occipital 3

temporomedial 5

parahippocampal 1

OT Group*:

occipital 4

temporo‐occipital 3

temporomedial 5

parahippocampal 1

*numbers presented in paper do not add up to 15 (?)

RT group

L stroke 7

R stroke 8

CT group

L stroke 5

R stroke 10

OT group:

L stroke 5

R stroke 10

Plow 2010

12

Mean 59.6 years, SEM 3.5 years

M 5

F 7

Mean 39.8 ± 16.2 months, range 3 to 192 months

NS

Stroke 8 (7 infarct, 1 haemorrhage)

Surgical trauma 2

L‐affected side 4

R‐affected side 8

Poggel 2004

20 participants recruited. Baseline data only available for 19 (data for one dropout not reported)

Group 1

Mean 41.9 years

Range 20 to 67 years

Group 2

Mean 43.2 years

Range 30 to 61 years

Group 1

M 6

F 3

Group 2

M 6

F 4

Group 1

Mean 49.1 months, SEM ?, Range 6.7 to 189.9 months

Group 2

Mean 24.1 months, SEM 5.0, Range 6.8 to 58.3 months

NS

Group 1

vascular 1
infarct 8

cortical and radiations 4

cortical 5

Group 2

vascular 2
infarct 7,

traumatic brain injury 1
cortical and radiation 5
cortical 3

radiation 2

Group 1

L 5/R 4

Group 2

L 5/R 5

Rossi 1990

39

Group 1

Mean 72.6 years, SEM 1.8

Group 2

Mean 63.3 years, SEM 2.5

Group 1

M 10

F 8

Group 2

M 9

F 12

Group 1

Mean 4.4 weeks, SEM 0.3

Group 2

Mean 4.7 weeks, SEM 0.6

NS

Group 1

15 infarct
3 haemorrhage

Group 2

18 infarct
3 haemorrhage

Group 1

16 R/2 L

Group 2

13 R/8 L

Roth 2009

30 participants recruited (data available for 28; 2 dropouts)

Group 1

Mean 60.5 years, SD 11.0, Median 65

Group 2

Mean 60.3 years, SD 11.7,

Median 63

Group 1

4 F

11 M

Group 2

F 7

M 8

Group 1

Mean 39.20 months, SD 54.59, Median 26

Group 2

Mean 87.87 months, SD 186.66, Median16

NS

Group 1

Stroke 11
Haemorrhage 1
Head injury 1
Abscess 1
AVM 1

Group 2

Stroke 11
Haemorrhage 3
Cyst 1

Affected side

Group 1

L 8/R 7

Group 2

L 7/R 8

Rowe 2010

87 participants recruited (full results for 70 participants at 26 weeks)

Group 1

Mean 69.9 years, SD 12.9, median 68.8, IQR, 14.4

Group 2

Mean 70.9 years, SD 11.2, median 72.9, IQR, 15.2

Group 3

Mean 66.2 years, SD 11.3, median 68.2, IQR, 16.2

Group 1

4 F

22 M

Group 2

13 F

17 M

Group 3

9 F

20 M

Group 1

Mean 75.5 days, SD 45.3, median 64.5, IQR 78.0

Group 2

Mean 73.8 days, SD 49.2, median 69.0, IQR 97.0

Group 3

Mean 81.2 days, SD 48.0, median 67.0, IQR 61.0

Barthel Index score

Group 1

Mean 97.5, SD 5.5, median 100.0, IQR 0.0

Group 2

Mean 92.7, SD 11.9, median 100.0, IQR 15.0

Group 3

Mean 93.3, SD 14.7, median 100.0, IQR 5.0

Group 1

25 ischaemic
1 haemorrhage

Group 2

28 ischaemic
2 haemorrhage

Group 3

28 ischaemic
1 haemorrhage

Side of infarct

Group 1

L 9/R 16/bilateral 1

Group 2

L 17/R 13/bilateral 0

Group 3

L 11/R 17/bilateral 1

Schuett 2012

36

Group 1

Mean 64.0 years, SD 11.1, range 44 to 81

Group 2

Mean 63.7 years, SD 13.3, range 42 to 83

Group 1

3 F

15 M

Group 2

3 F

15 M

Group 1

Mean 26.6 weeks, SD 14.5, range 6 to 57

Group 2

Mean 20.1 weeks, SD 18.8, range 4 to 74

NS

Group 1

17 posterior infarction
1 tumour operation

Group 1

17 posterior infarction
1 tumour operation

Side of field loss

Group 1

L 9/R 9

Group 2

L 7/R 11

Spitzyna 2007

22

Age at symptom onset

Group 1

Range 5 to 67 years, mean 42.5*, SD 20.5*

Group 2

Range 39 to 78 years, mean 63.1*, SD 12.2*

Group 1

M 6

F 5

Group 2

M 7

F 1

Time since symptoms onset

Group 1

Range 1 to 37 years, mean 7.5*, SD 10.9*

Group 2

Range 3 months to 5 years, mean 1.6*, SD 1.7*

NS

Group 1

3 infarct

1 tuberous sclerosis

2 traumatic brain injury, 2 tumour

2 haemorrhage

1 cyst

Group 2

8 infarct

Group 1

All R

Group 2

All R

Szlyk 2005

10

Group 1

Range 16 to 74 years, mean 50.6*, SD 22.5*

Group 2

Range 34 to 73 years, mean 54.0, SD 14.4

Group 1

5 M

Group 2

5 M

NS

NS

Group 1

4 CVA

1 tumour: all occipital lobe

Group 2

4 CVA

1 AVM: all occipital lobe

Group 1

L 3/R 2

Group 2

L 4/R 1

* Figures calculated from raw data supplied in papers
AVM: arteriovenous malformation
CVA: cerebrovascular accident
CT: compensatory training
F: female
IQR: interquartile range
L: left
LL: lower left
M: male
NS: not stated
OT: occupational therapy
R: right
RT: restitutive training
SD: standard deviation
SEM: standard error of the mean
UL: upper left
UR: upper right

Open in table viewer
Table 3. Demographics of included studies: visual problems of included participants

Study

Methods of visual field assessment

Type/extent of field loss

Macular sparing

Presence of neglect?

Aimola 2011

Unspecified kinetic perimeter

Esterman measures of static superthreshold

Group 1

Hemianopia 20, quadrantanopia 8

Group 2

Hemianopia 20, quadrantanopia 4

Group 1

Mean 1.92° (SD 1.44)

Group 2

Mean 2.45° (SD 1.85)

Yes: stated "Three patients (2 in the intervention group, 1 in the control group) had comorbid neglect as confirmed with the bells test".

Bainbridge 1994

Harrington Flocks Visual Screener

Confrontation

Not stated

Not stated

Yes: no details of inclusion criteria or participants provided, but objective stated "To study the effect of ... on visual neglect or hemianopsia following stroke".

Bowers 2014

Goldmann perimetry

Not stated

Not stated

No: stated "no visual neglect". Visual neglect diagnosed with Bells test and Schenkenberg Line Bisection Test.

De Haan 2015

Goldmann perimetry

Training group

Functional field score 58 ± 7.8

Quadrantanopia 5 (3 lower left, 1 upper left, 1 lower right)

Hemianopia 21

Control group

Functional field score 64 ± 11.4

Quadrantanopia 5 (3 lower left, 2 upper left)

Hemianopia 18

Not stated

No: stated "Neglect was excluded based on the Balloons, drawings, Line Bisection and Rey Complex Figure Test."

Elshout 2016

Goldman perimetry

Humphrey perimetry

Right field loss: hemifield 4, incomplete hemifield 5, quadrant 2, scotoma 1

Left field loss: hemifield 2, incomplete hemifield 9, quadrant 1, scotoma 2

Bilaterial field loss

Incomplete: 1

"All subjects
had macular sparing of at least 2°"

No: patients with visual neglect were excluded (based on line bisection test)

Gall 2013

Standard automated perimetry

Not stated

Not stated

Not stated

Jarvis 2012

Confrontation

Ocular diagnosis:

low vision 30

visual field loss 38

eye movement deficit 41

perceptual impairment 24

("Note: patients may have had an isolated visual impairment or combined visual deficits")

Not stated

Yes: all patients with a "post‐stroke visual impairment were eligible for inclusion".

Jobke 2009

Standard automated perimetry

High resolution perimetry (HRP)

NB: It did not state whether participants had visual neglect or whether this was diagnosed.

Group 1

2 diffuse, 2 full homonymous hemianopia, 1 partial homonymous hemianopia, 1 full quadrantanopia

2 partial quadrantanopia

Group 2

4 diffuse, 2 full homonymous hemianopia, 2 partial homonymous hemianopia, 1 full quadrantanopia,

1 partial quadrantanopia

Group 1

7 sparing, 1 not sparing

Group 2

10 sparing

Not stated

Kasten 1998

Tubinger automated perimetry (TAP)

High resolution perimetry (HRP)

NB: data were presented for full group of 38 participants (including participants in parallel trial)

Group 1

TAP 90° ‐ border position, mean 3.51° (degrees of visual angle from zero vertical meridian), SEM 1.0

TAP 90° ‐ number of misses, mean 53.0, SEM 9.1

Group 2

TAP 90° ‐ border position, mean 3.43° (degrees of visual angle from zero vertical meridian), SEM 0.99

TAP 90° ‐ number of misses, mean 69.2, SEM 11.2

Not stated

No: participants with neglect were excluded. Method of diagnosis of neglect not stated.

Kasten 2007

Tubinger automated perimetry (TAP)

High resolution perimetry (HRP)

TAP 90° (number of blind stimuli positions)

Group 1

Right eye ‐ mean 46.6, SD 6.9, left eye ‐ mean 43.9, SD 3.7

Group 2

Right eye ‐ mean 50.3, SD 8.7, left eye ‐ mean 43.1, SD 7.6

Group 3

Right eye ‐ mean 32.9, SD 6.8, left eye ‐ mean 37.9, SD 7.1

Not stated

No: participants with neglect were excluded. Method of diagnosis of neglect not stated

.Keller 2010

Goldmann perimetry

Goldmann suprathreshold

Group 1

4 left hemianopia

3 right hemianopia

1 UL quandrantanopia

1 LL quandrantanopia

1 UR quandrantanopia

Group 2

3 left hemianopia

3 right hemianopia

3 UL quandrantanopia

1 LL quandrantanopia

Group 1

6 with 0° macular sparing

4 with < 5° macular sparing

Group 2

6 with 0° macular sparing

4 with < 5° macular sparing

No: participants with neglect were excluded. 3 neglect tests were used: "line bisection, Mesulam test, draw a clock face test".

Modden 2012

Visual field assessment from the Test Battery of Attentional Performance

RT Group

10 hemianopia

5 quadrantanopia

TAP alertness without cueing, ms; mean 304.2, SD 80.8

TAP conjunction search, omissions; mean 9.1, SD 9.0

CT Group

12 hemianopia

3 quadrantanopia

TAP alertness without cueing, ms; mean 383.7, SD 205.2

TAP conjunction search, omissions; mean 10.7, SD 6.7

OT Group

10 hemianopia

5 quadrantanopia

TAP alertness without cueing, ms; mean 308.1, SD 58.6

TAP conjunction search, omissions; mean 10.3, SD 5.6

RT Group

3/15 participants with less than 2° sparing

CT Group

3/15 participants with less than 2° sparing plus 1 participant with no sparing

OT Group

3/15 participants with less than 2° sparing

No: participants with neglect were excluded. Method of diagnosis of neglect not stated.

Plow 2010

Subjective topographic measure of perceived visual field defect

High resolution perimetry (HRP)

7 hemianopia

5 quadrantanopia

Not stated

Not stated

Poggel 2004

Tubinger automated perimetry (TAP)

High resolution campimetry

High resolution perimetry (HRP)

Group 1

Upper attention field (size of area of residual vision, %), mean 18.2, SEM 4.0

Lower probe field (size of area of residual vision, %), mean 21.3, SEM 3.1

Total visual field (size of area of residual vision, %), mean 7.3, SEM 1.9

Group 2

Upper attention field (size of area of residual vision, %), mean 16.9, SEM 2.4

Lower probe field (size of area of residual vision, %), mean 15.5, SEM 4.0

Total visual field (size of area of residual vision, %), mean 6.7, SEM 1.3

Not stated

No: participants with neglect were excluded. Method of diagnosis of neglect not stated.

Rossi 1990

Harrington Flocks Visual Screener

Tangent screen measures

Group 1

Homonymous hemianopia 12

(Visual neglect 6)

Group 2

Homonymous hemianopia 15

(Visual neglect 6)

Not stated

Yes: participants with "homonymous hemianopia or visual neglect were recruited ....". Method of diagnosis of neglect was Harrington Flocks Visual Screener.

39 participants recruited: 27 had homonymous hemianopia; 12 had visual neglect.

Roth 2009

Tubinger automated perimetry (TAP)

Scanning laser ophthalmoscopy

Group 1

Homonymous hemianopia 12, quadrantanopia 3

Group 2

Homonymous hemianopia 12, quadrantanopia 3

Not stated

No: participants with neglect were excluded. Method of diagnosis of neglect was clock‐drawing and line‐bisection tests.

Rowe 2010

Goldmann perimetry

Esterman measures of static superthreshold

Group 1

Homonymous hemianopia left partial 8, Homonymous hemianopia right partial 3, Homonymous hemianopia left complete 9, Homonymous hemianopia right complete 6

Group 2

Homonymous hemianopia left partial 5, Homonymous hemianopia right partial 9, Homonymous hemianopia left complete 8, Homonymous hemianopia right complete 8

Group 3

Homonymous hemianopia left partial 8, Homonymous hemianopia right partial 5, Homonymous hemianopia left complete 10, Homonymous hemianopia right complete 6

Not stated

No: participants with neglect were excluded. Method of diagnosis was clinical assessment: "as assessed by the orthoptist".

Schuett 2012

Tubingen kinetic perimetry

Group 1

Hemianopia 15, quadranopia 1, paracentral scotoma 2

Group 2

Hemianopia 10, quadranopia 4, paracentral scotoma 4

Group 1

Mean 2.3° (SD 1.4)

Group 2

Mean 2.3° (SD 1.2)

No: participants with neglect were excluded. Method of diagnosis described as: "as assessed by tests in accordance with the Behavioural Inattention Test (line bisection, letter and star cancellation, figure and shape copying, drawing from memory; Halligan et al, 1991)."

Spitzyna 2007

Goldmann perimetry

Humphrey automated perimetry

Group 1

Full homonymous hemianopia 8, partial homonymous hemianopia 1, lower quadrantanopia 1, upper quadrantanopia 1

Group 2

Full homonymous hemianopia 6, lower quadrantanopia 1, upper quadrantanopia 1

Macular sparing defined as 2° of sparing

Group 1

Sparing 5, non‐sparing 6

Group 2

Sparing 3, non sparing 5

No: only participants with right‐sided homonymous hemianopic were included; therefore, presence of neglect was assumed unlikely.

Szlyk 2005

Goldmann perimetry

Group 1

Goldmann III4e, range 45.2 to 125, mean 59.12*, SD 22.07*

Goldmann V4e, range 48.8 to 115, mean 70.56*, SD 26.15*

Group 2

Goldmann III4e, range 46.8 to 123.8, mean 68.0*, SD 31.71*

Goldmann V4e, range 50.67 to 132, mean 73.73*, SD 33.10*

Figures were calculated for the affected side only.

Not stated

Not stated; however, although it was not stated whether the participants may have had visual neglect, neglect is unlikely in occipital lesions, and only participants with occipital lesions were included.

* Figures calculated from raw data supplied in papers
combined: combined etiology
HRP: high resolution perimetry
LL: lower left
LR: lower right
ms: milliseconds
SD: standard deviation
SEM: standard error of the mean
TAP: Tübingen automated perimeter
UL: upper left
UR: upper right

Open in table viewer
Table 4. Outcome measures within included studies

Study

Functional ability in ADL

Visual field

Outcome category (measure)

Functional ability in EADL

Reading

Falls

Quality of life

Visual scanning

Adverse events

Other

Outcomes with data included within meta‐analyses

Aimola 2011

Kinetic Perimetry (unspecified kinetic perimeter)

Static Superthreshold (Esterman measures of static superthreshold)

(NB not clear if recorded as outcome or not; no results provided for visual field data)

Reading (corrected reading speed)

1. VFQ 25
2. VIQ ‐ Visual Impairments questionnaire
3. Subjective Reasons questionnaire

1. visual search ‐ find the number (computer ‐based)
2. visuomotor search ‐ find items on a shelf

Tasks simulating ADL ‐ 1. driving hazard perception (mean score per hazard), 2. obstacle avoidance (completion time), 3. visuomotor search (time)

Attention tasks ‐ 1. sustained attention to response (mean percentage error score), 2. test of everyday attention

Reading: Analysis 2.3

Visual search: time to complete Analysis 2.5

QoL: data not included as only available for individual questionnaire items

Bainbridge 1994

Gross visual screening (Harrington‐Flocks Visual Field Score)

Line Cancellation Test

Motor Free Visual Perception Score

Line Bisection Test

No data included in meta‐analyses (as no control group). See Table 5

Bowers 2014

Mobility questionnaire

Question: "If the study were to end today, would you want to continue with these prism glasses (i.e. the prism glasses worn in that period)?"

