Interventions for people with perceptual disorders after stroke: the PIONEER scoping review, Cochrane systematic review and priority setting project

What is perception?

Perception is the brain’s ability to integrate and interpret information detected by the different sensory systems including hearing (auditory), taste (gustatory), touch (tactile), smell (olfactory), somatosensory and visual systems. Somatosensation is a mixed sense, relating to temperature, pressure, vibration and body position (proprioception); some consider touch a component of this also. Perception involves multiple steps in processing sensory information: organising, assigning meaning and creating an understandable representation of the sensory landscape. 2 Perception is an umbrella term for various abilities, that are both successive and interactive. 3 In vision, for example, perceptual abilities range from perceiving simple physical characteristics in a scene, such as shape and colour, to ‘higher’ level skills including recognition and visuoconceptual processing. 4

PIONEER definition of perception

Detailed, working definitions of perception are a challenge, with variations in the scope and components included: consensus remains elusive. The delineation between perception and sensation, attention and cognition is one aspect of this – sensation and perception can be conflated, but perception can also be considered one of several cognitive abilities. 2,5 Conceptual differences often vary by disciplinary background, theoretical approach, research methodology and geographic location, and these also vary with time. The lack of an agreed definition of perception has been a challenge encountered by previous reviews of interventions for perceptual disorders after stroke. 6

Working with our Lived Experience and Clinical Expert stakeholder groups we explored the issue and developed a feasible, working definition for our project [see sections Stakeholder activities (what happened) and Types of participants: defining perception]. We used the World Health Organization’s (WHO) International Classification of Functioning, Disability and Health (ICF) definition of perception – the ‘specific mental functions of recognizing and interpreting sensory stimuli’7,8 (ICF code b156). This definition (1) provides a very clear distinction between perception and other closely related functions of sensation and cognition; (2) is applicable to all senses and not just vision; and (3) is internationally accepted. We excluded sensory disorders (ICF code b2) and disorders of attention (ICF code b140, encompassing visual neglect), which have a separate evidence base. 9,10

Using this definition meant automatic exclusion of a range of functions (‘consciousness, orientation, attention, memory; language; seeing; hearing; and additional sensory functions'). 7,8 We therefore excluded disorders relating to (1) sensation (ICF code b2; including visual field loss) and (2) attention (ICF code b140, encompassing visual neglect), which have been classified as ‘perceptual’ in prior research. While we recognise both are closely associated with perception, we considered each to be inherently different from it: sensation arises ‘earlier’ in the process of acquiring and processing visual information and it relates to detecting the presence of sensory information while attention is the ability to attend to sensory information. In addition, sensory and attentional disorders will present and will often be assessed and treated differently from perceptual ones. Further, both sensory and attentional deficits have existing, separate evidence bases. 9,10 Thus, a review that focuses on perception as defined by WHO provides an evidence base that is unique, and maximally clinically relevant.

Perceptual disorder aetiology

Perceptual disorders can arise from a range of conditions. Neurological aetiologies are prevalent, with perceptual disorders associated with sudden onset conditions of stroke or head trauma, and also neurological disease, including meningitis, Parkinson’s disease,11 Alzheimer’s and Lewy-body dementia. 12,13 In children, developmental disorders, including cerebral palsy,14 can affect perception, while increasingly research suggests that processing of sensory information, especially relating to vision and touch15 may be impacted among people with autism. 16 Psychiatric conditions13 including schizophrenia and depression are associated with visual, taste and smell disorders. 11

The PIONEER project focuses on perceptual disorders occurring because of stroke.

Nature, incidence and prevalence

An estimated 100,000 people living in the UK have a stroke each year,17,18 a number forecasted to rise by 59% over the next two decades. 19 Over 1.2 million UK adults live with the long term-consequences of their stroke. 20 The average age of an individual with stroke ranges from 71 to 78 years, but around one-quarter of strokes now occur in a person of working age. 17 Overall, the presentation of stroke-related perceptual disorders and natural recovery is poorly understood. 21 While significant improvement may occur the first 6 months after onset22 longer-term prevalence data are highly variable. Available data suggest a wide range of prevalence figures, from 5% to 75% (see below). Using examples from somatosensation, we estimate that around one-fifth, or 240,000 stroke survivors in the UK may have a perceptual disorder. 23,24

Information on the incidence, prevalence and natural history of perceptual disorders after stroke has been neglected in research. We are aware however that deficits can affect a broad range of perceptual skills relating to an isolated sense, or alongside disorders in other sensory modalities.