Functional ability in EADL: Analysis 3.3

De Haan 2015

Kinetic perimetry (Goldmann Perimetry, Functional Field score)

Independent Mobility questionnaire

1. Radner reading test; (a) Radner average reading speed (wpm), (b) minimal readable text size (LogRad)

2. Text reading test; (a) text reading speed (wpm), (b) text correct answers

1. NEI‐VFQ‐25 (Visual Functional questionnaire)
2. Cerebral Visual Disorders questionnaire

1. visual scanning ‐ dots test
2. visual search ‐ letters (parallel search test)
3. visual search ‐ letters (serial search)

Not reported as an outcome measure, but stated no adverse events in either group

Visual acuity, contrast sensitivity, hazard perception, simulating driving/tracking task, obstacle course

Visual field: Functional Field score Analysis 2.1

Reading: Radner average reading speed Analysis 2.3

Visual scanning: Parallel search test, time Analysis 2.5

QoL: NEI VFQ Analysis 2.4

Functional ability in EADL: mobility questionnaire Analysis 2.2

Elshout 2016

Goldman perimetry

Humphrey perimetry

Reading speed (words per minute) ‐ 15 point Arial font, 88 and 165 words

No data included (as data not available for before the cross‐over)

Gall 2013

Static Threshold Perimetry (Standard automated perimetry)

1. NEI VFQ 39 (vision‐related)
2. SF‐12 (health‐related)

No data included in meta‐analyses (as no suitable data presented in abstract)

Jarvis 2012

FIM

1. Functional mobility (timed walk)

2. Non‐validated questionnaire giving qualitative information
about their treatment approach

Functional ability in ADL: Functional Independence Measure Analysis 4.1

Jobke 2009

Static Threshold Perimetry (Standard automated perimetry)

Resolution Perimetry (High resolution perimetry)

Radner reading test

NEI VFQ

Zahlen‐Verbindungs
test (ZVT) for measuring the speed of connecting numbers in
a paper‐pencil test

No data included in meta‐analyses (as no control group). See Table 5 (for data available before the cross‐over)

Kasten 1998

Resolution Perimetry (High resolution perimetry)

Static Threshold Perimetry (Tubinger automated perimetry)

Quality of life questionnaire

Visual acuity: Landolt ring to give minimum angle of resolution

Visual field: Tubinger automated perimetry: border position in degrees of visual angle from zero vertical meridian Analysis 1.1

Quality of life Analysis 1.2

Kasten 2007

Resolution Perimetry (High resolution perimetry: number of hits, learning effects, fixation ability, false hits)

Static Threshold Perimetry (Tubinger automated perimetry: no of hits, fixation ability)

Subjective visual ability questionnaire

1. Eye movements: "Chronos Vision Eye Tracker"

2. Visual acuity

3. "Zahlen‐Verbindungs Test" of visuo‐spatial attention

4. "Alters‐Konzentrationstest" attention test for older people

5. "testbatterie zur Aufmerksamkeitspruefung" ability to improve attention

No data included in meta‐analyses (as no control group). See Table 5 (for available data comparing group outcomes)

Keller 2010

Kinetic Perimetry (Goldmann perimetry)

Static Superthreshold (Goldmann suprathreshold)

Reading time (standardised reading test)

OT administered questionnaire (based on Kerkhoff's self‐evaluation of ADL)

1. Visual exploration test (number of omissions)

2. Search task (search time)

Electro‐oculography

No data included in meta‐analyses (as no control group). See Table 5

Modden 2012

Extended Barthel Index (German)

Gross Visual Screening (Test Battery of Attentional Performance:

visual field assessment)

Reading ‐ Weschler memory tests (errors)

1. Visual scan: from the Test Battery of Attentional Performance
2. Visual search: cancellation tasks from Behavioural Inattention Test (BIT) (omissions)

Attention: Test Battery of Attentional Performance (alertness)

No data included in meta‐analyses (as no control group). See Table 5

Plow 2010

Resolution Perimetry (High resolution perimetry: position of visual field border and stimulus detection accuracy)

Gross Visual Screening (subjective topographic
measure of perceived visual field deficit)

Impact of Vision Impairment (IVI) profile

Low Vision‐
Visual Functional Questionnaire (LV‐VFQ)

Measure of fixation performance

No data included in meta‐analyses (as no control group). See Table 5

Poggel 2004

Static Threshold Perimetry (Tubinger automated perimetry (TAP))

Static Superthreshold (High resolution campimetry)

Resolution Perimetry (High resolution perimetry (HRP))

No data included in meta‐analyses (as no control group). See Table 5

Rossi 1990

Barthel Index

Gross Visual Screening (Harrington Flocks Visual Screener)

Static Superthreshold (Tangent screen examination)

Number of falls

Line cancellation task

Modified Mini Mental Status Examination,

Motor Free Visual Perceptual Test,

Line Bisection Task

ADL: Barthel Index: Analysis 3.1

Visual Field: Analysis 3.2

Falls: number of falls Analysis 3.5

Visual scanning: cancellation Analysis 3.7

Roth 2009

Static Threshold Perimetry (Tubinger automated perimetry (TAP))

Resolution Perimetry (Scanning laser ophthalmoscopy)

Reading speed

QoL: World Health Organisation questionnaire WHOQOL‐BREF

1. Digit search task (response time)

2. Natural search task (table test)(response time)

3. Natural scene exploration

4. Fixation stability (video eye tracker)

No data included in meta‐analyses (as no control group). See Table 5

Rowe 2010

Kinetic Perimetry (Goldmann perimetry)

Static Superthreshold (Esterman measures of static superthreshold)

Nottingham extended activities of daily living (NEADL)

Reading ability (Radner test)

1. VFQ 25‐10 (vision related)
2. EQ‐5D
3. SF‐12

Number of participants and number of adverse events

Rivermead Mobility Index

Visual Field: relative change in visual field area Analysis 2.1 and Analysis 3.2

QoL: VFQ 25‐10 Analysis 2.4 and Analysis 3.6

Adverse events:Analysis 2.6; Analysis 3.8

Schuett 2012

Kinetic Perimetry (Tubingen kinetic perimetry) (NB not recorded immediately after first phase)

Reading (speed and errors)

Cancellation (speed and errors)

No data included in meta‐analyses (as no control group). See Table 5

Spitzyna 2007

Kinetic Perimetry (Goldmann perimetry)

Static Threshold Perimetry (Humphrey 10‐2 central threshold programme)

(NB not recorded immediately after first phase)

1. Text reading speeds

2. Single word reading speeds

Eye movement characteristics:

‐ spatial characteristics of saccadic amplitude, incoming saccade amplitude and landing position

‐ temporal characteristics

Reading (text reading speed): Analysis 2.3

Visual field ‐data not included as not collected before cross‐over.

Szlyk 2005

Kinetic Perimetry (Goldman Perimetry)

Indoor functional assessment

Outdoor functional assessment

Driving skills assessment

Psychophysical assessment

Satisfaction

Prisms use at 2 years

No data included in meta‐analyses(as no control group).

No data reported for before the cross‐over.

ADL: activities of daily living
BIT: behavioural inattention test
EADL: extended activities of daily living
EQ‐5D:standardised EuroQol health‐related quality of life instrument
FIM: Functional Independence Measure
HRP: high resolution perimetry
IVI: impact of vision impairment
LogRad: a scale of reading acuity
LV‐VFQ: low vision functional questionnaire
NB: note
NEADL: Nottingham Extended Activities of Daily Living
NEI‐VFQ‐25: National Eye Institution Visual Function Questionnaire
OT: occupational therapy
QoL: quality of life
SF‐12: 12‐Item Short Form Health Survey
TAP: tubinger automated perimetry
VFQ: visual function questionnaire
VIQ: visual impairment questionnaire
WHOQOL‐BREF: World Health Organisation quality of life questionnaire
wpm: words per minute
ZVT: Zahlen‐Verbindungs test

Study design

Fifteen of the included studies were parallel‐group randomised controlled trials (RCTs), and five were randomised cross‐over studies (Bowers 2014; Elshout 2016; Jobke 2009; Schuett 2012; Szlyk 2005).

Sixteen of the included studies randomised participants to one of two treatment groups; three had three treatment groups (Kasten 2007; Modden 2012; Rowe 2010); and one was a cross‐over AB/BA design, where A was an active treatment and B a placebo; however participants were also randomised to receive one of two different active treatments, each of which had a related sham treatment meaning that there were effectively four different treatment groups (active 1, active 2, sham 1 and sham 2) (Bowers 2014).

Comparison versus control

Ten of the 20 included studies had a control (no treatment, standard care, or placebo) group, comparing 10 active treatments with control (Rowe 2010 had two active treatment groups):

In Rowe 2010, the two active treatment groups (compensatory and substitutive interventions) were compared with each other.

Ten of the 20 included studies did not have a control group. Nine compared two different active treatments, and one had three active treatment groups (Modden 2012):

Interventions studied
Restitutive interventions

Nine studies (239 randomised participants) investigated the effect of restitutive interventions.

In eight of these studies, the restitutive intervention studied was a form of computer‐based vision restoration therapy:

  • Kasten 1998 and Elshout 2016 compared visual restitution therapy with a placebo intervention;

  • Jobke 2009 and Kasten 2007 compared the effectiveness of two (or more) types of visual restitution therapy;

  • Plow 2010 explored the effect of adding transcranial direct current stimulation (tDCS) to visual restitution therapy;

  • Poggel 2004 explored the effect of adding attentional cueing to visual restitution therapy;

  • Modden 2012 compared computerised restitution therapy with two different compensatory interventions;

  • Roth 2009 compared 'flicker‐stimulation training', which the authors described as a "potential" restitutive intervention, with a compensatory intervention.

In one of these nine included studies, the restitutive intervention studied was a form of non‐invasive brain stimulation using alternating current stimulation. This was compared with a placebo intervention (Gall 2013).

Compensatory interventions

Eight studies (347 randomised participants) investigated the effect of compensatory interventions:

  • Aimola 2011 compared computer‐based compensatory training with a control;

  • Spitzyna 2007 compared computer‐based reading training ("optokinetic nystagmus inducing reading therapy", involving reading scrolling right to left text) with a control;

  • De Haan 2015 compared a compensatory scanning training programme with a control;

  • Rowe 2010 compared paper‐based visual scanning training with a control (and with a substitutive intervention);

  • Keller 2010 compared two types of compensatory training ‐ audiovisual exploratory training and visual exploration training;

  • Schuett 2012 compared two types of compensatory training ‐ visual exploration training and reading training;

  • Modden 2012 compared computerised scanning training, an occupational therapy compensatory training program, and a computerised restitutive therapy;

  • Roth 2009 compared computer‐based scanning training with restitutive training.

Substitutive interventions

Five studies (227 randomised participants) investigated the effect of substitutive interventions. In all five studies, the substitutive intervention studied was a type of prism:

  • Rossi 1990 compared 15 diopter Fresnel prisms with no prisms;

  • Rowe 2010 compared 40 diopter Fresnel prisms with no treatment, and with a compensatory intervention;

  • Bowers 2014 compared 57 diopter oblique prism glasses and horizontal prism glasses with five diopter sham prism glasses;

  • Bainbridge 1994 compared full‐field 15 diopter Fresnel prisms with hemi‐field 15 diopter Fresnel prisms;

  • Szlyk 2005 compared 20 diopter Fresnel prisms with 18.5 dioptre Gottlieb VFAS (Visual Field Awareness System) prisms.

Assessment and screening interventions

One study (64 randomised participants) investigated the effectiveness of an assessment and screening intervention on relevant outcomes:

  • Jarvis 2012 compared the effect of providing therapy staff with information from an orthoptic assessment with no intervention.

Populations studied

The reported diagnoses of the participants within the 20 included studies were as follows:

  • 14 studies recruited participants with mixed diagnoses including stroke, trauma, surgery, and infections;

  • five studies included participants with stroke only; and

  • one study ‐ the cause of the lesion was unclear.

(See details in Characteristics of included studies and Table 2). Despite the high number of studies including participants with mixed diagnoses, the majority of participants in the studies in this review did have stroke (across all studies, 85% (520/611) of participants with data reported had a diagnosis of stroke).

Fourteen of the 20 included studies included participants with visual field defects only (no visual neglect); seven of 14 studies clearly stated the method of diagnosis of visual neglect (Bowers 2014; De Haan 2015; Elshout 2016; Keller 2010; Roth 2009; Rowe 2010; Schuett 2012), four of the 14 studies stated that participants with neglect were excluded but did not state the method of diagnosis of visual neglect (Kasten 1998; Kasten 2007; Modden 2012; Poggel 2004), and in the three remaining studies it was assumed (but not clearly stated) that participants with visual field defects only were included (Jobke 2009; Spitzyna 2007; Szlyk 2005). Four of the included studies included participants who had visual neglect in addition to, or instead of, visual field defects (Aimola 2011; Bainbridge 1994; Jarvis 2012; Rossi 1990). In two studies, it was unclear whether the participants had visual neglect or not (Gall 2013; Plow 2010). See Table 3.

Visual field measurement

All 20 included studies reported a measurement of the visual field in order to inform participant inclusion or provide baseline information relating to visual field defect, or both. Seven reported one visual field measure (De Haan 2015; Gall 2013; Jarvis 2012; Modden 2012; Plow 2010; Schuett 2012; Szlyk 2005), 11 reported two visual field measures (Aimola 2011; Bainbridge 1994; Bowers 2014; Elshout 2016; Jobke 2009; Kasten 1998; Keller 2010; Rossi 1990; Roth 2009; Rowe 2010; Spitzyna 2007), and two reported three visual field measures (Kasten 2007; Poggel 2004). Spitzyna 2007 reported a second measure only where the first perimetric results had poor reliability.

In three studies, the perimetry equipment was unclear (Aimola 2011; Gall 2013; Jobke 2009), and in Rowe 2010, either of two types of perimetry was reported (Esterman static programme or Goldmann kinetic).

Visual field measurements were categorised as:

Sample size

On average, included studies randomised 36 participants (standard deviation, 22 participants) into the trial prior to attrition. This ranged from just 10 participants (Szlyk 2005), to 87 participants (Rowe 2010). Only five of 19 studies recruited more than 50 participants: Aimola 2011 (n = 70), Bowers 2014 (n = 73), De Haan 2015 (n = 54), Jarvis 2012 (n = 64), and Rowe 2010 (n = 87). A total of 732 participants were recruited across the 20 included studies, with data available for 638 participants, of whom 547 were stroke patients. See Table 2 for recruitment numbers across all included studies.

Outcome measures

Table 4 summarises the outcome measures within the included studies, and highlights which studies had data which was suitable for inclusion in meta‐analyses within this review.

Primary outcome

  • Functional ability in activities of daily living. Three studies included a measure of functional ability: Rossi 1990 reported the Barthel Index, Jarvis 2012 reported the FIM, and Modden 2012 reported the extended Barthel Index (German version).

Secondary outcomes

Excluded studies

We excluded 158 papers after assessment of the full paper (see Figure 1). Sixty‐two of the 158 clearly did not meet the inclusion criteria, and 30 of the 158 were duplicate publications of excluded studies. Fifty‐seven of the 158 papers required more in‐depth appraisal prior to exclusion; we have provided our reasons for exclusion of these studies in the Characteristics of excluded studies table. We excluded the majority of these because the intervention was not specifically targeted at the ability of the participant to cope with visual field loss (32/57), or because the study was not randomised (22/57). We found one study did not include participants with stroke, one was focused on central alexia, and one was exploring agreement using a visual screening tool.

We included three studies in the previous (2011) version of this review (Carter 1983, Weinberg 1977and Weinberg 1979), but they have been excluded in this update: in the 2011 version, we included studies that investigated the effectiveness of visual scanning training and techniques even if the population of participants had not been clearly defined as having visual field defects. Carter 1983, Weinberg 1977and Weinberg 1979 included populations of participants with 'visual scanning' problems, who may have had either visual field defects or visual neglect, or both. For this latest update, we reconsidered and reversed this decision: we have excluded populations of participants with 'visual scanning problems', but no confirmed visual field defect, to ensure that all included studies are focused on stroke survivors with confirmed visual field defects.

Studies awaiting classification

Two studies are awaiting classification (Ghandehari 2011; Sand 2017). Ghandehari 2011 compared two different pharmacological interventions (Neuroaid and Piracetam) without a control group. The selection criteria, which were prestated for this review, did not clearly state whether pharmacological interventions were relevant for inclusion, but discussion amongst review authors supports the conclusion that it would be appropriate to include any trials of pharmacological interventions if they are specifically focused on improving outcomes in participants with visual field defects. However, it remains unclear whether this study is a randomised controlled trial or not, as there is inconsistent reporting of the study design, and we await clarification from the study authors. Prior to any future updates of this review, we will clarify the selection criteria and methods to ensure that the inclusion (or exclusion) of pharmacological interventions is addressed. Sand 2017 is an ongoing study; however, it was not possible to determine from available information whether this was a randomised controlled trial or not.

Ongoing studies

Seven studies are listed as ongoing studies. Two ongoing studies compare a form of visual restitutive training with control (Feldon 2017; NCT02886663); one is comparing a form of transcranial electrical stimulation with control (Gall 2015); one is comparing a pharmacological intervention focused on restitution of the visual field with a placebo (NCT02737930); and one is comparing two different modes of delivery of computer‐based compensatory scanning interventions with a control (ISRCTN16023965). Two of these studies, listed as ongoing in the previous version of this review, are now complete (Hayes 2010; Komm 2009), but we have been unable to obtain results from the authors. If results are not available at the time of the next update, we will exclude these studies.

Risk of bias in included studies

We have described the assessment of risk of bias for individual studies in the 'Risk of bias' tables in Characteristics of included studies, summarised in Figure 2.

Figure 2
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Although all 20 included studies were described as randomised controlled trials, only 10 of them reported sufficient information on the method of randomisation to assess whether the randomisation and allocation concealment methods were adequate, and, in two of these studies, we judged them to be at high risk of bias: in Keller 2010 some of the allocation appeared to have been alternate, rather than random, and, in Modden 2012, randomisation was through throwing of a dice, with no allocation concealment.

Blinding

In the majority of included studies, the nature of the intervention meant that it was not possible to mask (blind) participants or people involved in their care. We judged blinding as low risk of bias if there was evidence that the outcome assessor was masked to the treatment allocation of the participants ‐ six of the 20 studies clearly reported having a masked outcome assessor. In eight of the 20 studies this information was unclear, and in six of the 20 studies we judged that there was a high risk of bias as the outcome assessor was not masked.

Incomplete outcome data

Lack of information about the management of incomplete outcome data, and whether or not there had been dropouts or participants excluded from the studies, made it difficult to assess risk of bias relating to incomplete outcome data for all the studies. In four of the 20 studies, we considered that adequate reasons for dropouts were provided, numbers of dropouts were even between groups, or intention‐to‐treat analysis was carried out. However, in five of the 20 studies where we felt that incomplete outcome data were not adequately addressed, and that dropouts were either unbalanced or might be related to the intervention studied (e.g. because people dropped out or were excluded because of low compliance), we judged risk of bias to be high. For the remaining 11 studies, we were unclear as to whether or not incomplete outcome data were adequately addressed.

Other potential sources of bias

Lack of information and details of methodology within the included studies generally made it difficult for us to assess the potential risk due to other biases. However, four of the studies of visual restitution therapy were carried out by researchers who have acknowledged a financial interest in this intervention (as stated in Poggel 2004) (Jobke 2009; Kasten 1998; Kasten 2007; Poggel 2004), and Bowers 2014 declared a financial conflict of interest relating to the prisms that were being investigated. We assessed this to potentially introduce a source of bias. We identified concerns that the increased attention given to the treatment group, as opposed to the control group, in De Haan 2015 may create a high risk of bias. For two of the studies, only abstracts were available and there was insufficient information on which to assess bias (Bainbridge 1994; Gall 2013). We judged the remaining 12 studies as unlikely to be at risk of other potential sources of bias, but based this assessment on the absence of information suggesting bias rather than the presence of information indicating that the study was free from bias.

Studies included in meta‐analyses within this review

From the 20 studies included in this review, there are only eight studies included in meta‐analyses within this review (Aimola 2011; Bowers 2014; De Haan 2015; Jarvis 2012; Kasten 1998; Rossi 1990; Rowe 2010; Spitzyna 2007). These eight studies randomised 428 participants. The studies relevant to our prestated comparisons (see Objectives), and included within the planned meta‐analyses are:

1. Restitutive interventions versus no treatment, placebo, or control

  • Kasten 1998 (visual restitution therapy versus placebo, n = 19). We have included data from this study in relevant analyses.

2. Compensatory interventions versus no treatment, placebo, or control

3. Substitutive interventions versus no treatment, placebo, or control

  • Rossi 1990 (prisms versus no treatment, n = 39) and Rowe 2010 (prisms versus control, n = 59). We have included data from these studies in relevant analyses.

  • Bowers 2014 (prisms versus placebo, n = 67). This is a cross‐over study and data were principally presented after the cross‐over; however, there were data for one outcome (extended activities of daily living) presented before the cross‐over, with data suitable for inclusion in analysis.

4. Assessment or screening versus no treatment, placebo, or control

  • Jarvis 2012 (assessment versus control, n = 39). We have included data from this study in relevant analyses.

Studies not included in meta‐analyses

Ten of the studies included in this review did not have a control group and are therefore not included in meta‐analyses.

These 10 studies each compared one active intervention with another active intervention:

The results from relevant outcomes from these studies are summarised in Table 5 and Table 6 and in a narrative synthesis below.

Open in table viewer
Table 5. Results of studies comparing two similar active interventions (i.e. two interventions from the same category)

Study

Interventions

Outcome

Mean (or other reported result if no mean available)

Standard deviation

Number of participants

Statistical test/results

Restitution: one restitution intervention versus another restitution intervention

Jobke 2009

Extrastriate VRT

Visual field (high‐resolution perimetry, HRP)

increase from baseline of 5.9% (percentage of HRP hits)

8

significant increase: t = ‐5.262, P = 0.0005

Standard VRT

increase from baseline of 2.9% (percentage of HRP hits)

10

significant increase: t = ‐2.373, P = 0.021

Kasten 2007

Parallel co‐stimulation

Visual field (high‐resolution perimetry)

increase of 2.4% detected stimuli

7

No significant difference "confirmed by nonparametric Kruskal–Wallis ANOVA"

Moving co‐stimulation

increase of 6.5% detected stimuli

7

Single stimulus

increase of 3.9% detected stimuli

9

Plow 2010

VRT + tDCS

Visual field (high‐resolution perimetry)

shift from baseline to post‐test from
4.11° ± 1.50° to 8.37° ± 2.29°, Wilcoxon signed‐rank test = 0, P = 0.068

4

Mann‐Whitney U = 0, P = 0.021 (significantly greater shift in the visual field border with VRT + tDCS than VRT alone)

VRT + sham tDCS

shift from baseline to post‐test from 6.33° ± 2.59° to 7.03° ± 2.51°, Wilcoxon signed‐rank test = 1, P = 0.144

4

VRT + tDCS

Functional ability in ADL (LV‐VFQ)

shift from baseline to post‐test from 32.25
± 5.30 to 28.25 ± 5.07; Wilcoxon signed‐rank test = 0; P = 0.068

4

Mann‐Whitney U = 5.5; P = 0.468 (non‐significant)

VRT + sham tDCS

shift from baseline to post‐test from 28 ± 2.34 to 25.25 ± 1.11; Wilcoxon signed‐rank test = 1; P = 0.285

4

VRT + tDCS

Functional ability in ADL (LV‐VFQ) ‐ 6‐month follow‐up

29.00

3.58

5

Wilcoxon signed‐rank test = 4; P = 0.343

VRT + sham tDCS

26.80

2.11

VRT + tDCS

Quality of life ‐ 6‐month follow‐up

23.20

7.83

5

Wilcoxon
signed‐rank test = 2.5; P = 0.357

VRT + sham tDCS

16.8

4.62

Poggel 2004

VRT + attentional cueing

Visual field (high‐resolution perimetry) ‐ percentage improvement, attention field

8.3

SEM 1.5

9

P = 0.001 (in favour of attentional cueing)

VRT with no attentional cueing

2.9

SEM 0.8

10

Compensation: one compensation intervention versus another compensation intervention

Schuett 2012

Visual exploration training

Reading speed

105.3

33.8

18

Not reported; calculated as MD ‐19.30 (‐43.32 to 4.72) (see Figure 3)

Reading training

124.6

39.5

18

Visual exploration training

Cancellation test (exploration time)

18.5

4.9

18

Not reported; calculated as MD 18.30 (14.28 to 22.32) (see Figure 3)

Reading training

36.8

7.2

18

Keller 2010

Audiovisual exploration training (AVT)

Functional ability in ADL (ADL test total score)

1.5

(SE displayed on graph only)

10

ANOVA P = 0.036 (in favour of AVT)

Visual exploration training

5.0

(SE displayed on graph only)

10

Audiovisual exploration training (AVT)

Reading time (seconds)

75

(SE displayed on graph only)

10

ANOVA P = 0.03 (in favour of AVT)

Visual exploration training

178

(SE displayed on graph only)

10

Audiovisual exploration training (AVT)

Visual scanning (percentage hits)

85.3

(SE displayed on graph only)

10

ANOVA P = 0.01 (in favour of AVT)

Visual exploration training

64.1

(SE displayed on graph only)

10

Modden 2012

Computer‐based compensation therapy

Visual field enlargement (visual field assessment from Test Battery of Attentional Performance)

2.9

4.0

15

Pre‐ to post‐treatment significant field expansion (P = 0.013)

Standard occupational therapy (compensation)

1.3

4.7

15

Pre‐ to post‐treatment: no significant field expansion (P = 0.316)

Computer‐based compensation therapy (CT)

Functional ability in ADL (improvement in Extended Barthel Index)

3.3

3.6

15

"No significant treatment effects were found when comparing ... CT/OT".

Standard occupational therapy (OT) (compensation)

1.8

2.0

15

Computer‐based compensation therapy (CT)

Reading ‐ Improvement in reading performance, reduction in number of errors (from baseline)

‐0.9

1.1

15

"Compared with OT"... "CT did not significantly reduce reading errors."

Standard occupational therapy (OT) (compensation)

‐0.7

1.0

15

Computer‐based compensation therapy (CT)

Visual scanning ‐ reduction in number of omissions from baseline, cancellation tasks of the Test Battery of Attentional Performance

‐5.4

5.2

15

"Compared with OT"... "CT did not result in superior improvements".

Standard occupational therapy (OT) (compensation)

‐2.3

5

15

Substitution: one substitution intervention versus another substitution intervention

Bainbridge 1994

Full‐field Fresnel Prisms

Visual Field (Harrington Flocks Visual Field Score)

2.9

2

10

States full‐field more improved

Hemi‐field Fresnel Prisms

7.2

3

8

Full‐field Fresnel Prisms

Scanning (Line cancellation test errors)

4.7

1.3

10

P < 0.01, Student's t‐test (in favour of full‐field prisms)

Hemi‐field Fresnel Prisms

0.3

0.6

8

Szlyk 2005

18.5 dioptre Gottlieb Visual field awareness system prisms

Visual skills category assessment battery

"There was improvement within all categories with both of the prism systems ranging from 36% for mobility (with the Fresnel prisms) to 13% for recognition (with the Gottlieb VFAS)."

10 (data only available for after cross‐over)

"There were no statistically significant differences between improvements
with the Gottlieb VFAS compared with the
Fresnel prisms."

Press‐on TM 20 Diopter Fresnel prisms

ADL: activities of daily living
ANOVA: analysis of variance (statistical test of)
AVT: audiovisual exploration training
CT: compensation therapy
HRP: high‐resolution perimetry
LV‐VFQ: Low Vision Visual Functioning Questionnaire
MD: mean difference
OT: occupational therapy
SE: standard error
SEM: standard error of the mean
tDCS: transcranial direct current stimulation
VFAS: visual field awareness system
VRT: visual restitution therapy

Open in table viewer
Table 6. Results of studies comparing two different types of active interventions (i.e. interventions from different categories)

Study

Interventions

Outcome

Mean (or other reported result if no mean available)

Standard deviation

Number of participants

Statistical test/results

Compensation intervention versus restitution intervention

Modden 2012

Computer‐based restitution therapy

Visual field enlargement (visual field assessment from Test Battery of Attentional Performance)

3.9

4.9

15

Pre‐ to post‐treatment significant field expansion (P = 0.003)

Computer‐based compensation therapy

2.9

4.0

15

Pre‐ to post‐treatment significant field expansion (P = 0.013)

Computer‐based restitution therapy (RT)

Functional ability in ADL (improvement in Extended Barthel Index)

1.5

2.8

15

"No significant treatment effects were found when comparing ... RT/CT".

Computer‐based compensation therapy (CT)

3.3

3.6

15

Computer‐based restitution therapy (RT)

Reading: improvement in reading performance, reduction in number of errors (from baseline)

‐0.9

2.4

15

"There were no differences between RT and CT."

Computer‐based compensation therapy (CT)

‐0.9

1.1

15

Computer‐based restitution therapy (RT)

Visual scanning: reduction in number of omissions from baseline, cancellation tasks of the Test Battery of Attentional Performance

‐5.3

10.5

15

"... the improvement of the CT compared with the RT group did not meet the defined significance level after Bonferroni correction (P = .023)."

Computer‐based compensation therapy (CT)

‐5.4

5.2

15

Roth 2009

Explorative scanning training (EST) (compensation)

Visual field: Tubingen automated perimetry

44.4

13.1

15

"Neither the EST group nor the FT group showed any differences in their TAP or SLO outcomes, quantified as the total number of stimuli detected in the blind hemifield (lowest P = 0.204)."

Flicker stimulation training (FT)(restitution)

35.7

15.2

13

Explorative scanning training (EST) (compensation)

Quality of life (WHOQOL‐BREF)

12.93

1.67

15

"The EST group reported greater improvements (T2 minus T1 scores) in the WHOQOL social‐relationships domain (t test; t(20) = 2.217, P = 0.038)" (but no significant differences for other domains).

Flicker stimulation training (FT) (restitution)

13.23

1.3

13

Explorative scanning training (EST) (compensation)

Reading (reading speed)

99.7

34.7

15

"Although the EST and FT groups differed in their reading speeds at T1, this difference remained unchanged [main effect of group, F(1,26) = 133.074, P < 0.0001, interaction, F < 1]".

Flicker stimulation training (FT)(restitution)

140.2

20.9

13

Compensation intervention versus substitution intervention

Rowe 2010*

Fresnel prisms (substitution)

Visual Field (relative change in visual field area)

0.052

0.1396

24

ANOVA results: no significant differences between groups (P = 0.55, for comparison across 3 treatment groups)

Visual search training (compensation)

0.0815

0.1488

24

Fresnel prisms (substitution)

Extended activities of daily living (change in EADL from baseline)

15.2

4.8

22

"No evidence of differences ..."

Visual search training (compensation)

15.2

4.4

22

Fresnel prisms (substitution)

Reading (change in Radner reading speed)

17.4

21.3

24

"No evidence of differences ..."

Visual search training (compensation)

13.0

13.1

25

Fresnel prisms (substitution)

Quality of life (VFQ‐25 total score)

68.2

18.4

24

"Visual function (using the VFQ 25‐10) improved at 26 weeks in the visual search training arm (60 [SD 19] to 68.4 [SD 20]) when compared to the Fresnel prisms (68.5 [SD 16.4] to 68.2 [18.4]) and standard care arms (63.7 [SD 19.4] to 59.8 [SD 22.7]: Table 6, ANCOVA P = 0.05)."

Visual search training (compensation)

68.4

20.0

25

Fresnel prisms (substitution)

Adverse events (number of participants with reported adverse events during study)

18 participants

26

"Given the extent and range of adverse events reported with prism wear, caution must be exercised if prescribing prism glasses as an intervention for homonymous hemianopia."

Visual search training (compensation)

2 participants

30

*Rowe 2010 also had a control (standard care) group, and data were included in relevant meta‐analyses for compensatory and substitution interventions versus control.
ADL: activities of daily living
ANCOVA: analysis of covariance (statistical test of)
ANOVA: analysis of variance (statistical test of)
CT: compensation therapy
EADL: extended activities of daily living
EST: explorative scanning training
FT: flicker stimulation training
RT: restitution therapy
SD: standard deviation
SLO: Scanning Laser Ophthalmoscope
T1: outcome asssessment timepoint 1
T2: outcome assessment timepoint 2
TAP: Tuebingen automated perimetry
VFQ: visual function questionnaire
WHOQOL‐BREF: World Health Organization Quality of Life Instrument

One study did have a relevant control group (alternating current stimulation versus placebo, n = 39) but was published as an abstract only and we have been unable to identify data suitable for inclusion (Gall 2013).

One study did have a relevant control group (visual restitution training versus placebo, n = 30) but was a randomised cross‐over trial and we have been unable to obtain data from the first phase only, in order to include data in meta‐analyses (Elshout 2016).

Effects of interventions

See: Summary of findings 1 Summary of findings: Restitutive interventions versus control; Summary of findings 2 Summary of findings: Compensative interventions versus control; Summary of findings 3 Summary of findings: Substitutive interventions versus control; Summary of findings 4 Summary of findings: Assessment/screening interventions versus control

Restitutive interventions versus no treatment, placebo, or control

There was one included study: Kasten 1998 (see summary of findings Table 1). Data were available for visual field and quality of life outcomes, but not for any other outcomes of interest to this review. Stroke survivors with visual neglect were not included in Kasten 1998, therefore these analyses relate to participants with visual field defects only (no co‐existing visual neglect).

Visual field

See Analysis 1.1. Data from Kasten 1998 (19 participants) showed that there was no statistically significant effect of a restitutive intervention as compared to control (MD 1.02, 95% CI ‐1.37 to 3.41) for the visual field outcome (confrontation). We judged this evidence to be of very low quality.

Quality of life

See Analysis 1.2. Data from Kasten 1998 showed that there was a statistically significant effect of a restitutive intervention as compared to control (OR 13.00, 95% CI 2.07 to 81.48). The data used in this analysis were derived from 30 randomised participants, and included data from participants with optic nerve injury who had also received the same interventions in a separate (but parallel) trial. A total of 38 participants were randomised, of whom 19 had stroke and 19 had optic nerve injury; data were only available for 30 of 38 of these participants for this outcome. Separate data were not available for participants with stroke only. We judged this evidence to be of very low quality.

Compensatory interventions versus no treatment, placebo, or control

Included studies: Aimola 2011; De Haan 2015; Rowe 2010; Spitzyna 2007 (see summary of findings Table 2). Stroke survivors with visual neglect were not included in De Haan 2015; Rowe 2010; Spitzyna 2007, while participants in Aimola 2011 could have co‐existing neglect: data are presented in subgroups relating to the inclusion of participants with neglect.

Visual field

See Analysis 2.1. Data from two studies (95 participants) showed that there was no statistically significant effect of compensatory interventions compared with control (SMD ‐0.11, 95% CI ‐0.92 to 0.70, heterogeneity: I² = 75%) (De Haan 2015; Rowe 2010). We judged this evidence to be of very low quality. Several factors could contribute to the substantial heterogeneity, including different inclusion criteria (e.g. participants had to be < 26 weeks post stroke in Rowe 2010) and very different interventions (see Characteristics of included studies). However a key factor to note, which limits confidence in these findings, is that there was a significant difference in baseline assessment between groups in De Haan 2015).

Extended activities of daily living

See Analysis 2.2. Data from two studies (97 participants) showed that there was no statistically significant benefit in favour of compensatory interventions compared with control (SMD 0.49, 95% CI ‐0.01 to 0.99, heterogeneity: I² = 25%) (De Haan 2015, Rowe 2010). We judged this evidence be of very low quality.

Reading

See Analysis 2.3. Data from four studies (162 participants) showed that there was no statistical significant effect of compensatory interventions compared with control (SMD 0.26, 95% CI ‐0.05 to 0.58, heterogeneity: I² = 0%) (Aimola 2011; De Haan 2015; Rowe 2010; Spitzyna 2007). We judged this evidence to be of low quality. Although some of the participants in Aimola 2011 may have had neglect, there was no downgrade for indirectness as the test for subgroup differences demonstrated no significant differences between the studies including or not including participants with neglect (P = 0.43).

Quality of life

See Analysis 2.4. Data from two studies (96 participants) showed that there was a statistically significant effect of compensatory interventions compared with control (MD 9.36, 95% CI 3.10 to 15.62), with minimal heterogeneity (I² = 0%), based on the results for the total score for the VFQ‐25 assessment (De Haan 2015; Rowe 2010). We judged this evidence to be of low quality.

Scanning ‐ cancellation

See Analysis 2.5. Data from two studies (97 participants) showed that there was no statistically significant effect of compensatory interventions compared with control (SMD ‐0.01, 95% CI ‐0.40 to 0.39), with minimal heterogeneity ( I² = 0%) (Aimola 2011; De Haan 2015). We judged this evidence to be of low quality. Although some of the participants in Aimola 2011 may have had neglect, there was no downgrade for indirectness as the test for subgroup differences demonstrated no significant differences between the studies including or not including participants with neglect (P = 0.55). Substituting the De Haan 2015 'parallel search test' data for 'all trials' with other presented data, including the 'serial search test' data and 'dot counting test', or substituting the 'target present' data for the 'all trials' data, did not change the non‐significant result.

Adverse events

See Analysis 2.6. Rowe 2010 collected and reported data relating to adverse events, stating that: "Two patients (6.7%) in the visual search training arm experienced seven adverse events (six fatigue and one headache). No adverse events were recorded for the standard care arm". De Haan 2015 did not report adverse events as an outcome measure, but stated that: "No important harms caused by the training or the assessments were encountered, nor reported by the participants". Data from Rowe 2010 (see Analysis 2.6), showed that there was no difference in the odds of a participant having an adverse event with compensatory scanning training, when compared to control (OR 15.18, 95% CI 0.24 to 112.57). We judged this evidence to be of low quality.

Substitutive interventions versus no treatment, placebo, or control

Included studies: Bowers 2014; Rossi 1990; Rowe 2010 (see summary of findings Table 3). Stroke survivors with visual neglect were not included in Bowers 2014 and Rowe 2010, while participants in Rossi 1990 could have co‐existing neglect. Substitutive prisms were worn during the outcome assessment by participants in Bowers 2014 and Rossi 1990, but were not worn during outcome assessment by participants in Rowe 2010. As the wearing of the substitutive intervention during outcome assessment should theoretically expand the size of the visual field, the data arising from these different approaches (wearing or not wearing prisms) were presented as subgroups, and no pooled total was calculated.

Functional activities of daily living (primary outcome)

See Analysis 3.1. Data from one study (39 participants, wearing prisms during assessment; participants may have co‐existing neglect) showed that there was no statistically significant effect of a substitutive intervention compared with control for the primary outcome of functional activities of daily living (MD ‐4.00, 95% CI ‐17.86 to 9.86) (Rossi 1990). We judged this evidence to be of very low quality.

Visual field

See Analysis 3.2. Data from one study (46 participants; no neglect) showed that there was no statistically significant effect of a substitutive intervention compared with control when the substitutive intervention (prism) was not being worn (SMD 0.12, 95% CI ‐0.46 to 0.70) (Rowe 2010). Data from one study (39 participants; possibly co‐existing neglect) showed that there was a statistically significant effect of a substitutive intervention compared with control when the substitutive intervention (prism) was being worn (SMD 1.12, 95% CI 0.44 to 1.80) (Rossi 1990). There was a statistically significant difference between the subgroups in which participants did and did not wear prisms during assessment (P = 0.03). We judged this evidence to be of very low quality.