Impact of perceptual disorders after stroke

Perceptual disorders will reduce an individual’s ability to understand their environment and respond appropriately to it; for example, stroke survivors with visual perceptual disorders may not recognise family members, while spatial difficulties may cause disorientation and anxiety in busy environments, leading to reluctance to leave the home. 37 Visual perceptual dysfunction is associated with reduced abilities in activities of daily living (ADLs),38 greater disability, poorer quality of life (QoL)39 and can predict self-care difficulties. 40

Auditory perceptual disorders impact on listening skills and are likely to contribute to poorer auditory comprehension and communication abilities. Stroke assessments and interventions are typically based on healthcare professionals’ spoken instruction. Stroke survivors that have trouble communicating will also experience difficulties in diagnosis, and rehabilitation participation. 28,41,42

Taste dysfunction can lead to subjective unpleasantness when eating, impaired appetite, dietary changes, malnutrition and weight loss. Stroke survivors who are malnourished have poorer outcomes and require longer hospital admissions. 43,44 The inability to smell negatively impacts on eating, social communication and safety (e.g. detecting a gas leak). 45

Altered perception of the various components of somatosensation can lead to poorer performance of motor tasks, particularly control of fine motor skills in the hand (such as grip control, touch, pressure, proprioception),46 greater risk of accidents and injuries such as scalds and burns (temperature), increased incidence of falls28 (proprioception, learned non-use of limbs)47 and is linked with poor recovery of motor function and reduced independence in ADLs. 48 Those affected by altered somatosensory perception have greater activity limitations and longer hospital stays. 49

While there is evidence of the effect of perceptual impairments on stroke survivors’ rehabilitation outcomes and ability in everyday tasks, there is very limited exploration of the lived experience of such disorders, across the different senses. Stroke survivors and carers may well not recognise or understand that a problem they experience is due to impaired perception, and such find it a ‘puzzling and disabling.’50 Where perceptual disorders are not assessed and/or identified (see Poor documentation and variability) the nature of the issue can be mistakenly attributed to disorders of communication, memory, balance or motor skills. Where stroke survivors are aware of their perceptual impairment(s), they can have difficulty articulating their experiences of this51 but can detail the extra time and effort needed to accomplish tasks ‘You just have to be so methodical, so slow and it takes me forever to do stuff’52 and associated frustration. In addition, there are a range of emotional consequences: despair, anger, changes in self-confidence, feelings of worthlessness, vulnerability and changes in personal identity. 52 A number of online resources exist to inform and support carers and stroke survivors affected. 53,54

Interventions for perceptual disorders after stroke

The current literature offers some proposed interventions for the management of perceptual disorders after stroke: common to all six sensory areas are screening and assessment interventions to enable timely and accurate diagnosis. 28,33,55–57 Treatment approaches are primarily rehabilitative, aiming to compensate for the loss of function, but these vary depending on the sense affected and the nature of the dysfunction.

Therapeutic approaches to visual disorders may include sensory stimulation (visuo‐perceptual tasks),58,59 functional training (everyday tasks)60 and strategy training (alternate strategies to achieve goals) including the use of other senses to do so. 61–63 More recent interventions have used computer-based virtual reality training,64,65 incorporated visual and auditory feedback,66 or used transcranial direct current stimulation (tDCS) to stimulate the brain. 67 For auditory perception disorders, approaches may include environmental modifications, assistive listening devices,68 development of compensatory strategies or auditory training, which aims to improve the affected auditory process(es) through challenging listening tasks. 69

Few stroke-related olfactory and gustatory dysfunction treatments are reported:70 Pharmacological approaches have been considered,71 as well as referral to a dietitian for advice. 72 For impaired touch perception, interventions have focused on the upper limb, and include retaining sensory recognition and discrimination using specialist equipment. 73 Interventions for somatosensory perceptual deficit vary dependent on the specific function targeted but may include courses of sensory retraining using a range of stimuli,74 or targeted physiotherapy, which may incorporate robotic or highly specialised equipment. 75