Extended activities of daily living.

See Analysis 3.3. Data from one study (48 participants, no neglect) showed that there was no statistically significant effect of a substitutive intervention compared with control when the substitutive intervention (prism) was not being worn (SMD 0.20, 95% CI ‐0.44 to 0.85) (Rowe 2010). Data from one study (61 participants, no neglect) showed that there was no statistically significant effect of a substitutive intervention compared with control for measures of extended activities of daily living, using a mobility score (SMD 0.24, 95% CI ‐0.26 to 0.75) (Bowers 2014). There was no statistically significant difference between the subgroups that did and did not wear prisms during assessment (P = 0.92). We judged this evidence to be of very low quality.

Reading

See Analysis 3.4. Data from one study (45 participants, no neglect) showed that there was no statistically significant effect of a substitutive intervention compared with control when the substitutive intervention (prism) was not being worn (MD 2.80, 95% CI ‐7.13 to 12.73) (Rowe 2010). We judged this evidence to be of low quality.

Falls

See Analysis 3.5. Data from one study (39 participants, possible co‐existing neglect) showed that there was no statistically significant effect of a substitutive intervention compared with control for the risk of falls (OR 1.21, 95% CI 0.26 to 5.76) (Rossi 1990). We judged this evidence to be of very low quality.

Quality of Life

See Analysis 3.6. Data from one study (43 participants, assessed not wearing prisms; no neglect) showed that there was no statistically significant effect of a substitutive intervention compared with control for a measure of quality of life (MD 8.40, 95% CI ‐4.18 to 20.98) (Rowe 2010). We judged this evidence to be of low quality.

Scanning (cancellation)

See Analysis 3.7. Data from one study (39 participants, assessed wearing prisms; possibly co‐existing neglect) showed that there was a statistically significant effect of a substitutive intervention compared with control for measures of scanning (MD 9.80, 95% CI 1.91 to 17.69) (Rossi 1990). We judged this evidence to be of very low quality.

Adverse events

Rowe 2010 collected and reported data relating to adverse events, stating that "Eighteen patients (69.2%) in the Fresnel prisms arm experienced a total of 42 adverse events of which 28 were classified as headache. No adverse events were recorded in the standard care arm". The reported adverse events in the group wearing prisms were: headache (28 events in six participants); diplopia (five events in five participants); visual confusion (four events in three participants); difficulty with navigation (two events in two participants); dizziness (two events in one participant); optical glare/aberrations (one event in one participant). Analysis 3.8 shows that there was an increased odds of a participant having an adverse event if they were wearing prisms (OR 87.32, 95% CI 4.87 to 1564.66). We judged this evidence to be of low quality.

Assessment or screening interventions versus no treatment, placebo, or control

Included study ‐ Jarvis 2012 (see summary of findings Table 4).

Functional activities of daily living (primary outcome)

See Analysis 4.1. Data from Jarvis 2012 (37 participants) showed that there was no statistically significant effect of an assessment intervention compared with control for measures of functional activities of daily living (MD ‐6.97, 95% CI ‐23.78 to 9.84). We judged this evidence to be of very low quality.

Subgroup and sensitivity analyses

As there were no more than six trials in any single meta‐analysis, as prestated in our protocol, we did not carry out the planned subgroup or sensitivity analyses. There were two exceptions to this:

  • we carried out subgroup analysis to explore the effect of including studies in which participants may have had neglect in addition to visual field defects. This decision was made as there was one study in which some of the participants may have had neglect (Aimola 2011), and we considered the implications for including this group of participants of key clinical importance. Results of these analyses are reported above;

  • we carried out subgroup analyses to explore different ways of doing assessment in studies investigating the use of prisms (i.e. assessment wearing or not wearing prisms). This was considered central to the clinical relevance of the pooled result. Results of these analyses are reported above.

One active intervention versus another active intervention (narrative synthesis)

As planned, no meta‐analyses have been carried out to pool data from studies that compared one active intervention with another active intervention. A summary of results for relevant outcomes for these comparisons is provided in Table 5 and Table 6, and a brief narrative summary below.

Studies comparing two similar active interventions

(See Table 5)

One restitutive intervention versus another restitutive intervention

We found two studies that compared the effectiveness of two (or more) types of visual restitution therapy (Jobke 2009; Kasten 2007). Jobke 2009 compared extrastriate visual restitution therapy with conventional visual restitution therapy in a cross‐over study including 18 participants. After the first phase of the study, the extrastriate visual restitution therapy group had improved in measures of visual field (detection performance in HRP) by 5.9% and the conventional visual restitution therapy group had improved by 2.9%. No significant changes were found for either group for Radner reading scores. Kasten 2007 compared three treatment groups: 1) standard visual restitution therapy with single stimulation, 2) visual restitution therapy with parallel co‐stimulation, and 3) visual restitution therapy with moving co‐stimulation. We found no statistically significant differences between groups for measures of the visual field (perimetry).

We found two studies that investigated the addition of another intervention to visual restitution therapy (Plow 2010; Poggel 2004). Plow 2010 investigated the effect of adding transcranial direct current stimulation (tDCS) to visual restitution therapy and reported a greater increase in detection accuracy (perimetry) in the group receiving tDCS (9.37% increase compared to 5.55% increase in control group). Poggel 2004 compared standard visual restitution therapy with visual restitution therapy plus attentional cueing in 19 participants; results were primarily presented graphically and as whole‐group data: the authors concluded that the visual field border increased significantly more in the participants in the attentional cueing group.

One compensative intervention versus another compensative intervention

We found three studies that compared different compensative interventions (Keller 2010; Modden 2012; Schuett 2012). Keller 2010 compared audiovisual training and visual exploration training in 10 participants, reporting that audiovisual training was better than visual exploration training for outcomes of activities of daily living, reading, and visual scanning. Schuett 2012 compared reading training and visual exploration training in 36 participants within a cross‐over study. We used data reported from before the cross‐over period to calculate effect sizes for Schuett 2012 (see Figure 3); this demonstrated that visual exploration training was significantly more beneficial than reading training at improving scanning outcomes, and there was a trend towards reading training improving reading outcomes more significantly than visual exploration training. Modden 2012 compared 45 participants within three groups; one receiving computer‐based restitution therapy, one computer‐based compensation therapy, and one standard occupational therapy. We judged the occupational therapy intervention to be a compensatory intervention. We found no differences between the groups receiving the computer‐based compensatory therapy and occupational therapy interventions for measures of visual field enlargement, reading, or visual scanning.

Figure 3
One compensatory intervention versus another compensatory intervention.Effect sizes for Schuett 2012 (see )

One compensatory intervention versus another compensatory intervention.

Effect sizes for Schuett 2012 (see Table 5)

One substitutive intervention versus another substitutive intervention

We found two studies that compared the effectiveness of two types of prism (Bainbridge 1994; Szlyk 2005). Bainbridge 1994 compared the effect of full‐field and hemi‐field prisms in 18 participants, and reported a significant effect in favour of full‐field prisms for the cancellation test and Harrington Flocks Visual Field score. Szlyk 2005 compared Fresnel prisms and the Gottlieb Visual Field Awareness System in 10 participants and found no statically significant differences in outcomes between the two groups, although this was a cross‐over study with no data presented for the period of time before the cross‐over.

Studies comparing two different types of active intervention

(see Table 6)

Compensation intervention versus restitutive intervention

Two studies compared the effectiveness of a compensation intervention and a restitutive intervention (Modden 2012; Roth 2009). Roth 2009 compared explorative scanning training (a compensation intervention) with flicker stimulation training (a restitution intervention) in 29 participants, finding no significant differences between groups for key outcomes (although there were differences between groups at baseline for some outcomes). Modden 2012 compared 45 participants within three groups; one receiving computer‐based restitution therapy, one computer‐based compensation therapy, and one standard occupational therapy. In Table 6, we compared the results of the groups receiving computer‐based training. There were no significant differences reported for these groups for measures of visual field enlargement, activities of daily living, reading, or visual scanning.

Compensation intervention versus substitution intervention

One study compared the effectiveness of a compensative intervention (paper‐based visual search training) and a substitutive intervention (Fresnel prisms) (Rowe 2010). This was a three‐armed study, with a control (standard care) treatment; data from the active treatment arms compared to the control treatment have been included in meta‐analyses within this review. We found no significant differences between the compensative and substitutive intervention for measures of visual field, extended activities of daily living, and reading. We found a statistically significant difference in favour of the compensative intervention for the measure of quality of life (using the VFQ‐25).

Discussion

Summary of main results

We found 20 studies (732 randomised participants, with data for 638 participants, 547 of whom were participants with stroke) that investigated interventions for visual field defects in people with stroke. However, only 10 of these studies compared the effect of an intervention with a placebo, control, or no treatment group, which were the comparisons of interest to the review question, and only eight had data suitable for inclusion within meta‐analyses. Only two of these eight studies presented data relating to our primary outcome of functional abilities in activities of daily living, and there was a lack of consistency in outcome measures used across studies, which limited our ability to draw generalised conclusions.

Effect of restitutive interventions

Three studies (88 participants) compared a restitutive intervention with a control, but data were only available for one (19 participants). There was very low‐quality evidence that visual restitution therapy had no effect on visual field outcomes, and a statistically significant positive effect on quality of life. However, the data relating to the quality of life outcome must be interpreted with caution as the data used for analysis combined the population of interest with an additional 19 participants with optic nerve injury who had been included in a separate (but parallel) trial. These participants had damage to the anterior visual pathway, a population which was specifically excluded from this review, and in none of the participants was the optic nerve damage due to stroke. We, therefore, do not believe that the findings based on data from this population are applicable to the population of patients with visual field loss due to post‐chiasmal stroke. There is, therefore, insufficient evidence to draw any conclusions about the effectiveness of visual restitution therapy as compared to placebo, control, or no treatment. There was also some very limited evidence from two small studies which compared different types of restitutive interventions that there may be some benefits to adding either attentional cueing or tDCS to visual restitutive therapy, while a further two small studies found no difference in different modes of delivering visual restitution therapy.

Effect of compensatory interventions

Four studies (193 participants) compared a compensatory intervention with a control. There was low‐quality evidence of a beneficial effect on measures of quality of life. However, there was low‐ or very low‐quality evidence of no effect on measures of visual field, extended activities of daily living, reading, and scanning ability. Findings from a small study comparing two different types of compensatory therapy conflicts with the evidence of no effect on scanning outcome, demonstrating a beneficial effect of visual exploration therapy on scanning outcomes when compared to reading training. There is, therefore, some limited low‐quality evidence that compensatory scanning training may improve an important outcome (quality of life) in people with visual field defects following stroke, but 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. Studies comparing different active interventions provide some very limited evidence that there may be benefits associated with audiovisual training, rather than visual exploration training alone, and suggest that compensatory training, as delivered during standard clinical practice, may be as effective as computer‐based training interventions.

Effect of substitutive interventions

Three studies (166 participants) compared a substitutive intervention (a type of prism) with a control. There was low‐ or very low‐quality evidence that prisms did not have an effect on measures of activities of daily living, extended activities of daily living reading, falls, or quality of life, but that they may have an effect on scanning ability. Evidence from one study indicates that people using prisms may have a higher odds of adverse events, particularly headache. However, it is important to note a fundamental difference between these studies in that, in two of the studies, participants in the treatment group wore prisms during the outcome assessments (Bowers 2014; Rossi 1990), while in one study none of the participants wore prisms during outcome assessment by participants (Rowe 2010). Evidence relating to measures of visual field demonstrated a significant difference between studies that measured outcome whilst wearing or not wearing the substitutive device, with a beneficial effect when the prisms were worn during testing but not without. However, due to the quality and quantity of evidence, we remain uncertain about the benefits of prisms.

Effect of screening/assessment interventions

One study (39 participants) compared the effect of assessment by an orthoptist to standard care (no assessment) and found very low‐quality evidence that there was no effect on measures of activities of daily living. However, due to the quality and quantity of evidence, we remain uncertain about the benefits of assessment interventions.

Ten studies compared the effect of two of more different active interventions and did not include a control group.

In summary, this review has identified:

  • a lack of evidence relating to the effect of interventions on our primary outcome (functional ability in activities of daily living);

  • low‐quality evidence that compensatory scanning training is more beneficial than placebo or control at improving quality of life, but not other outcomes;

  • insufficient evidence to reach any generalised conclusions about the effect of restitutive interventions, substitutive interventions (prisms), or screening/assessment interventions as compared to placebo, control or no treatment;

  • low‐quality evidence that prisms may result in an increase in the number of people experiencing adverse events (particularly headache).

Overall completeness and applicability of evidence

Studies with control, placebo, or no treatment group

Although we identified 20 studies that investigated the effect of interventions for visual field defects after stroke, only 10 of these studies (and only eight with data) compared the effect of an intervention with a control, placebo, or no treatment group, which was the question of interest to this review. The remaining 10 studies compared different types of interventions, with nine of these studies comparing interventions of the same 'type' (i.e. comparing two substitutive interventions or two restitutive interventions); arguably, comparisons of different active interventions have little merit until such time as the benefits (and harms) of active interventions, as compared to control, placebo, or no treatment have been established. Three studies did compare the relative effects of two different types of intervention (i.e. compensatory versus restitutive, or compensative versus substitutive). Thus, although we made the decision to include all the studies which investigated the effectiveness of interventions for visual field defects, in fact only nine of these studies included comparisons that were directly relevant to the review question, focusing on a comparison with a control, placebo, or no treatment group. Five of the 20 included studies, including two of the 10 studies with a control/placebo comparison, were cross‐over studies. The reporting of data from after the cross‐over in studies with this design limited our ability to incorporate data within meta‐analyses in this review, and there is a risk that outcomes from after the cross‐over are affected by the treatments administered prior to the cross‐over.

Restitutive interventions versus no treatment, placebo, or control

Kasten 1998 was the one study comparing a restitutive intervention with a control or placebo, with data suitable for inclusion in our meta‐analyses. This study included 19 participants, only 10 of whom had stroke. The study did not measure our primary outcome of interest (functional ability in activities of daily living). There was an uneven distribution of stroke patients between the two groups (with two stroke patients out of the nine participants in the treatment group and eight stroke patients out of 10 participants in the control group). This uneven distribution means that only two patients with stroke received an active intervention, providing evidence from which it would be inappropriate to generalise. Furthermore, for quality of life outcome data, the 19 participants in this study had been combined with the results of 19 participants in another similar study. However, the additional 19 participants all had optic nerve injury rather than post‐chiasmal injury. Due to the nature of the participants included in this study, it would not be appropriate to make generalisations from this evidence to the population of stroke patients with visual field defects after stroke. Our confidence in the findings from this evidence is very low.

Compensatory interventions versus no treatment, placebo, or control

We were able to combine data from four studies (162 participants) comparing compensatory (scanning) interventions with control or placebo in meta‐analysis. None of these studies measured our primary outcome of interest (functional ability in activities of daily living). Visual field defects in the participants included in these studies were confirmed using perimetry (see Table 3). Three of the studies included participants with visual field defects and no co‐existing visual neglect; two of these studies excluded participants with neglect based on clinical testing (De Haan 2015; Rowe 2010), while one only included participants with a left‐sided cerebrovascular accident where patients rarely experience persistent neglect (Spitzyna 2007). One study included participants who had visual field defects but possibly also co‐existing visual neglect (Aimola 2011); clinical testing confirmed that only three of the 52 participants had confirmed neglect. Participants included those from a mixed population (i.e. stroke and other neurological conditions) for three of the four studies; 41 of the 49 participants had stroke in De Haan 2015, 39 of 52 had stroke in Aimola 2011, and 13 of 22 had stroke in Spitzyna 2007. All participants had stroke in Rowe 2010. The majority of participants in these studies, therefore, had visual field defects following stroke and did not have co‐existing neglect (as confirmed by clinical testing); therefore, it would be appropriate to generalise from these results to the population of stroke survivors with visual field defects and no neglect.

The nature of the scanning training in the four studies combined within the meta‐analyses varied considerably; two investigated computer‐based scanning training, one focused on visual exploration training (Aimola 2011), one on reading training using scrolling horizontal text (Spitzyna 2007), one was a training programme, primarily delivered face‐to‐face by an occupational therapist (De Haan 2015), and one was a self‐delivered paper‐based scanning exercise (Rowe 2010). It is likely that the nature of these interventions will result in varied scanning movements of the eye. As well as differences in the mode of delivery of the scanning training, there were also differences in the amount of training. These differences in the compensatory interventions reduce confidence in any findings, and limit ability to generalise from these findings. During the GRADE assessment of quality of evidence, we applied a downgrade to each comparison combining results from these studies due to the variations in the interventions studied. We, therefore, have low‐ to very‐low confidence in the findings from this evidence.

Substitutive interventions versus no treatment, placebo, or control

Two studies compared a substitutive intervention (prisms) with a no treatment control (Rossi 1990; Rowe 2010). Rossi 1990 measured our primary outcome of interest (functional ability in activities of daily living). Both studies only included participants with stroke, but Rossi 1990 included participants with visual neglect in additional to visual field defect. The studies both investigated the effect of Fresnel prisms. There was a fundamental difference between these studies relating to the assessment of outcome: Rossi 1990 measured outcomes whilst participants in the intervention group wore the assigned prisms, whilst Rowe 2010 measured outcomes while participants were not wearing any assigned prisms. It has been argued that the rationale for prisms is that they provide visual field expansion when in use, and that consequently outcomes from clinical trials exploring the effectiveness of prisms should be measured with participants wearing the prisms (Bowers 2014). In contrast, Rowe 2010 measured outcomes without use of assigned prisms in order to preserve blinding of outcome assessor and enable direct comparison of study treatment groups. It was, therefore, not appropriate to pool data from these different studies of prisms, as the wearing of prisms during assessment in one study but not the other makes the results incomparable. This is an important issue which must be appropriately considered in future trials. Due to the methodological limitations of these studies, and the inability to combine results, we have low to very low confidence relating to this evidence.

Assessment/screening intervention versus control

One study explored the effect of implementing a full visual assessment by an orthoptist and sharing the results with hospital staff (Jarvis 2012). This study measured our primary outcome of interest (functional ability in activities of daily living). The evidence was judged to be very low quality, limiting our confidence in the findings from this study.

Quality of the evidence

For this updated review, we judged the quality of evidence using the GRADE approach. We judged all evidence included within meta‐analyses to be of low or very low quality. Key factors contributing to downgrading of the evidence within these comparisons included:

Risk of bias

We identified concerns about the methodology for the majority of included studies, and there were often insufficient details available from incomplete reporting of methodological details. We judged only eight of the 20 studies to be at low risk of bias for allocation concealment, six of the 20 for blinding of outcome assessment, and four of the 20 for incomplete outcome assessment. Furthermore, five of the studies were carried out by researchers who have acknowledged a financial interest in intervention (Bowers 2014; Jobke 2009; Kasten 1998; Kasten 2007; Poggel 2004); we assessed this as potentially introducing a source of bias.