Poor documentation and variability

Descriptions of current screening, assessment, treatment and referral pathways for stroke survivors with perceptual disorders are limited and variable. As perceptual impairments can affect all six senses, where services exist, they may be delivered by one of several healthcare professions (HCPs), including occupational therapists (OTs), physiotherapists, doctors, psychologists, orthoptists, audiologists and ear, nose and throat (ENT) services. HCPs and members of the public may be unaware of the range of perceptual impairments across hearing, smell, somatosensation, taste, touch and vision that may present after stroke and disorders may go unreported, under diagnosed or untreated. 43,76–78 Where UK service data are available, provision for visual disorders varies greatly76,79,80 with lack of standardised tests and procedures. 81 There are several barriers to effective service delivery, one of which is an evidence base on which to base treatment decisions. 79,80,82,83

Limited research informing clinical guidelines

Guidelines highlight the paucity of perceptual disorder intervention research on which to base clinical recommendations. 3,84,85 UK stroke clinical guidelines for adults (which are in update) refer to perceptual disorders, but not all sensory modalities are mentioned: three consider vision, one considers sensation (appearing to include tactile perception), and none make recommendations on hearing, taste or smell dysfunction. 3,84,85 The Royal College of Physicians (RCP) clinical guideline for stroke is the most comprehensive but focuses solely on visual perception, and agnosia (impaired object recognition), a specific visual disorder. The existing guidelines based their recommendations on a historic Cochrane Review6 which found no evidence of benefits for perceptual disorders after stroke.

Orthoptists’ clinical guidelines refer only to visual agnosia and hallucinations (recommending the provision of information) and the evidence underpinning these recommendations is unclear. 86 Paediatric stroke guidelines noted the absence of relevant research evidence and recommend the assessment of vision and hearing. Specialist support services and functional impacts, and tactile stimulation was also suggested for children with altered upper limb sensation. 87 These best practice recommendations were based on the opinion of the guideline development group.

Existing reviews of interventions for perceptual impairments in stroke

A comprehensive review of the evidence relating to all perceptual disorders after stroke has not been conducted to date. The reviews relevant to this topic have limitations as they may:

1. Include interventions for non-perceptual deficits, such as visual perceptual disorder reviews which include attentional deficits88 and those of touch and somatosensation which include sensory impairments. 89

2. Relate to a small subset of perceptual impairments within one sense, rather than considering all perceptual disorders relating to the sense as a whole. 90

3. Have a clear focus on exploring interventions for visual disorders, to exclusion of other senses.

4. Include non-stroke populations, making clinical interpretation challenging. 6

Systematic scoping review with evidence gap mapping

We undertook a scoping review of the literature to identify all evidence relating to interventions for perceptual disorders after stroke (see Aims and objectives). Scoping reviews map a broad field of literature, an approach ideally suited to this objective, given the variety of perceptual problems occurring post stroke, and the wide range of potential interventions.

Adhering to published guidance we followed a six-stage scoping review framework, including thorough searching and broad study design inclusion criteria. 94 We searched the relevant research comprehensively, working alongside our stakeholder groups and information specialist (JC) to develop a rigorous exploration of the existing literature. 95 We included all interventions, all participant age groups, settings, study designs (including quantitative and qualitative methods) and outcomes relating to interventions for perceptual disorders. Results were tabulated and summarised narratively. Details of the systematic scoping review methods (see Chapter 4) and results (see Chapter 5) are provided later.

Cochrane systematic review

We undertook a Cochrane systematic review and meta-analysis, exploring RCT evidence of the effectiveness of interventions for perceptual disorders after stroke (Objective 2). Cochrane Review methodologies provide the highest quality approach for synthesis of evidence intervention effectiveness. 96 We revised, expanded and updated an existing Cochrane Review6 published in 2011. We narrowed the review’s participant eligibility criteria to focus on stroke participant populations while expanded the intervention eligibility criteria to include all treatment approaches.

We identified, appraised and synthesised the relevant evidence from RCTs to determine where sufficient evidence exists of the benefits of a specific intervention for a perceptual disorder. We used the Template for Intervention Description and Replication (TIDieR) guidelines to maximise the clarity of our intervention data extraction and reporting97 and the grading quality of evidence and strength of recommendations (GRADE) approach to appraise evidence certainty. 98 Details of the Cochrane systematic review methods and results are provided in Chapters 6 and 7.

Integration and priority setting

We explored the implications of our scoping and Cochrane systematic review findings relating to interventions for people with perceptual disorders after stroke and the HCPs working with them (Objective 3). We aimed to maximise the relevance and applicability of the synthesised evidence to clinical practice and future research, and to identify any barriers to the uptake of that evidence. 99

Working with our Clinical Expert and Lived Experience Groups we interpreted our findings in the context of current clinical practice and stroke survivor experiences. 100 Using structured methods of involvement101 and priority setting,102 we agreed on the implications for clinical care in relation to: (1) stroke survivors and their carers; (2) HCPs providing care; and (3) policy-makers. We also prioritised the research gaps identified and developed recommendations for future research. Details of the methods and results are provided in Chapters 3 and 8.

Who was involved?

The Lived Experience Group were individuals aged 18 years or over, with personal experience of stroke-related perceptual problems, or as the parent or carer of a stroke survivor with perceptual problems.

The Clinical Expert Group were HCPs with expert knowledge of at least one sense addressed in this project. Clinical expertise was sought to complement that of the clinical research team, ensuring input relating to all senses, ages and healthcare settings, as well as a range of geographic locations.

How were people recruited?

When were they involved?

Stakeholder involvement can occur at any stage of a systematic review. 107 We used two involvement approaches: stakeholders participated in planned meetings at set time points, and informal communication occurring as needed throughout the review.

Planned meetings: these were held in person or online to inform decision-making for: planning methods, search development, selecting studies, analysing data, interpreting findings and writing the review (Figure 2). Stakeholders met virtually or in person at least every 3 months.

FIGURE 2.

The involvement of stakeholders in this project. 2

Informal communication: we provided regular updates to, and gained input from, the stakeholder groups throughout the project via e-mail or online updates. Individual stakeholders were available for consultation throughout and were contacted as required; for example, specific clinical experts were consulted regarding a study’s eligibility criteria during the study selection stage. Group members were invited to comment on draft abstracts, lay summaries and evidence gap maps via e-mail.

Level of involvement

We sought different levels of stakeholder involvement across the project:

• Controlling aspects of the review process – we pre-planned that several decisions would be made by the stakeholder group working in partnership with the research team. For example, the stakeholders controlled decisions about the outcomes of interest.

• Influencing the review activities and outputs – the stakeholders had a role in the review process and an opportunity to directly influence the review but stopping short of final decision-making control. For example, the stakeholders influenced the wording of statements relating to clinical implications.

• Contributing throughout the review – stakeholders had the opportunity to take part in meetings, respond to e-mails and provide input which could indirectly impact on decisions made within the review.

We recorded group members’ involvement and impact at each review stage. We also asked stakeholders to describe their perception of involvement for each of their activities. We have reported involvement using the GRIPP2 tool. 108

Stakeholder activities (what happened)

Meetings adhered to the key principles of research coproduction, creating an environment that recognised everyone’s contributions and in which people worked together to achieve a shared understanding. 109 Ground rules were agreed at the start of meetings which highlighted respect, inclusivity and joint ownership of decisions. We used practical techniques that facilitated participants’ input (experienced facilitators, a timekeeper) and ‘devolved’ some decisions to the participants, ensuring that they had control over decision-making. 101

We sought stakeholders’ input on six planned review-related activities (see Figure 2):

1. definition of key terms

2. outcome measurement identification and prioritisation

3. interpretation of the scoping review results

4. interpretation of the Cochrane systematic review results

5. identifying clinical implications of the review findings

6. determining research gaps and priorities (see Report Supplementary Materials 1–8).

Types of participants: defining perception

The definition of perception was determined via discussion and voting with our stakeholder groups [see Stakeholder activities (what happened), Activity 1 and Report Supplementary Material 1]. Members agreed to use the WHO ICF definition of perception ‘specific mental functions of recognizing and interpreting sensory without amendment or addition stimuli’, with the creation of a lay definition of ‘processing and understanding information from the senses’ to assist with clearly communicating findings. Members discussed, created and voted to accept definitions of the individual senses included (see Box 1). It was decided that although touch could be considered a component of somatosensation this would be treated as distinct, to support the accessibility of the findings since touch is one of the ‘traditional’ five senses.