Imprecision

The number of participants within the included studies was small, ranging from 10 to 87 participants, with only five of the 20 studies including more than 50 participants. While there were a total of 732 randomised participants, variations in studies made it inappropriate for the majority of study data to be combined within analyses, and the maximum number of participants with data combined in a single analysis was 162 (Analysis 2.3). The small number of participants within the included studies and suitable for combination within meta‐analyses limits the conclusions that can be drawn from this evidence.

Indirectness

A number of factors contributed to indirectness of the data included within meta‐analyses. In particular:

  • population: there was considerable heterogeneity between the populations recruited to individual studies. In addition to stroke‐related differences, such as time post‐stroke, initial impairment, and the presence of other stroke‐related impairments (e.g. communication, mobility), this was confounded by the inclusion of participants with conditions other than stroke, and in opposing the decision to either include or exclude participants with visual neglect. The variations in populations contributed to decisions to downgrade the quality of evidence, as this reduced our certainty in the reported findings;

  • interventions: this review aimed to synthesise evidence relating to a wide range of different interventions for visual field defects following stroke and preplanned categories to support appropriate combination of evidence. However, we found substantial variations in the interventions within these different categories in relation to the details of the delivered interventions. In particular, the compensative interventions had considerable variation in the mode of delivery, with interventions varying from computer‐based scanning training and reading training, paper‐based scanning training, to face‐to‐face scanning and mobility training. What is being delivered in terms of the eye movements being trained with these different interventions is likely to vary considerably, and ‐ while data from these 'scanning' interventions have been combined ‐ the variations in interventions limit our ability to be confident about the pooled result;

  • outcomes: as is highlighted in Table 4, there was a lack of consistency in the outcomes assessed by individual studies, and in the assessment tools used to do this. While there is an argument that outcome measures should be carefully selected according to the anticipated action of, or scientific rationale for, the intervention (Bowers 2017), the variations in interventions do not fully explain the lack of consistency between the outcome measures. Within the studies of prisms, the difference in choice of outcome measure was further confounded by opposing decisions relating to whether outcomes were assessed wearing or not wearing the prisms. The variations in outcome measures limited the ability to pool data from individual studies in a meaningful way, and where measures have been pooled, limited our certainty in the result.

In summary, we judged the quality of the evidence synthesised within this review to be low to very low, and this limits our confidence in the results. Future research needs to address the factors which contribute to this level of evidence, in order to produce results which are useful and meaningful.

Potential biases in the review process

Publication bias

Through a thorough searching process we are quite confident that we should have identified all relevant published studies. However, at the peer review stage of this review update, we were alerted to the fact that we had erroneously excluded a relevant study (Elshout 2016): while this was corrected prior to publication, it does highlight the potential for human error in our process of screening titles. A limitation of our search strategy for this update is that we have not searched a number of trials databases beyond March 2015 (Current Controlled Trials, Health Service Research Projects in Progress, National Eye Institute Clinical Studies Database, Stroke Trials Registry); this may have limited our ability to identify ongoing trials, but ought not to have impacted on our identification of completed trials. It must be acknowledged that there is a small possibility that there are additional studies (published and unpublished) that we did not identify. We had planned to explore the effect of publication type using sensitivity analyses; however, all data included in meta‐analyses were from peer‐reviewed journals.

Categorisation of interventions

Although we anticipated that we may experience difficulty in categorising the interventions studied into our predefined categories of restitutive, compensative, and substitutive interventions, this was not the case and the categorisation process was a clear and unambiguous process. This was because the studies we identified were primarily visual restitution therapy (restitutive), compensatory scanning training (compensative), or prisms (substitutive). We are, therefore, confident that our categorisation of interventions has not introduced bias into the review process. However, we did find substantial differences between the interventions within each category (see discussion above), and decisions to combine data from varied interventions may reduce applicability of these results. Future updates of this review should, therefore, consider and preplan which interventions it is clinically relevant to combine. Involvement of key stakeholders to inform this decision making for future updates would be an advantage.

Inclusion criteria: participants

In the previous version of this review, we reported that the inclusion criterion that was judged as most difficult to assess by the independent review authors was the participants. The particular difficulty encountered was with studies that did not appear to include the diagnosis of visual field defects as an inclusion criterion. In the previous version, we identified and included several studies that either used a clinical assessment of 'scanning' as an inclusion criterion without formally assessing or diagnosing either visual field defects or visual neglect, or which included participants with a right‐sided cerebrovascular accident, making the assumption that these participants would have visual neglect (and possibly also visual field defects). For this update of the review, we addressed this difficulty by only including studies that reported a method for diagnosing visual field defects at the recruitment stage. This led to the exclusion of a number of studies that had previously been included, but in which the participants did not have confirmed visual field defects (Carter 1983; Weinberg 1977; Weinberg 1979). We made this change between the previous version and this updated version of the review in order to reduce potential bias in the review process; however, we acknowledge that this may have led to the exclusion of studies in which some participants had visual field defects that were not confirmed through clinical diagnosis (e.g. instead manifesting as a scanning problem).

Outcomes

Categorisation of some reported outcome assessments into our predefined outcomes of interest was difficult in some cases. For this update of the review, we added a table to report our categorisation of outcomes to ensure transparency in this process (Table 4). We also reconsidered categorisation of all outcomes from trials included in the previous version of the review, and made a number of changes through a process of consensus. For example, Plow 2010 reported the Veterans Affairs Low Vision‐Visual Functional Questionnaire (LV‐VFQ) which "assesses an individual's visual ability to perform ADLs across 4 domains, including reading, mobility, visual motor function, and visual processing". This measure arguably relates to both extended activities of daily living and quality of life. In the first version of this review, this was listed as a measure of EADL: however, this was changed to being listed as a measure of QoL outcome for subsequent updates, following consensus discussion between review authors.

The primary outcome for this review was functional ability in activities of daily living, measured using standardised scales. It has been argued that measurement of effectiveness in rehabilitation ought to take into account patients' individual goals (Turner‐Stokes 2009). There is growing evidence that goal attainment scaling (a standardised method of scoring performance of patient‐specific tasks) may provide a valid, reliable, sensitive method of evaluating outcomes that are of greatest importance to individual patients (Krasny‐Pacini 2016). This scale was not considered for inclusion within this review, and could be considered for future updates.

As has been discussed under Quality of the evidence, variations in outcomes and outcomes assessment tools between included studies created challenges for the synthesis of evidence within this review. The need to make judgements and decisions relating to categorisation and pooling of outcomes created potential for the introduction of bias into the review process. We have aimed for transparency in our decision making and reporting in an attempt to avoid the introduction of bias. However, to support the creation of meaningful evidence syntheses and meta‐analyses, there is a need for consensus between stroke survivors, their families and carers, health professionals, and researchers in relation to core outcomes for trials relating to interventions for visual field defects after stroke as recommended by the COMET Initiative.

Agreements and disagreements with other studies or reviews

Agreements and disagreements between this updated version and previous version

In the previous 2011 version of this review, we stated the key conclusions and implications for practice arising from the evidence as follows.

  • There is limited evidence which supports the use of compensatory scanning training for patients with visual field defects (and possibly co‐existing visual neglect) to improve visual field, scanning, and reading outcomes.

  • There is insufficient evidence to reach a conclusion about the impact of compensatory scanning training on functional activities of daily living.

  • There is also insufficient evidence to reach generalised conclusions about the benefits of visual restitution therapy (restitutive intervention) or prisms (substitutive intervention) for patients with visual field defects after stroke.

Key changes in the methods between different versions of this review include:

  • updated searches in this updated version, increasing the number of included studies;

  • amended inclusion criteria, leading to the exclusion of studies in which participants did not have confirmed visual field defects;

  • the use of the GRADE approach to systematically assess quality of evidence in this updated version.

These changes have highlighted further uncertainty around previous limited evidence supporting the use of compensatory interventions and have introduced evidence relating to adverse events associated with substitutive interventions, but have not resulted in any changes in conclusions relating to restitutive or substitutive interventions.

  • There is limited low‐quality evidence that compensatory scanning training may improve an important outcome (quality of life) in patients with visual field defects following stroke, but 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. There is insufficient evidence to reach a conclusion about the impact of compensatory scanning training on other outcomes.

  • There remains insufficient evidence to reach generalised conclusions about the benefits of visual restitution therapy (restitutive intervention) or prisms (substitutive intervention), and there is insufficient evidence to reach conclusions about the effect of screening or assessment interventions for patients with visual field defects after stroke.

  • There is now some low‐quality evidence from one study that prisms may result in an increased number of adverse events, particularly headache.

Agreements and disagreements with other published reviews

The Royal College of Physicians updated the evidence included in the earlier version of this review, concluding that there is "insufficient evidence regarding the effectiveness of interventions aimed at improving function in people with visual field defects" (ISWP 2016). This updated review is in agreement with the conclusions drawn from the evidence in the guideline.

The Scottish Intercollegiate Guideline Network (SIGN) guidelines for stroke rehabilitation state that there is "limited poor quality evidence suggesting that visual scanning compensatory training techniques may be effective in improving functional outcomes after stroke" (SIGN 2011). This SIGN guideline (updated in 2010) is based on a number of other reviews (Barrett 2009; Bouwmeester 2007; Jones 2006). While the previous version of our review was in agreement with the recommendations made by the Royal College of Physicians and SIGN guidelines, our updated review has highlighted further uncertainty relating to the effect of compensatory interventions. The National Institute for Clinical Excellence (NICE) stroke guidelines recommends that "eye movement therapy" is provided to "people who have persisting hemianopia after stroke and who are aware of the condition" (NICE 2013): our updated review does not directly support this recommendation.

Our review is in agreement with the conclusions in other reviews that the evidence relating to the effectiveness of visual restoration therapy is inconsistent and of poor quality (Barrett 2009; Bouwmeester 2007), and that few studies have assessed functional ability in activities of daily living as an outcome.

Our review is in agreement with narrative reviews of evidence for interventions for visual problems after stroke (Lane 2008, Hanna 2017), which have concluded that there is a need for high‐quality studies of the effectiveness of interventions for visual field defects.

Study flow diagram.

Figures and Tables -
Figure 1

Study flow diagram.

Navigate to figure in ReviewOpen in new tab

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figures and Tables -
Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Navigate to figure in ReviewOpen in new tab

One compensatory intervention versus another compensatory intervention.Effect sizes for Schuett 2012 (see )

Figures and Tables -
Figure 3

One compensatory intervention versus another compensatory intervention.

Effect sizes for Schuett 2012 (see Table 5)

Navigate to figure in ReviewOpen in new tab

Comparison 1: Restitutive interventions versus control, placebo or no intervention, Outcome 1: Visual field

Figures and Tables -
Analysis 1.1

Comparison 1: Restitutive interventions versus control, placebo or no intervention, Outcome 1: Visual field

Navigate to figure in ReviewOpen in new tab

Comparison 1: Restitutive interventions versus control, placebo or no intervention, Outcome 2: Quality of life

Figures and Tables -
Analysis 1.2

Comparison 1: Restitutive interventions versus control, placebo or no intervention, Outcome 2: Quality of life

Navigate to figure in ReviewOpen in new tab

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 1: Visual field

Figures and Tables -
Analysis 2.1

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 1: Visual field

Navigate to figure in ReviewOpen in new tab

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 2: Extended activities of daily living

Figures and Tables -
Analysis 2.2

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 2: Extended activities of daily living

Navigate to figure in ReviewOpen in new tab

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 3: Reading

Figures and Tables -
Analysis 2.3

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 3: Reading

Navigate to figure in ReviewOpen in new tab

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 4: Quality of life

Figures and Tables -
Analysis 2.4

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 4: Quality of life

Navigate to figure in ReviewOpen in new tab

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 5: Scanning ‐ cancellation

Figures and Tables -
Analysis 2.5

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 5: Scanning ‐ cancellation

Navigate to figure in ReviewOpen in new tab

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 6: Adverse events

Figures and Tables -
Analysis 2.6

Comparison 2: Compensative interventions versus control, placebo or no intervention, Outcome 6: Adverse events

Navigate to figure in ReviewOpen in new tab

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 1: Functional Activities of Daily Living

Figures and Tables -
Analysis 3.1

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 1: Functional Activities of Daily Living

Navigate to figure in ReviewOpen in new tab

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 2: Visual field

Figures and Tables -
Analysis 3.2

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 2: Visual field

Navigate to figure in ReviewOpen in new tab

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 3: Extended activities of daily living

Figures and Tables -
Analysis 3.3

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 3: Extended activities of daily living

Navigate to figure in ReviewOpen in new tab

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 4: Reading

Figures and Tables -
Analysis 3.4

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 4: Reading

Navigate to figure in ReviewOpen in new tab

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 5: Falls

Figures and Tables -
Analysis 3.5

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 5: Falls

Navigate to figure in ReviewOpen in new tab

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 6: Quality of life

Figures and Tables -
Analysis 3.6

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 6: Quality of life

Navigate to figure in ReviewOpen in new tab

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 7: Scanning ‐ cancellation

Figures and Tables -
Analysis 3.7

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 7: Scanning ‐ cancellation

Navigate to figure in ReviewOpen in new tab

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 8: Adverse events

Figures and Tables -
Analysis 3.8

Comparison 3: Substitutive interventions versus control, placebo or no intervention, Outcome 8: Adverse events

Navigate to figure in ReviewOpen in new tab

Comparison 4: Assessment or screening versus control, placebo or no intervention, Outcome 1: ADL

Figures and Tables -
Analysis 4.1

Comparison 4: Assessment or screening versus control, placebo or no intervention, Outcome 1: ADL

Navigate to figure in ReviewOpen in new tab
Summary of findings 1. Summary of findings: Restitutive interventions versus control

Restitutive interventions compared with control for visual field defects in patients with stroke

Patient or population: stroke survivors with visual field defects

Settings: any rehabilitation setting

Intervention: restitutive interventions

Comparison: control, placebo, or no intervention

Outcomes

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Functional ability in activities of daily living

(no data)

No studies

Insufficient evidence

Visual field

(TAP border position in degrees of visual angle from zero vertical meridian)

After intervention

MD 1.02 (‐1.37 to 3.41)

19
(1 study, Kasten 1998)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study had high ROB for 'other bias' (relating to potential financial interest in the intervention), study had uncertain ROB for allocation concealment and incomplete outcome data

  • Indirectness ‐ included participants with diagnoses other than stroke

  • Imprecision ‐ very small study population (n = 19)

Extended activities of daily living

(no data)

No studies

Insufficient evidence

Reading ability

(no data)

No studies

Insufficient evidence

Falls

(no data)

No studies

Insufficient evidence

Quality of life

(improved or not improved ‐ derived from percentage of those who reported subjective improvements of vision)

OR 13.00 (2.07 to 81.48)

30*

(1 study, Kasten 1998)

*The data used in this analysis were derived from 30 of the original 38 participants, which included data from an additional 19 participants with optic nerve injury who had also received the same interventions in a separate (but parallel) trial. Participants with optic nerve injury do not meet the inclusion criteria for this review.

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study had high ROB for 'other bias' (relating to potential financial interest in the intervention), study had uncertain ROB for allocation concealment and incomplete outcome data

  • Indirectness ‐ analysis contained data from a subset of participants from a separate trial, who were not relevant to this review

  • Indirectness ‐ included participants with diagnoses other than stroke

  • Imprecision ‐ very small study population (n = 19)

Scanning ‐ cancellation

(no data)

No studies

Insufficient evidence

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.

CI: confidence interval
MD: mean difference
n: number
OR: odds ratio
ROB: risk of bias
TAP: Tuebingen Automated Perimeter

Figures and Tables -
Summary of findings 1. Summary of findings: Restitutive interventions versus control
Navigate to table in Review
Summary of findings 2. Summary of findings: Compensative interventions versus control

Compensative interventions compared with control for visual field defects in patients with stroke

Patient or population: stroke survivors with visual field defects

Settings: any rehabilitation setting

Intervention: compensative interventions

Comparison: control, placebo, or no intervention

Outcomes

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Functional ability in activities of daily living

(no data)

No studies

Insufficient evidence

Visual field

(Functional field score and relative change in visual field score, combined)

After intervention

SMD ‐0.11 (‐0.92 to 0.70

(no significant effect)

95

(2 studies, De Haan 2015; Rowe 2010)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ one study judged as high risk of bias for at least one domain

  • Inconsistency ‐ one study had baseline differences

  • Inconsistency ‐ I2 = 75%

  • Indirectness ‐ studies explored very different compensatory interventions

Extended activities of daily living

(Mobility questionnaire and change in Nottingham EADL, combined)

After intervention

SMD 0.49 (‐0.01 to 0.99)

(no significant effect)

97

(2 studies, De Haan 2015; Rowe 2010)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ one study judged as high risk of bias for at least one domain

  • Indirectness ‐ outcome measures were very different; for one study outcome was a mobility measure, rather than a general measure of EADL

  • Indirectness ‐ studies explored very different compensatory intervention

s

Reading ability

(Reading speed; various tests)

After intervention

SMD 0.26 (‐0.05 to 0.58)

(no significant effect)

162

(4 studies, Aimola 2011; De Haan 2015; Rowe 2010; Spitzyna 2007)

⊕⊕⊝⊝
low

Reasons for downgrades:

  • Risk of bias ‐ three studies judged as high risk of bias for at least one domain

  • Indirectness ‐ studies explored very different compensatory intervention

Falls

(no data)

No studies

Insufficient evidence

Quality of life

(National Eye Institute Visual Function Questionnaire (NEI ‐ VFQ‐25) total score)

After intervention

MD 9.36 (3.10 to 15.62)

(favours compensatory)

96

(2 studies, De Haan 2015; Rowe 2010)

⊕⊕⊝⊝
low

Reasons for downgrades:

  • Risk of bias ‐ two studies judged as high risk of bias for at least one domain

  • Indirectness ‐ studies explored very different compensatory interventions

Scanning ‐ cancellation

(cancellation tests ‐ time to complete)

After intervention

SMD ‐0.01 (‐0.40 to 0.39)

(no significant effect)

97

(2 studies, Aimola 2011; De Haan 2015)

⊕⊕⊝⊝
low

Reasons for downgrades:

  • Risk of bias ‐ two studies judged as high risk of bias for at least one domain

  • Indirectness ‐ studies explored very different compensatory interventions

Adverse events

(number of participants with reported events during intervention period)

OR 5.18 (0.24 to 112.57

(favours control)

108

(2 studies, De Haan 2015; Rowe 2010)

(NB. no events recorded in De Haan 2015, which did not explicitly report adverse events as an outcome measure)

⊕⊕⊝⊝
low

Reason for downgrades:

  • Inconsistency ‐ no events from one study, means pooled result was not estimable for that study; large confidence intervals

  • Indirectness ‐ studies explored very different compensatory interventions

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.