Discussion between the research team and stakeholders led to a priori decisions relating to the inclusion of specific disorders that were either complex in nature or their categorisation as a perceptual disorder was unclear. These covered Pusher syndrome (a disorder of perception of body position)118 (included), hallucinations (included); balance, vestibular disorders and neglect (excluded). Studies that combined stroke and non-stroke populations were included and coded to indicate they were a mixed population (see Participants included). Studies that combined perceptual impairments with other disorders such as sensory or cognitive impairments were included and coded to indicate this (as perceptual, perceptual-sensory, perceptual-cognitive, mixed or unclear if the exact perceptual impairment could not be identified).

Types of interventions

Where interventions addressed perceptual disorders across more than one sense (e.g. smell and taste), these were included and coded to indicate the mixed grouping.

Types of outcome measures

Seventeen outcomes of interest were identified in collaboration with our stakeholder groups [see Stakeholder activities (what happened), Activity 2 and Report Supplementary Material 2]. These were prioritised by stakeholders and the research teams (see Table 1).

Electronic searches

Electronic bibliographic databases and clinical trial registers were searched from inception (unless otherwise indicated) to 7 February 2020 including the Cochrane Stroke Group Register, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews (CDSR) in the Cochrane Library (2020, Issue 1), MEDLINE (Ovid; from 1996) and EMBASE (Ovid; from 1980) (see Report Supplementary Material 9).

Searching other resources

We also conducted an extensive search for published and unpublished studies, via:

• OpenGrey [www.opengrey.eu/ (accessed September 2022)]

• Grey Matters: a practical tool for searching health-related grey literature [www.cadth.ca/grey-matters-practical-tool-searching-health-related-grey-literature (accessed September 2022)]

• Google Scholar [https://scholar.google.com/ (accessed September 2022)]

• NIHR Clinical Research Network [www.nihr.ac.uk/explore-nihr/support/clinical-research-network.htm (accessed September 2022)]

• Physiotherapy Evidence Database (PEDro) [https://pedro.org.au/ (accessed September 2022)]

• OTseeker [www.otseeker.com/ (accessed September 2022)]

• PROSPERO International prospective register of systematic reviews [www.crd.york.ac.uk/prospero/ (accessed September 2022)].

National and international guidelines, government, HCP, relevant charities and patient support organisations websites were searched. Research, professional associations, foundations and experts in the field were also contacted (see Report Supplementary Material 10). An initial database search identified few paediatric studies. In consultation with a paediatric specialist (LD) and reflecting the iterative nature of scoping review searches, further sources specific to children were searched (see Report Supplementary Material 10). Forward citation tracking using Science Citation Index and Google Scholar was also completed (last searched 24 November 2020) together with searching reference lists of included studies.

Applying selection criteria

Searches were imported into EndNote software (v8) and de-deduplicated. 120 Titles were screened for inclusion with ineligible titles or duplicates excluded (KMcG). Potentially relevant abstracts were independently screened using Covidence systematic review management system (KMcG, CH). Based on the inclusion criteria, abstracts were independently categorised as ‘relevant’, ‘irrelevant’ or ‘unsure’. Studies ranked as irrelevant by both reviewers were excluded. The full text of the remaining studies was retrieved and assessed independently (KMcG, CH). Where disagreement occurred, or a study was categorised as unsure, an expert in that sensory area was consulted. Exclusion reasons were recorded and reported. 117

Screening of grey literature or studies identified during supplementary searches was conducted by one researcher (KMcG, PC or CH) and key identifiers were entered into Microsoft Excel^®. Details were checked by a second researcher.

Perception terminology and decision-making

Perceptual terminology is complex (see PIONEER definition of perception) and specific challenges we encountered in determining whether a study population had a perceptual disorder included:

• the lack of universal agreement on a definition of perception and the need to determine authors’ precise meaning, and delineation between disorders of perception, sensation, attention and cognition

• differences in terminology between the senses, populations (adult and paediatric) and disciplinary fields.

Challenges were compounded by poor reporting of the disorder – both whether a study population had a perceptual disorder and whether an intervention addressed a perceptual disorder. Reviewers considered all relevant and available information, including reported details on type and location of stroke, the tools used to assess perceptual function and theoretical frameworks underpinning intervention design when distinguishing perceptual from other disorders. The key phrase ‘distinguish, discriminate, recognise and interpret’ from our coproduced definition [see Stakeholder activities (what happened)] supported the determination of whether a disorder reported related to perception or sensation: specifically, we sought evidence of processing of sensory information as opposed to simply detecting the presence or absence of sensory stimulation. Our research team and Clinical Expert Group specialists provided valuable third reviewer input as required in applying our inclusion criteria.

Data extraction forms

Data extraction forms were developed in Microsoft Excel, based on the TIDieR checklist97 and forms used in prior complex evidence syntheses. Data extraction forms were independently piloted (CH, KMcG) on five studies with different research designs, populations and interventions. Completed forms were then discussed and refined to ensure that all relevant data were extracted in an efficient manner. Data charting was completed (KMcG) and cross-checked (CH). Where studies included mixed populations, stroke-specific data were extracted where possible.

Data items extracted

We extracted the following data for each included study:

• Demographics: author, year of publication, type of publication, country.

• Design and methods: aim, design, number of recruitment sites, stakeholder involvement, participant numbers, withdrawals or lost to follow-up.

• Participant characteristics: age, sex (% female), stroke severity measurement, stroke type, hemisphere affected, other stroke-related impairment, time since stroke, inclusion of non-stroke survivors (n).

• Perceptual disorder: sense, diagnosis.

• Intervention and comparator characteristics: details extracted using the TIDieR checklist,97 including number of interventions in the study, intervention approach, theory supporting the intervention, materials used, reporting of intervention procedure, who provided the intervention, mode of delivery, location, duration, single/multiple sessions, adaptation, other interventions tested, if usual care was provided.

• Assessed outcomes: outcome measures/tools for each eligible outcome; other outcomes specified and reported and data collection time points.

• Additional qualitative data: we planned to extract any descriptive themes relating to intervention effect/impact, costs or implementation, including the name and description of the content and meaning. 121

Intervention categorisation

Selected data were categorised using drop-down menus within the data extraction form, to facilitate interpretation and maximise the clarity of reporting. Intervention categorisation was based on the underlying therapeutic approach and used an established approach. 32,33

Interventions were categorised as:

• pharmacological – relating to the use of drugs

• surgical – relating to suture, incision, excision, manipulation or other invasive procedure that usually, but not always, requires local, regional or general anaesthesia122

• non-invasive brain stimulation (NIBS) – technologies and techniques used to modulate the excitability of the brain via transcranial stimulation (such as tDCS)123

• rehabilitation – designed to optimise functional ability and reduce disability in individuals with health conditions, in interaction with their environment. 124 Rehabilitation interventions were subcategorised as:

  • restitution (direct training of the impaired function)

  • compensation (via training to use a spared function)

  • substitution (use of an external device or modification)125

  • mixed (a combination of the above approaches).

Categorisation was undertaken by a researcher (KMcG), checked by a second (CH) with expert input as required (DG, SMH). See Report Supplementary Material 11 for full list of data categorisation options.

Collating and summarising the results

We collated the evidence identified in the scoping review into tables using the extracted and charted data. 94,95 Data were organised by study characteristics (design, year, continent), population (side of stroke, duration of stroke, sense affected, age, sex) and intervention (using TIDieR-based descriptors including approach, materials, who delivered, modality, location, duration, number of sessions, tailoring). The frequency of outcome measures across included studies was also tabulated. We created graphs and numeric summaries. Comparisons explored the nature and breadth of the findings, and identified key areas where data were poorly reported or absent. Results were discussed with the Lived Experience Group [see Stakeholder activities (what happened)].

Reporting the results

We created narrative data summaries94 based on the following:

• The number and characteristics of included studies, study design, year of publication and country of origin.

• The participant population, including the details of their stroke and the senses affected.

• Intervention summaries were created reflecting the key TIDieR97 checklist categories and grouped according to the sense affected.

• The nature and number of outcomes in each category were presented, noting gaps.

Interactive evidence gap mapping

A series of interactive evidence gap maps116,126,127 provided a simple and accessible visual summary of the evidence using Tableau and Evidence for Policy and Practice (EPPI)-mapper software packages. Maps were shared with the Lived Experience Group to elicit their opinions, identify the most important information and how best to organise the information. Group members preferred a 2 × 2 matrix (in EPPI-mapper), which showed the relationship between the volume of evidence for an intervention category and the associated study outcomes. Group members recommended that the maps should be uncluttered, but sufficiently detailed to provide a legacy database, supporting future perceptual disorder research. Evidence gaps were highlighted.