CI: confidence intervals
EADL: extended activities of daily living
MD: mean difference
NEI‐VFQ‐25: National Eye Institute Visual Function Questionnaire
OR: odds ratio
SMD: standardised mean difference

Figures and Tables -
Summary of findings 2. Summary of findings: Compensative interventions versus control
Navigate to table in Review
Summary of findings 3. Summary of findings: Substitutive interventions versus control

Substitutive interventions compared with control for visual field defects in patients with stroke

Patient or population: stroke survivors with visual field defects

Settings: any rehabilitation setting

Intervention: compensative interventions

Comparison: control, placebo, or no intervention

Outcomes

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Functional ability in activities of daily living

(Barthel Index)

After 4 weeks of treatment

Wearing prisms

MD ‐4.00 (‐17.86 to 9.86)

(no significant effect)

39

(1 study, Rossi 1990)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study judged as high risk of bias for at least one domain

  • Indirectness ‐ included data from participants with neglect

  • Imprecision ‐ small study population (n = 39)

Visual field

(change in visual field area & change in error scores, from baseline)

After intervention

Not wearing prisms

SMD 0.12 (‐0.46 to 0.70)

Wearing prisms

SMD 1.12 (0.44 to 1.80)

85

(2 studies, Rossi 1990; Rowe 2010)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ one study judged as high risk of bias for at least one domain

  • Indirectness ‐ included data from participants with neglect

  • Indirectness ‐ studies cannot be combined due to differences in testing (wearing/not wearing prisms)

Extended activities of daily living

(Change in EADL from baseline; mobility improvement scores, in Logits)

After intervention

Not wearing prisms

SMD 0.20 (‐0.44 to 0.85)

Wearing prisms

SMD 0.24 (‐0.26 to 0.75)

99

(2 studies, Bowers 2014; Rowe 2010)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ one study judged as high risk of bias for at least one domain

  • Indirectness ‐ one study outcome was a mobility measure, rather than a general measure of EADL

  • Indirectness ‐ included participants with diagnoses other than stroke

  • Indirectness ‐ studies cannot be combined due to differences in testing (wearing/not wearing prisms)

Reading ability

Not wearing prisms

MD 2.80 (‐7.13 to 12.73)

(no significant effect)

45

(1 study, Rowe 2010)

⊕⊕⊝⊝
Low

Reasons for downgrades:

  • Imprecision ‐ small study population (n = 45)

  • Imprecision ‐ wide confidence intervals

Falls

(number of falls)

After intervention

Wearing prisms

OR 1.21, (0.26 to 5.76)

(no significant difference)

39

(1 study, Rossi 1990)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study judged as high risk of bias for at least one domain

  • Indirectness ‐ included data from participants with neglect

  • Imprecision ‐ small study population (n = 39)

Quality of life

(Visual Function Questionnaire (VFQ‐25))

After intervention

Not wearing prisms

MD 8.40 (‐4.18 to 20.98)

(no significant effect)

43

(1 study, Rowe 2010)

⊕⊕⊝⊝
Low

Reasons for downgrades:

  • Imprecision ‐ small study population (n = 43)

  • Imprecision ‐ wide confidence intervals

Scanning ‐ cancellation

(line cancellation errors)

After intervention

Wearing prisms

MD 9.80 (1.91 to 17.69)

(favours substitutive)

39

(1 study, Rossi 1990)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study judged as high risk of bias for at least one domain

  • Indirectness ‐ included data from participants with neglect

  • Imprecision ‐ small study population (n = 39)

  • Imprecision ‐ wide confidence intervals

Adverse events

(number of participants with reported events during intervention period)

OR 87.32 (4.87 to 1564.66)

(favours control)

59

(1 study, Rowe 2010)

⊕⊕⊝⊝
Low

Reason for downgrades:

  • Inconsistency ‐ large confidence intervals

  • Imprecision ‐ data from only one study

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.

EADL: extended activities of daily living
MD: mean difference
OR: odds ratio
SMD: standardised mean difference
VFQ‐25: Visual function questionnaire

Figures and Tables -
Summary of findings 3. Summary of findings: Substitutive interventions versus control
Navigate to table in Review
Summary of findings 4. Summary of findings: Assessment/screening interventions versus control

Assessment/screening interventions compared with control for visual field defects in patients with stroke

Patient or population: stroke survivors with visual field defects

Settings: any rehabilitation setting

Intervention: assessment/screening interventions

Comparison: control, placebo, or no intervention

Outcomes

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Functional ability in activities of daily living

(FIM)

After intervention

MD ‐6.97 (‐23.78 to 9.84)

(no significant effect)

37

(1 study, Jarvis 2012)

⊕⊝⊝⊝
very low

Reasons for downgrades:

  • Risk of bias ‐ study judged as high risk of bias for at least one domain

  • Imprecision ‐ small study population (n = 37)

  • Imprecision ‐ wide confidence intervals

Visual field

(no data)

No studies

Insufficient evidence

Extended activities of daily living

(no data)

No studies

Insufficient evidence

Reading ability

(no data)

No studies

Insufficient evidence

Falls

(no data)

No studies

Insufficient evidence

Quality of life

(no data)

No studies

Insufficient evidence

Scanning ‐ cancellation

(no data)

No studies

Insufficient evidence

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.

FIM: Functional Independence Measure
MD: mean difference

Figures and Tables -
Summary of findings 4. Summary of findings: Assessment/screening interventions versus control
Navigate to table in Review
Table 1. Demographics of included studies: settings of included studies

Study

Country

Number of centres

Setting for intervention

Trial registration

Aimola 2011

UK

Multicentre ("from local hospitals or as self‐referrals")

Community

(participants' own homes)

UK Clinical Research Network Portfolio (UKCRN, ID 7144)

Bainbridge 1994

USA

Single

NS

NS

Bowers 2014

UK, USA

Multicentre (13 study sites)

University, hospital, private practice for fitting of prisms

Then use at home (participants' own homes)

clinicaltrials.gov

NCT00494676

De Haan 2015

Netherlands

2 ("Royal Dutch Visio and Bartiméus, the two centers of expertise for blind and partially sighted people in the Netherlands")

Training ... "was provided in Dutch at nine locations of Royal Dutch Visio and one location of Bartiméus in the Netherlands".

Participants were also given homework assignments.

ISRCTN Registry

ISRCTN16833414

Elshout 2016

Netherlands

Unclear ("Patients throughout the Netherlands could sign up for our study voluntarily by filling in a form on our website")

Community

(participants' own homes)

NS

Gall 2013

Not clear

NS

NS

NS

Jarvis 2012

UK

Single

Stroke unit, Warring and Halton Hospitals,

NHS Foundation Trust

NS

Jobke 2009

Germany

NS

NS

NS

Kasten 1998

Germany

NS

Community

(participants' own homes)

NS

Kasten 2007

Germany

NS

Community

(participants' own homes)

NS

Keller 2010

Germany

Single

Neurological clinic

NS

Modden 2012

Germany

Single

Rehabilitation centre (inpatients)

NS

Plow 2010

USA

Single

Outpatient

(University clinic)

clinicaltrials.gov NCT00921427

Poggel 2004

Germany

Single

Community

(participants' own homes)

NS

Rossi 1990

USA

Single

Rehabilitation

(inpatient)

NS

Roth 2009

Germany

NS

Community

(participants' own homes)

NS

Rowe 2010

UK

Multicentre ("from stroke units based in 15 United Kingdom (UK) National Health Service (NHS) trusts")

Any (hospital, community)

Current Controlled Trials ISRCTN05956042.

Schuett 2012

Unclear. Authors from Austria, UK and Germany. "All participants were native German speakers."

NS

NS

NS

Spitzyna 2007

UK

NS

Community

(participants' own homes)

NS

Szlyk 2005

USA

Single; university

Outpatient clinic

NS

NHS: National Health Service
NS: not stated
UK: United Kingdom
USA: United States of America

Figures and Tables -
Table 1. Demographics of included studies: settings of included studies
Navigate to table in Review
Table 2. Demographics of included studies: demographics of included participants

Study

Number of participants

Age

Gender

Time since stroke/lesion

Initial functional ability

Type of stroke/lesion

Side of stroke/lesion

Aimola 2011

70 participants recruited, 52 participants included in analyses

Group 1

Mean 61.4 years, SD 10.3

Group 2

Mean 63.0 years, SD 10.9

Group 1

9 F

19 M

Group 2

7 F

17 M

NS

NS

Group 1

19 ischaemic stroke

4 haemorrhagic,

4 traumatic brain injury

1 tumour

Group 2

20 ischaemic stroke

2 haemorrhagic

2 traumatic brain injury

0 tumour

Side of field defect

Group 1

L 15/R 13

Group 2

L11/R 13

Bainbridge 1994

18

NS

NS

NS

NS

NS

NS

Bowers 2014

73 randomised; 67 completed first phase (before cross‐over); 61 completed second phase (after cross‐over)

For 61 participants included after the cross‐over:

median 58 years (range 18 to 89)

For 61 participants included after the cross‐over:

M 40

F 21

For 61 participants included after the cross‐over:

median 18 months (range 3 to 396)

Overall baseline mobility difficulty, for 61 participants included after the cross‐over:

mean ‐0.17 (SD 2.31) logits for n = 31 using oblique prisms

mean ‐0.06 (SD 1.89) logits for n = 30 using horizontal prisms

For 61 participants included after the cross‐over:

hemianopia was caused by stroke for 47 (77%)

For 61 participants included after the cross‐over:

L hemianopia 39 (64%)

De Haan 2015

54 randomised; data from 49 analysed (training group n = 26, control group n = 23)

Training group 55 ± 10.1 years

Control group 57 ± 13.0 years

M 32

F 17

Training group 18 ± 22.5 months

Control group 22 ± 24.6 months

NS

Ischaemic CVA 36

Haemorrhagic CVA 5

Traumatic brain injury 3

Penetrating head trauma 1

AVM extirpation 1

combined 3

L hemianopia 33

R hemianopia 16

Elshout 2016

40 recruited; data presented from first 3 cohorts of 10 only (n = 30); data from 27 analysed

Mean age 51.2 years (range 29 to 74)

M 22

F 5

Mean 26.3 months

(range 11 to 111)

NS

5 haemorrhagic stroke

22 ischaemic stroke

L‐sided field defect 14

R‐sided field defect 13

Gall 2013

39 (alternating current stimulation n = 15, sham n = 14)

NS

NS

NS

NS

NS ("patients with post‐chiasmatic visual pathway lesions")

NS

Jarvis 2012

64 randomised (experimental group n = 33, control n = 31)

Experimental: mean 70.4 years (SD 10.8)

Control: mean 69.4 years (SD 14.5)

M 40

F 24

NS

NS

Ischaemic 56

Haemorrhage 7

Combined 1

R‐sided stroke 41

L‐sided stroke 19

Bilateral 4

Jobke 2009

21

Group 1

Mean 51.5 years, SD 14.8

Group 2

Mean 47.3 years, SD 13.4

Group 1

M 7

F 1

Group 2

M 6

F 4

Group 1

Mean 89.0 months, SD 59.9

(range 67 to 225 months)

Group 2

Mean 89.4, SD 57.6

(range 40 to 236 months)

NS

Group 1

5 stroke/ischaemia

1 brain injury

1 tumour

1 surgery

Group 2

5 stroke/ischaemia

1 meningitis;

1 injury

3 surgery

NS

Kasten 1998

19

(plus 19 with pre‐chiasmal damage)

Data are presented for full group of 38

Group 1

? Mean 47.7 years, ? SD 12.9

Group 2

? Mean 55.3 years, ? SD 16.2

It is assumed the data presented are mean and SD, but this was not stated

Group 1

M 11

F 8

Group 2

M 13

F 6

Group 1

? Mean 6.8 months, ? SD11.4

Group 2

? Mean 7.2 months, ? SD 6.3

It is assumed the data presented are mean and SD, but this was not stated

NS

19 participants with post chiasmal injury; 10 were due to stroke, 4 due to trauma and 5 due to other reasons

NS

Kasten 2007

23

Group 1

Mean 41.1 years, SD 16.9

Group 2

Mean 39.3 years, SD 10.9

Group 3

Mean 44.3 years, SD 9.1

Group 1

M 5

F 2

Group 2

M 6

F 1

Group 3

M6

F3

Group 1

10 to 83 months, Mean 34.2*, SD 30.1*

Group 2

13 to 477 months, Mean 92.7*, SD 170.6*

Group 3

10 to 143 months, Mean 47.6*, SD 54.4*

NS

Group 1

4 stroke

1 trauma

1 cerebral aneurysmal bleeding

1 hypoxia

Group 2

3 stroke

3 trauma

1 surgery

Group 3

3 stroke

2 trauma

1 surgery

1 hypoxia

2 other

NS

Keller 2010

20

Group 1

Mean 54.7 years. SD 20.4

Group 2

Mean 63.6 years, SD 13.8

Group 1

M 6

F 4

Group 2

M 6

F 4

Group 1

Mean 8.5 weeks, SD 6.7

Group 2

Mean 4.2 weeks, SD 2.1

NS

Group 1

9 vascular

1 tumour

Group 2

9 vascular

1 traumatic

Group 1

4 left hemianopia

3 right hemianopia

1 UL quandrantanopia

1 LL quandrantanopia

1 UR quandrantanopia

Group 2

3 left hemianopia

3 right hemianopia

3 UL quandrantanopia

1 LL quandrantanopia

Modden 2012

45

RT Group: Mean 58.3 ± 11.4 years

CT group: Mean 57.1 ± 8.3 years

OT group: Mean 59.0 ± 11.1 years

RT group:

M 10

F 5

CT group:

M 9

F 6

OT group:

M 7

F 8

RT group:

Mean 4.7 weeks

CT group:

Mean 4.9 weeks

OT group:

Mean 4.3 weeks

"Patients were recruited on average about 4 weeks after their stroke."

NS

RT Group:

occipital 7

temporo‐occipital 2

temporomedial 5

parahippocampal 1

CT Group:

occipital 6

temporo‐occipital 3

temporomedial 5

parahippocampal 1

OT Group*:

occipital 4

temporo‐occipital 3

temporomedial 5

parahippocampal 1

*numbers presented in paper do not add up to 15 (?)

RT group

L stroke 7

R stroke 8

CT group

L stroke 5

R stroke 10

OT group:

L stroke 5

R stroke 10

Plow 2010

12

Mean 59.6 years, SEM 3.5 years

M 5

F 7

Mean 39.8 ± 16.2 months, range 3 to 192 months

NS

Stroke 8 (7 infarct, 1 haemorrhage)

Surgical trauma 2

L‐affected side 4

R‐affected side 8

Poggel 2004

20 participants recruited. Baseline data only available for 19 (data for one dropout not reported)

Group 1

Mean 41.9 years

Range 20 to 67 years

Group 2

Mean 43.2 years

Range 30 to 61 years

Group 1

M 6

F 3

Group 2

M 6

F 4

Group 1

Mean 49.1 months, SEM ?, Range 6.7 to 189.9 months

Group 2

Mean 24.1 months, SEM 5.0, Range 6.8 to 58.3 months

NS

Group 1

vascular 1
infarct 8

cortical and radiations 4

cortical 5

Group 2

vascular 2
infarct 7,

traumatic brain injury 1
cortical and radiation 5
cortical 3

radiation 2

Group 1

L 5/R 4

Group 2

L 5/R 5

Rossi 1990

39

Group 1

Mean 72.6 years, SEM 1.8

Group 2

Mean 63.3 years, SEM 2.5

Group 1

M 10

F 8

Group 2

M 9

F 12

Group 1

Mean 4.4 weeks, SEM 0.3

Group 2

Mean 4.7 weeks, SEM 0.6

NS

Group 1

15 infarct
3 haemorrhage

Group 2

18 infarct
3 haemorrhage

Group 1

16 R/2 L

Group 2

13 R/8 L

Roth 2009

30 participants recruited (data available for 28; 2 dropouts)

Group 1

Mean 60.5 years, SD 11.0, Median 65

Group 2

Mean 60.3 years, SD 11.7,

Median 63

Group 1

4 F

11 M

Group 2

F 7

M 8

Group 1

Mean 39.20 months, SD 54.59, Median 26

Group 2

Mean 87.87 months, SD 186.66, Median16

NS

Group 1

Stroke 11
Haemorrhage 1
Head injury 1
Abscess 1
AVM 1

Group 2

Stroke 11
Haemorrhage 3
Cyst 1

Affected side

Group 1

L 8/R 7

Group 2

L 7/R 8

Rowe 2010

87 participants recruited (full results for 70 participants at 26 weeks)

Group 1

Mean 69.9 years, SD 12.9, median 68.8, IQR, 14.4

Group 2

Mean 70.9 years, SD 11.2, median 72.9, IQR, 15.2

Group 3

Mean 66.2 years, SD 11.3, median 68.2, IQR, 16.2

Group 1

4 F

22 M

Group 2

13 F

17 M

Group 3

9 F

20 M

Group 1

Mean 75.5 days, SD 45.3, median 64.5, IQR 78.0

Group 2

Mean 73.8 days, SD 49.2, median 69.0, IQR 97.0

Group 3

Mean 81.2 days, SD 48.0, median 67.0, IQR 61.0

Barthel Index score

Group 1

Mean 97.5, SD 5.5, median 100.0, IQR 0.0

Group 2

Mean 92.7, SD 11.9, median 100.0, IQR 15.0

Group 3

Mean 93.3, SD 14.7, median 100.0, IQR 5.0

Group 1

25 ischaemic
1 haemorrhage

Group 2

28 ischaemic
2 haemorrhage

Group 3

28 ischaemic
1 haemorrhage

Side of infarct

Group 1

L 9/R 16/bilateral 1

Group 2

L 17/R 13/bilateral 0

Group 3

L 11/R 17/bilateral 1

Schuett 2012

36

Group 1

Mean 64.0 years, SD 11.1, range 44 to 81

Group 2

Mean 63.7 years, SD 13.3, range 42 to 83

Group 1

3 F

15 M

Group 2

3 F

15 M

Group 1

Mean 26.6 weeks, SD 14.5, range 6 to 57

Group 2

Mean 20.1 weeks, SD 18.8, range 4 to 74

NS

Group 1

17 posterior infarction
1 tumour operation

Group 1

17 posterior infarction
1 tumour operation

Side of field loss

Group 1

L 9/R 9

Group 2

L 7/R 11

Spitzyna 2007

22

Age at symptom onset

Group 1

Range 5 to 67 years, mean 42.5*, SD 20.5*

Group 2

Range 39 to 78 years, mean 63.1*, SD 12.2*

Group 1

M 6

F 5

Group 2

M 7

F 1

Time since symptoms onset

Group 1

Range 1 to 37 years, mean 7.5*, SD 10.9*

Group 2

Range 3 months to 5 years, mean 1.6*, SD 1.7*

NS

Group 1

3 infarct

1 tuberous sclerosis

2 traumatic brain injury, 2 tumour

2 haemorrhage

1 cyst

Group 2

8 infarct

Group 1

All R

Group 2

All R

Szlyk 2005

10

Group 1

Range 16 to 74 years, mean 50.6*, SD 22.5*

Group 2

Range 34 to 73 years, mean 54.0, SD 14.4

Group 1

5 M

Group 2

5 M

NS

NS

Group 1

4 CVA

1 tumour: all occipital lobe

Group 2

4 CVA

1 AVM: all occipital lobe

Group 1

L 3/R 2

Group 2

L 4/R 1

* Figures calculated from raw data supplied in papers
AVM: arteriovenous malformation
CVA: cerebrovascular accident
CT: compensatory training
F: female
IQR: interquartile range
L: left
LL: lower left
M: male
NS: not stated
OT: occupational therapy
R: right
RT: restitutive training
SD: standard deviation
SEM: standard error of the mean
UL: upper left
UR: upper right

Figures and Tables -
Table 2. Demographics of included studies: demographics of included participants
Navigate to table in Review
Table 3. Demographics of included studies: visual problems of included participants

Study

Methods of visual field assessment

Type/extent of field loss

Macular sparing

Presence of neglect?

Aimola 2011

Unspecified kinetic perimeter

Esterman measures of static superthreshold

Group 1

Hemianopia 20, quadrantanopia 8

Group 2

Hemianopia 20, quadrantanopia 4

Group 1

Mean 1.92° (SD 1.44)

Group 2

Mean 2.45° (SD 1.85)

Yes: stated "Three patients (2 in the intervention group, 1 in the control group) had comorbid neglect as confirmed with the bells test".

Bainbridge 1994

Harrington Flocks Visual Screener

Confrontation

Not stated

Not stated

Yes: no details of inclusion criteria or participants provided, but objective stated "To study the effect of ... on visual neglect or hemianopsia following stroke".

Bowers 2014

Goldmann perimetry

Not stated

Not stated

No: stated "no visual neglect". Visual neglect diagnosed with Bells test and Schenkenberg Line Bisection Test.

De Haan 2015

Goldmann perimetry

Training group

Functional field score 58 ± 7.8

Quadrantanopia 5 (3 lower left, 1 upper left, 1 lower right)

Hemianopia 21

Control group

Functional field score 64 ± 11.4

Quadrantanopia 5 (3 lower left, 2 upper left)

Hemianopia 18

Not stated

No: stated "Neglect was excluded based on the Balloons, drawings, Line Bisection and Rey Complex Figure Test."

Elshout 2016

Goldman perimetry

Humphrey perimetry

Right field loss: hemifield 4, incomplete hemifield 5, quadrant 2, scotoma 1

Left field loss: hemifield 2, incomplete hemifield 9, quadrant 1, scotoma 2

Bilaterial field loss

Incomplete: 1

"All subjects
had macular sparing of at least 2°"

No: patients with visual neglect were excluded (based on line bisection test)

Gall 2013

Standard automated perimetry

Not stated

Not stated

Not stated

Jarvis 2012

Confrontation

Ocular diagnosis:

low vision 30

visual field loss 38

eye movement deficit 41

perceptual impairment 24

("Note: patients may have had an isolated visual impairment or combined visual deficits")

Not stated

Yes: all patients with a "post‐stroke visual impairment were eligible for inclusion".

Jobke 2009

Standard automated perimetry

High resolution perimetry (HRP)

NB: It did not state whether participants had visual neglect or whether this was diagnosed.

Group 1

2 diffuse, 2 full homonymous hemianopia, 1 partial homonymous hemianopia, 1 full quadrantanopia

2 partial quadrantanopia

Group 2

4 diffuse, 2 full homonymous hemianopia, 2 partial homonymous hemianopia, 1 full quadrantanopia,

1 partial quadrantanopia

Group 1

7 sparing, 1 not sparing

Group 2

10 sparing

Not stated

Kasten 1998

Tubinger automated perimetry (TAP)

High resolution perimetry (HRP)

NB: data were presented for full group of 38 participants (including participants in parallel trial)

Group 1

TAP 90° ‐ border position, mean 3.51° (degrees of visual angle from zero vertical meridian), SEM 1.0

TAP 90° ‐ number of misses, mean 53.0, SEM 9.1

Group 2

TAP 90° ‐ border position, mean 3.43° (degrees of visual angle from zero vertical meridian), SEM 0.99

TAP 90° ‐ number of misses, mean 69.2, SEM 11.2

Not stated

No: participants with neglect were excluded. Method of diagnosis of neglect not stated.

Kasten 2007

Tubinger automated perimetry (TAP)

High resolution perimetry (HRP)

TAP 90° (number of blind stimuli positions)

Group 1

Right eye ‐ mean 46.6, SD 6.9, left eye ‐ mean 43.9, SD 3.7

Group 2

Right eye ‐ mean 50.3, SD 8.7, left eye ‐ mean 43.1, SD 7.6

Group 3

Right eye ‐ mean 32.9, SD 6.8, left eye ‐ mean 37.9, SD 7.1

Not stated

No: participants with neglect were excluded. Method of diagnosis of neglect not stated

.Keller 2010

Goldmann perimetry

Goldmann suprathreshold

Group 1

4 left hemianopia

3 right hemianopia

1 UL quandrantanopia

1 LL quandrantanopia

1 UR quandrantanopia

Group 2

3 left hemianopia

3 right hemianopia

3 UL quandrantanopia

1 LL quandrantanopia

Group 1

6 with 0° macular sparing

4 with < 5° macular sparing

Group 2

6 with 0° macular sparing

4 with < 5° macular sparing

No: participants with neglect were excluded. 3 neglect tests were used: "line bisection, Mesulam test, draw a clock face test".

Modden 2012

Visual field assessment from the Test Battery of Attentional Performance

RT Group

10 hemianopia

5 quadrantanopia

TAP alertness without cueing, ms; mean 304.2, SD 80.8

TAP conjunction search, omissions; mean 9.1, SD 9.0

CT Group

12 hemianopia

3 quadrantanopia

TAP alertness without cueing, ms; mean 383.7, SD 205.2

TAP conjunction search, omissions; mean 10.7, SD 6.7

OT Group

10 hemianopia

5 quadrantanopia

TAP alertness without cueing, ms; mean 308.1, SD 58.6

TAP conjunction search, omissions; mean 10.3, SD 5.6

RT Group

3/15 participants with less than 2° sparing

CT Group

3/15 participants with less than 2° sparing plus 1 participant with no sparing

OT Group

3/15 participants with less than 2° sparing

No: participants with neglect were excluded. Method of diagnosis of neglect not stated.

Plow 2010

Subjective topographic measure of perceived visual field defect

High resolution perimetry (HRP)

7 hemianopia

5 quadrantanopia

Not stated

Not stated

Poggel 2004

Tubinger automated perimetry (TAP)

High resolution campimetry

High resolution perimetry (HRP)

Group 1

Upper attention field (size of area of residual vision, %), mean 18.2, SEM 4.0

Lower probe field (size of area of residual vision, %), mean 21.3, SEM 3.1

Total visual field (size of area of residual vision, %), mean 7.3, SEM 1.9

Group 2

Upper attention field (size of area of residual vision, %), mean 16.9, SEM 2.4

Lower probe field (size of area of residual vision, %), mean 15.5, SEM 4.0

Total visual field (size of area of residual vision, %), mean 6.7, SEM 1.3

Not stated

No: participants with neglect were excluded. Method of diagnosis of neglect not stated.

Rossi 1990

Harrington Flocks Visual Screener

Tangent screen measures

Group 1

Homonymous hemianopia 12

(Visual neglect 6)

Group 2

Homonymous hemianopia 15

(Visual neglect 6)

Not stated

Yes: participants with "homonymous hemianopia or visual neglect were recruited ....". Method of diagnosis of neglect was Harrington Flocks Visual Screener.

39 participants recruited: 27 had homonymous hemianopia; 12 had visual neglect.

Roth 2009

Tubinger automated perimetry (TAP)

Scanning laser ophthalmoscopy

Group 1

Homonymous hemianopia 12, quadrantanopia 3

Group 2

Homonymous hemianopia 12, quadrantanopia 3

Not stated

No: participants with neglect were excluded. Method of diagnosis of neglect was clock‐drawing and line‐bisection tests.

Rowe 2010

Goldmann perimetry

Esterman measures of static superthreshold

Group 1

Homonymous hemianopia left partial 8, Homonymous hemianopia right partial 3, Homonymous hemianopia left complete 9, Homonymous hemianopia right complete 6

Group 2

Homonymous hemianopia left partial 5, Homonymous hemianopia right partial 9, Homonymous hemianopia left complete 8, Homonymous hemianopia right complete 8

Group 3

Homonymous hemianopia left partial 8, Homonymous hemianopia right partial 5, Homonymous hemianopia left complete 10, Homonymous hemianopia right complete 6

Not stated

No: participants with neglect were excluded. Method of diagnosis was clinical assessment: "as assessed by the orthoptist".

Schuett 2012

Tubingen kinetic perimetry

Group 1

Hemianopia 15, quadranopia 1, paracentral scotoma 2

Group 2

Hemianopia 10, quadranopia 4, paracentral scotoma 4

Group 1

Mean 2.3° (SD 1.4)

Group 2

Mean 2.3° (SD 1.2)

No: participants with neglect were excluded. Method of diagnosis described as: "as assessed by tests in accordance with the Behavioural Inattention Test (line bisection, letter and star cancellation, figure and shape copying, drawing from memory; Halligan et al, 1991)."

Spitzyna 2007

Goldmann perimetry

Humphrey automated perimetry

Group 1

Full homonymous hemianopia 8, partial homonymous hemianopia 1, lower quadrantanopia 1, upper quadrantanopia 1

Group 2

Full homonymous hemianopia 6, lower quadrantanopia 1, upper quadrantanopia 1

Macular sparing defined as 2° of sparing

Group 1

Sparing 5, non‐sparing 6

Group 2

Sparing 3, non sparing 5

No: only participants with right‐sided homonymous hemianopic were included; therefore, presence of neglect was assumed unlikely.

Szlyk 2005

Goldmann perimetry

Group 1

Goldmann III4e, range 45.2 to 125, mean 59.12*, SD 22.07*

Goldmann V4e, range 48.8 to 115, mean 70.56*, SD 26.15*

Group 2

Goldmann III4e, range 46.8 to 123.8, mean 68.0*, SD 31.71*

Goldmann V4e, range 50.67 to 132, mean 73.73*, SD 33.10*

Figures were calculated for the affected side only.

Not stated

Not stated; however, although it was not stated whether the participants may have had visual neglect, neglect is unlikely in occipital lesions, and only participants with occipital lesions were included.

* Figures calculated from raw data supplied in papers
combined: combined etiology
HRP: high resolution perimetry
LL: lower left
LR: lower right
ms: milliseconds
SD: standard deviation
SEM: standard error of the mean
TAP: Tübingen automated perimeter
UL: upper left
UR: upper right

Figures and Tables -
Table 3. Demographics of included studies: visual problems of included participants
Navigate to table in Review
Table 4. Outcome measures within included studies

Study

Functional ability in ADL

Visual field

Outcome category (measure)

Functional ability in EADL

Reading

Falls

Quality of life

Visual scanning

Adverse events

Other

Outcomes with data included within meta‐analyses

Aimola 2011

Kinetic Perimetry (unspecified kinetic perimeter)

Static Superthreshold (Esterman measures of static superthreshold)

(NB not clear if recorded as outcome or not; no results provided for visual field data)

Reading (corrected reading speed)

1. VFQ 25
2. VIQ ‐ Visual Impairments questionnaire
3. Subjective Reasons questionnaire

1. visual search ‐ find the number (computer ‐based)
2. visuomotor search ‐ find items on a shelf

Tasks simulating ADL ‐ 1. driving hazard perception (mean score per hazard), 2. obstacle avoidance (completion time), 3. visuomotor search (time)

Attention tasks ‐ 1. sustained attention to response (mean percentage error score), 2. test of everyday attention

Reading: Analysis 2.3

Visual search: time to complete Analysis 2.5

QoL: data not included as only available for individual questionnaire items

Bainbridge 1994

Gross visual screening (Harrington‐Flocks Visual Field Score)

Line Cancellation Test

Motor Free Visual Perception Score

Line Bisection Test

No data included in meta‐analyses (as no control group). See Table 5

Bowers 2014

Mobility questionnaire

Question: "If the study were to end today, would you want to continue with these prism glasses (i.e. the prism glasses worn in that period)?"

Functional ability in EADL: Analysis 3.3

De Haan 2015

Kinetic perimetry (Goldmann Perimetry, Functional Field score)

Independent Mobility questionnaire

1. Radner reading test; (a) Radner average reading speed (wpm), (b) minimal readable text size (LogRad)

2. Text reading test; (a) text reading speed (wpm), (b) text correct answers

1. NEI‐VFQ‐25 (Visual Functional questionnaire)
2. Cerebral Visual Disorders questionnaire

1. visual scanning ‐ dots test
2. visual search ‐ letters (parallel search test)
3. visual search ‐ letters (serial search)

Not reported as an outcome measure, but stated no adverse events in either group

Visual acuity, contrast sensitivity, hazard perception, simulating driving/tracking task, obstacle course

Visual field: Functional Field score Analysis 2.1

Reading: Radner average reading speed Analysis 2.3

Visual scanning: Parallel search test, time Analysis 2.5

QoL: NEI VFQ Analysis 2.4

Functional ability in EADL: mobility questionnaire Analysis 2.2

Elshout 2016

Goldman perimetry

Humphrey perimetry

Reading speed (words per minute) ‐ 15 point Arial font, 88 and 165 words

No data included (as data not available for before the cross‐over)

Gall 2013

Static Threshold Perimetry (Standard automated perimetry)

1. NEI VFQ 39 (vision‐related)
2. SF‐12 (health‐related)

No data included in meta‐analyses (as no suitable data presented in abstract)

Jarvis 2012

FIM

1. Functional mobility (timed walk)

2. Non‐validated questionnaire giving qualitative information
about their treatment approach

Functional ability in ADL: Functional Independence Measure Analysis 4.1

Jobke 2009

Static Threshold Perimetry (Standard automated perimetry)

Resolution Perimetry (High resolution perimetry)

Radner reading test

NEI VFQ

Zahlen‐Verbindungs
test (ZVT) for measuring the speed of connecting numbers in
a paper‐pencil test

No data included in meta‐analyses (as no control group). See Table 5 (for data available before the cross‐over)

Kasten 1998

Resolution Perimetry (High resolution perimetry)

Static Threshold Perimetry (Tubinger automated perimetry)

Quality of life questionnaire

Visual acuity: Landolt ring to give minimum angle of resolution

Visual field: Tubinger automated perimetry: border position in degrees of visual angle from zero vertical meridian Analysis 1.1

Quality of life Analysis 1.2

Kasten 2007

Resolution Perimetry (High resolution perimetry: number of hits, learning effects, fixation ability, false hits)

Static Threshold Perimetry (Tubinger automated perimetry: no of hits, fixation ability)

Subjective visual ability questionnaire

1. Eye movements: "Chronos Vision Eye Tracker"

2. Visual acuity

3. "Zahlen‐Verbindungs Test" of visuo‐spatial attention

4. "Alters‐Konzentrationstest" attention test for older people

5. "testbatterie zur Aufmerksamkeitspruefung" ability to improve attention

No data included in meta‐analyses (as no control group). See Table 5 (for available data comparing group outcomes)

Keller 2010

Kinetic Perimetry (Goldmann perimetry)

Static Superthreshold (Goldmann suprathreshold)

Reading time (standardised reading test)

OT administered questionnaire (based on Kerkhoff's self‐evaluation of ADL)

1. Visual exploration test (number of omissions)

2. Search task (search time)

Electro‐oculography

No data included in meta‐analyses (as no control group). See Table 5

Modden 2012

Extended Barthel Index (German)

Gross Visual Screening (Test Battery of Attentional Performance:

visual field assessment)

Reading ‐ Weschler memory tests (errors)

1. Visual scan: from the Test Battery of Attentional Performance
2. Visual search: cancellation tasks from Behavioural Inattention Test (BIT) (omissions)

Attention: Test Battery of Attentional Performance (alertness)

No data included in meta‐analyses (as no control group). See Table 5

Plow 2010

Resolution Perimetry (High resolution perimetry: position of visual field border and stimulus detection accuracy)

Gross Visual Screening (subjective topographic
measure of perceived visual field deficit)

Impact of Vision Impairment (IVI) profile

Low Vision‐
Visual Functional Questionnaire (LV‐VFQ)

Measure of fixation performance

No data included in meta‐analyses (as no control group). See Table 5

Poggel 2004

Static Threshold Perimetry (Tubinger automated perimetry (TAP))

Static Superthreshold (High resolution campimetry)

Resolution Perimetry (High resolution perimetry (HRP))

No data included in meta‐analyses (as no control group). See Table 5

Rossi 1990

Barthel Index

Gross Visual Screening (Harrington Flocks Visual Screener)

Static Superthreshold (Tangent screen examination)

Number of falls

Line cancellation task

Modified Mini Mental Status Examination,

Motor Free Visual Perceptual Test,

Line Bisection Task

ADL: Barthel Index: Analysis 3.1

Visual Field: Analysis 3.2

Falls: number of falls Analysis 3.5

Visual scanning: cancellation Analysis 3.7

Roth 2009

Static Threshold Perimetry (Tubinger automated perimetry (TAP))

Resolution Perimetry (Scanning laser ophthalmoscopy)

Reading speed

QoL: World Health Organisation questionnaire WHOQOL‐BREF

1. Digit search task (response time)

2. Natural search task (table test)(response time)

3. Natural scene exploration

4. Fixation stability (video eye tracker)

No data included in meta‐analyses (as no control group). See Table 5

Rowe 2010

Kinetic Perimetry (Goldmann perimetry)

Static Superthreshold (Esterman measures of static superthreshold)

Nottingham extended activities of daily living (NEADL)

Reading ability (Radner test)

1. VFQ 25‐10 (vision related)
2. EQ‐5D
3. SF‐12

Number of participants and number of adverse events

Rivermead Mobility Index

Visual Field: relative change in visual field area Analysis 2.1 and Analysis 3.2

QoL: VFQ 25‐10 Analysis 2.4 and Analysis 3.6

Adverse events:Analysis 2.6; Analysis 3.8

Schuett 2012

Kinetic Perimetry (Tubingen kinetic perimetry) (NB not recorded immediately after first phase)

Reading (speed and errors)

Cancellation (speed and errors)

No data included in meta‐analyses (as no control group). See Table 5

Spitzyna 2007

Kinetic Perimetry (Goldmann perimetry)

Static Threshold Perimetry (Humphrey 10‐2 central threshold programme)

(NB not recorded immediately after first phase)

1. Text reading speeds

2. Single word reading speeds

Eye movement characteristics:

‐ spatial characteristics of saccadic amplitude, incoming saccade amplitude and landing position

‐ temporal characteristics

Reading (text reading speed): Analysis 2.3

Visual field ‐data not included as not collected before cross‐over.

Szlyk 2005

Kinetic Perimetry (Goldman Perimetry)

Indoor functional assessment

Outdoor functional assessment

Driving skills assessment

Psychophysical assessment

Satisfaction

Prisms use at 2 years

No data included in meta‐analyses(as no control group).

No data reported for before the cross‐over.

ADL: activities of daily living
BIT: behavioural inattention test
EADL: extended activities of daily living
EQ‐5D:standardised EuroQol health‐related quality of life instrument
FIM: Functional Independence Measure
HRP: high resolution perimetry
IVI: impact of vision impairment
LogRad: a scale of reading acuity
LV‐VFQ: low vision functional questionnaire
NB: note
NEADL: Nottingham Extended Activities of Daily Living
NEI‐VFQ‐25: National Eye Institution Visual Function Questionnaire
OT: occupational therapy
QoL: quality of life
SF‐12: 12‐Item Short Form Health Survey
TAP: tubinger automated perimetry
VFQ: visual function questionnaire
VIQ: visual impairment questionnaire
WHOQOL‐BREF: World Health Organisation quality of life questionnaire
wpm: words per minute
ZVT: Zahlen‐Verbindungs test

Figures and Tables -
Table 4. Outcome measures within included studies
Navigate to table in Review
Table 5. Results of studies comparing two similar active interventions (i.e. two interventions from the same category)

Study

Interventions

Outcome

Mean (or other reported result if no mean available)

Standard deviation

Number of participants

Statistical test/results

Restitution: one restitution intervention versus another restitution intervention

Jobke 2009

Extrastriate VRT

Visual field (high‐resolution perimetry, HRP)

increase from baseline of 5.9% (percentage of HRP hits)

8

significant increase: t = ‐5.262, P = 0.0005

Standard VRT

increase from baseline of 2.9% (percentage of HRP hits)

10

significant increase: t = ‐2.373, P = 0.021

Kasten 2007

Parallel co‐stimulation

Visual field (high‐resolution perimetry)

increase of 2.4% detected stimuli

7

No significant difference "confirmed by nonparametric Kruskal–Wallis ANOVA"

Moving co‐stimulation

increase of 6.5% detected stimuli

7

Single stimulus

increase of 3.9% detected stimuli

9

Plow 2010

VRT + tDCS

Visual field (high‐resolution perimetry)

shift from baseline to post‐test from
4.11° ± 1.50° to 8.37° ± 2.29°, Wilcoxon signed‐rank test = 0, P = 0.068

4

Mann‐Whitney U = 0, P = 0.021 (significantly greater shift in the visual field border with VRT + tDCS than VRT alone)

VRT + sham tDCS

shift from baseline to post‐test from 6.33° ± 2.59° to 7.03° ± 2.51°, Wilcoxon signed‐rank test = 1, P = 0.144

4

VRT + tDCS

Functional ability in ADL (LV‐VFQ)

shift from baseline to post‐test from 32.25
± 5.30 to 28.25 ± 5.07; Wilcoxon signed‐rank test = 0; P = 0.068

4

Mann‐Whitney U = 5.5; P = 0.468 (non‐significant)

VRT + sham tDCS

shift from baseline to post‐test from 28 ± 2.34 to 25.25 ± 1.11; Wilcoxon signed‐rank test = 1; P = 0.285

4

VRT + tDCS

Functional ability in ADL (LV‐VFQ) ‐ 6‐month follow‐up

29.00

3.58

5

Wilcoxon signed‐rank test = 4; P = 0.343

VRT + sham tDCS

26.80

2.11

VRT + tDCS

Quality of life ‐ 6‐month follow‐up

23.20

7.83

5

Wilcoxon
signed‐rank test = 2.5; P = 0.357

VRT + sham tDCS

16.8

4.62

Poggel 2004

VRT + attentional cueing

Visual field (high‐resolution perimetry) ‐ percentage improvement, attention field

8.3

SEM 1.5

9

P = 0.001 (in favour of attentional cueing)

VRT with no attentional cueing

2.9

SEM 0.8

10

Compensation: one compensation intervention versus another compensation intervention

Schuett 2012

Visual exploration training

Reading speed

105.3

33.8

18

Not reported; calculated as MD ‐19.30 (‐43.32 to 4.72) (see Figure 3)

Reading training

124.6

39.5

18

Visual exploration training

Cancellation test (exploration time)

18.5

4.9

18

Not reported; calculated as MD 18.30 (14.28 to 22.32) (see Figure 3)

Reading training

36.8

7.2

18

Keller 2010

Audiovisual exploration training (AVT)

Functional ability in ADL (ADL test total score)

1.5

(SE displayed on graph only)

10

ANOVA P = 0.036 (in favour of AVT)

Visual exploration training

5.0

(SE displayed on graph only)

10

Audiovisual exploration training (AVT)

Reading time (seconds)

75

(SE displayed on graph only)

10

ANOVA P = 0.03 (in favour of AVT)

Visual exploration training

178

(SE displayed on graph only)

10

Audiovisual exploration training (AVT)

Visual scanning (percentage hits)

85.3

(SE displayed on graph only)

10

ANOVA P = 0.01 (in favour of AVT)

Visual exploration training

64.1

(SE displayed on graph only)

10

Modden 2012

Computer‐based compensation therapy

Visual field enlargement (visual field assessment from Test Battery of Attentional Performance)

2.9

4.0

15

Pre‐ to post‐treatment significant field expansion (P = 0.013)

Standard occupational therapy (compensation)

1.3

4.7

15

Pre‐ to post‐treatment: no significant field expansion (P = 0.316)

Computer‐based compensation therapy (CT)

Functional ability in ADL (improvement in Extended Barthel Index)

3.3

3.6

15

"No significant treatment effects were found when comparing ... CT/OT".

Standard occupational therapy (OT) (compensation)

1.8

2.0

15

Computer‐based compensation therapy (CT)

Reading ‐ Improvement in reading performance, reduction in number of errors (from baseline)

‐0.9

1.1

15

"Compared with OT"... "CT did not significantly reduce reading errors."

Standard occupational therapy (OT) (compensation)

‐0.7

1.0

15

Computer‐based compensation therapy (CT)

Visual scanning ‐ reduction in number of omissions from baseline, cancellation tasks of the Test Battery of Attentional Performance

‐5.4

5.2

15

"Compared with OT"... "CT did not result in superior improvements".

Standard occupational therapy (OT) (compensation)

‐2.3

5

15

Substitution: one substitution intervention versus another substitution intervention

Bainbridge 1994

Full‐field Fresnel Prisms

Visual Field (Harrington Flocks Visual Field Score)

2.9

2

10

States full‐field more improved

Hemi‐field Fresnel Prisms

7.2

3

8

Full‐field Fresnel Prisms

Scanning (Line cancellation test errors)

4.7

1.3

10

P < 0.01, Student's t‐test (in favour of full‐field prisms)

Hemi‐field Fresnel Prisms

0.3

0.6

8

Szlyk 2005

18.5 dioptre Gottlieb Visual field awareness system prisms

Visual skills category assessment battery

"There was improvement within all categories with both of the prism systems ranging from 36% for mobility (with the Fresnel prisms) to 13% for recognition (with the Gottlieb VFAS)."

10 (data only available for after cross‐over)

"There were no statistically significant differences between improvements
with the Gottlieb VFAS compared with the
Fresnel prisms."

Press‐on TM 20 Diopter Fresnel prisms

ADL: activities of daily living
ANOVA: analysis of variance (statistical test of)
AVT: audiovisual exploration training
CT: compensation therapy
HRP: high‐resolution perimetry
LV‐VFQ: Low Vision Visual Functioning Questionnaire
MD: mean difference
OT: occupational therapy
SE: standard error
SEM: standard error of the mean
tDCS: transcranial direct current stimulation
VFAS: visual field awareness system
VRT: visual restitution therapy

Figures and Tables -
Table 5. Results of studies comparing two similar active interventions (i.e. two interventions from the same category)
Navigate to table in Review
Table 6. Results of studies comparing two different types of active interventions (i.e. interventions from different categories)

Study

Interventions

Outcome

Mean (or other reported result if no mean available)

Standard deviation

Number of participants

Statistical test/results

Compensation intervention versus restitution intervention

Modden 2012

Computer‐based restitution therapy

Visual field enlargement (visual field assessment from Test Battery of Attentional Performance)

3.9

4.9

15

Pre‐ to post‐treatment significant field expansion (P = 0.003)

Computer‐based compensation therapy

2.9

4.0

15

Pre‐ to post‐treatment significant field expansion (P = 0.013)

Computer‐based restitution therapy (RT)

Functional ability in ADL (improvement in Extended Barthel Index)

1.5

2.8

15

"No significant treatment effects were found when comparing ... RT/CT".

Computer‐based compensation therapy (CT)

3.3

3.6

15

Computer‐based restitution therapy (RT)

Reading: improvement in reading performance, reduction in number of errors (from baseline)

‐0.9

2.4

15

"There were no differences between RT and CT."

Computer‐based compensation therapy (CT)

‐0.9

1.1

15

Computer‐based restitution therapy (RT)

Visual scanning: reduction in number of omissions from baseline, cancellation tasks of the Test Battery of Attentional Performance

‐5.3

10.5

15

"... the improvement of the CT compared with the RT group did not meet the defined significance level after Bonferroni correction (P = .023)."

Computer‐based compensation therapy (CT)

‐5.4

5.2

15

Roth 2009

Explorative scanning training (EST) (compensation)

Visual field: Tubingen automated perimetry

44.4

13.1

15

"Neither the EST group nor the FT group showed any differences in their TAP or SLO outcomes, quantified as the total number of stimuli detected in the blind hemifield (lowest P = 0.204)."

Flicker stimulation training (FT)(restitution)

35.7

15.2

13

Explorative scanning training (EST) (compensation)

Quality of life (WHOQOL‐BREF)

12.93

1.67

15

"The EST group reported greater improvements (T2 minus T1 scores) in the WHOQOL social‐relationships domain (t test; t(20) = 2.217, P = 0.038)" (but no significant differences for other domains).

Flicker stimulation training (FT) (restitution)

13.23

1.3

13

Explorative scanning training (EST) (compensation)

Reading (reading speed)

99.7

34.7

15

"Although the EST and FT groups differed in their reading speeds at T1, this difference remained unchanged [main effect of group, F(1,26) = 133.074, P < 0.0001, interaction, F < 1]".

Flicker stimulation training (FT)(restitution)

140.2

20.9

13

Compensation intervention versus substitution intervention

Rowe 2010*

Fresnel prisms (substitution)

Visual Field (relative change in visual field area)

0.052

0.1396

24

ANOVA results: no significant differences between groups (P = 0.55, for comparison across 3 treatment groups)

Visual search training (compensation)

0.0815

0.1488

24

Fresnel prisms (substitution)

Extended activities of daily living (change in EADL from baseline)

15.2

4.8

22

"No evidence of differences ..."

Visual search training (compensation)

15.2

4.4

22

Fresnel prisms (substitution)

Reading (change in Radner reading speed)

17.4

21.3

24

"No evidence of differences ..."

Visual search training (compensation)

13.0

13.1

25

Fresnel prisms (substitution)

Quality of life (VFQ‐25 total score)

68.2

18.4

24

"Visual function (using the VFQ 25‐10) improved at 26 weeks in the visual search training arm (60 [SD 19] to 68.4 [SD 20]) when compared to the Fresnel prisms (68.5 [SD 16.4] to 68.2 [18.4]) and standard care arms (63.7 [SD 19.4] to 59.8 [SD 22.7]: Table 6, ANCOVA P = 0.05)."

Visual search training (compensation)

68.4

20.0

25

Fresnel prisms (substitution)

Adverse events (number of participants with reported adverse events during study)

18 participants

26

"Given the extent and range of adverse events reported with prism wear, caution must be exercised if prescribing prism glasses as an intervention for homonymous hemianopia."

Visual search training (compensation)

2 participants

30

*Rowe 2010 also had a control (standard care) group, and data were included in relevant meta‐analyses for compensatory and substitution interventions versus control.
ADL: activities of daily living
ANCOVA: analysis of covariance (statistical test of)
ANOVA: analysis of variance (statistical test of)
CT: compensation therapy
EADL: extended activities of daily living
EST: explorative scanning training
FT: flicker stimulation training
RT: restitution therapy
SD: standard deviation
SLO: Scanning Laser Ophthalmoscope
T1: outcome asssessment timepoint 1
T2: outcome assessment timepoint 2
TAP: Tuebingen automated perimetry
VFQ: visual function questionnaire
WHOQOL‐BREF: World Health Organization Quality of Life Instrument

Figures and Tables -
Table 6. Results of studies comparing two different types of active interventions (i.e. interventions from different categories)
Navigate to table in Review
Comparison 1. Restitutive interventions versus control, placebo or no intervention

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1.1 Visual field Show forest plot

1

19

Mean Difference (IV, Random, 95% CI)

1.02 [‐1.37, 3.41]

1.2 Quality of life Show forest plot

1

30

Odds Ratio (M‐H, Random, 95% CI)

13.00 [2.07, 81.48]
Figures and Tables -
Comparison 1. Restitutive interventions versus control, placebo or no intervention
Navigate to table in Review
Comparison 2. Compensative interventions versus control, placebo or no intervention

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

2.1 Visual field Show forest plot

2

95

Std. Mean Difference (IV, Random, 95% CI)

‐0.11 [‐0.92, 0.70]

2.1.1 Participants with visual field defects (no visual neglect)

2

95

Std. Mean Difference (IV, Random, 95% CI)

‐0.11 [‐0.92, 0.70]

2.1.2 Participants with visual field defects and (possibly) co‐existing visual neglect

0

0

Std. Mean Difference (IV, Random, 95% CI)

Not estimable

2.2 Extended activities of daily living Show forest plot

2

87

Std. Mean Difference (IV, Random, 95% CI)

0.49 [‐0.01, 0.99]

2.2.1 Participants with visual field defects (no visual neglect)

2

87

Std. Mean Difference (IV, Random, 95% CI)

0.49 [‐0.01, 0.99]

2.2.2 Participants with visual field defects and (possibly) co‐existing visual neglect

0

0

Std. Mean Difference (IV, Random, 95% CI)

Not estimable

2.3 Reading Show forest plot

4

162

Std. Mean Difference (IV, Random, 95% CI)

0.26 [‐0.05, 0.58]

2.3.1 Participants with visual field defects (no visual neglect)

3

110

Std. Mean Difference (IV, Random, 95% CI)

0.18 [‐0.20, 0.56]

2.3.2 Participants with visual field defects and (possibly) co‐existing visual neglect

1

52

Std. Mean Difference (IV, Random, 95% CI)

0.45 [‐0.10, 1.00]

2.4 Quality of life Show forest plot

2

96

Mean Difference (IV, Random, 95% CI)

9.36 [3.10, 15.62]

2.4.1 Participants with visual field defects (no visual neglect)

2

96

Mean Difference (IV, Random, 95% CI)

9.36 [3.10, 15.62]

2.4.2 Participants with visual field defects and (possibly) co‐existing visual neglect

0

0

Mean Difference (IV, Random, 95% CI)

Not estimable

2.5 Scanning ‐ cancellation Show forest plot

2

97

Std. Mean Difference (IV, Random, 95% CI)

‐0.01 [‐0.40, 0.39]

2.5.1 Participants with visual field defects (no visual neglect)

1

48

Std. Mean Difference (IV, Random, 95% CI)

0.12 [‐0.45, 0.68]

2.5.2 Participants with visual field defects and (possibly) visual neglect

1

49

Std. Mean Difference (IV, Random, 95% CI)

‐0.13 [‐0.69, 0.44]

2.6 Adverse events Show forest plot

2

108

Odds Ratio (M‐H, Random, 95% CI)

5.18 [0.24, 112.57]
Figures and Tables -
Comparison 2. Compensative interventions versus control, placebo or no intervention
Navigate to table in Review
Comparison 3. Substitutive interventions versus control, placebo or no intervention

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

3.1 Functional Activities of Daily Living Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Subtotals only

3.1.1 Participants not wearning prisms during assessment

0

0

Mean Difference (IV, Random, 95% CI)

Not estimable

3.1.2 Participants wearing prisms during assessment

1

39

Mean Difference (IV, Random, 95% CI)

‐4.00 [‐17.86, 9.86]

3.2 Visual field Show forest plot

2

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

3.2.1 Participants not wearing prisms during assessment

1

46

Std. Mean Difference (IV, Random, 95% CI)

0.12 [‐0.46, 0.70]

3.2.2 Participants wearing prisms during assessment

1

39

Std. Mean Difference (IV, Random, 95% CI)

1.12 [0.44, 1.80]

3.3 Extended activities of daily living Show forest plot

2

Std. Mean Difference (IV, Random, 95% CI)

Subtotals only

3.3.1 Participants not wearing prisms during assessment

1

38

Std. Mean Difference (IV, Random, 95% CI)

0.20 [‐0.44, 0.85]

3.3.2 Participants wearing prisms during assessment

1

61

Std. Mean Difference (IV, Random, 95% CI)

0.24 [‐0.26, 0.75]

3.4 Reading Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Subtotals only

3.4.1 Participants not wearing prisms during assessment

1

45

Mean Difference (IV, Random, 95% CI)

2.80 [‐7.13, 12.73]

3.4.2 Participants wearing prisms during assessment

0

0

Mean Difference (IV, Random, 95% CI)

Not estimable

3.5 Falls Show forest plot

1

Odds Ratio (M‐H, Random, 95% CI)

Subtotals only

3.5.1 Participants not wearing prisms during assessment

0

0

Odds Ratio (M‐H, Random, 95% CI)

Not estimable

3.5.2 Participants wearing prisms during assessment

1

39

Odds Ratio (M‐H, Random, 95% CI)

1.21 [0.26, 5.76]

3.6 Quality of life Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Subtotals only

3.6.1 Participants not wearing prisms during assessment

1

43

Mean Difference (IV, Random, 95% CI)

8.40 [‐4.18, 20.98]

3.6.2 Participants wearing prisms during assessment

0

0

Mean Difference (IV, Random, 95% CI)

Not estimable

3.7 Scanning ‐ cancellation Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Subtotals only

3.7.1 Participants not wearing prisms during assessment

0

0

Mean Difference (IV, Random, 95% CI)

Not estimable

3.7.2 Participants wearing prisms during assessment

1

39

Mean Difference (IV, Random, 95% CI)

9.80 [1.91, 17.69]

3.8 Adverse events Show forest plot

1

59

Odds Ratio (M‐H, Random, 95% CI)

87.32 [4.87, 1564.66]
Figures and Tables -
Comparison 3. Substitutive interventions versus control, placebo or no intervention
Navigate to table in Review
Comparison 4. Assessment or screening versus control, placebo or no intervention

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

4.1 ADL Show forest plot

1

37

Mean Difference (IV, Random, 95% CI)

‐6.97 [‐23.78, 9.84]
Figures and Tables -
Comparison 4. Assessment or screening versus control, placebo or no intervention
Navigate to table in Review