Design and location

Included studies were predominantly case reports (45%, 36/80) or RCTs (27.5%, 22/80) but also included N-of-1 designs (11.2%, 9/80), cohort studies (11.3%, 9/80) and controlled trials (2.5%, 2/80). No qualitative studies were included. Studies were conducted in Asia (33.8%, 27/80), Europe (32.5%, 26/80), North America (18.8%, 15/80), Australia (8.8%, 7/80) and South America (3.8%, 3/80). The setting for two studies (2.5%, 2/80) was not reported.

Recruitment

Few (41.2%, 33/80) reported recruitment data, perhaps reflecting the prevalence of single-participant studies (43.7%, 35/80). Only 5% (4/80) described recruitment over multiple sites.

Age of data

All 80 studies were published between 1973 and 2020: 34 (42.5%) in the last 5 years and 51 (63.8%) in the last 10 years. Most RCTs (12/22, 54.5%) were published between 2015 and 2020 (Figure 4).

FIGURE 4.

Studies included in the scoping review: study design and year of publication. Note: figure is based on one published in Stroke. 112

Reproduced with permission from Hazelton et al. 112 This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. The text below includes minor additions and formatting changes to the original text.

Attrition

Where reported, participant withdrawal and loss to follow-up was low; 28 (35%) reported no withdrawal, 3 (3.8%) reported 1 withdrawal, 2 (2.5%) reported 2 withdrawals, 1 (1.3%) reported 4 withdrawals and 1 (1.3%) reported 7 withdrawals.

Stroke survivor or family involvement in research

Stroke survivor or carer involvement was not reported in any studies’ design or implementation.

Age

Most participants were adults with 53.8% (43/80 studies) 18 to 65 years, and 31.3% > 65 years (25/80 studies); 5 children (6.3%; < 18 years) were included in five individual case studies.

Sex

Fewer females were represented within included studies, with mean percentage across the studies of 34.8%. This was not observed to alter by age of the data set.

Stroke severity, lesion and concurrent impairments

Few studies reported stroke severity (16.3%, 13/80). Right hemisphere lesions were common [39/80 studies (48.8%) recruited > 60% participants with right-sided lesions]. This was most apparent for studies of somatosensory deficits with 64.2% studies having > 60% participants with right-sided lesions. Concurrent stroke-related impairments were noted in 51.3% (41/80) of studies; however, this was not reported in 49.3% (39/80) of studies.

Time since stroke

Stroke survivors were recruited across the stroke trajectory: acute (15.6%, 139/893), subacute (38.1%, 340/893) and the chronic phases (35.6%, 318/893).

Diagnosis

Perceptual disorders were diagnosed using standardised tests (79.2%, 707/893), clinical assessments (see Included studies; 47/893) or both (2.1%, 19/893). For 26 participants (2.9%) their diagnosis was based on a combined clinical assessment and self-report. Two further participants’ perceptual disorder diagnosis was based on self-report (0.2%) while the method of diagnosis was unreported for 92 participants (10.3%, 92/893).

Visual perception

Stroke survivors with visual perceptual disorders accounted for 40% (357/893) of participants included in this review, across 34 studies (42.5%, 34/80); disorders included visual–spatial deficit (3.9%, 35/893), visual hallucination (including Charles Bonnet syndrome) (0.9%, 8/893), visual agnosia (0.3%, 3/893) or ‘other’ visual perceptual disorders (4.1%, 37/893). The largest group of participants were classed as experiencing a non-specific ‘visual perceptual deficit’ (30.6%, 274/893), which could include those with a mix of several perceptual issues, or diagnosis was based on perceptual test score, for example Motor-Free Visual Perception test (MVPT)129 (Figure 6) that did not specify the nature of the disorder detected.

FIGURE 6.

Included participants: by sense (inner ring) and perceptual disorder (outer ring). Note: figure is based on one published in Stroke. 112

Reproduced with permission from Hazelton et al. 112 This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. The text below includes minor additions and formatting changes to the original text.

Somatosensation

Somatosensation deficits were reported by a third of participants (33.9%, 303/893) across 28 studies (35%, 28/80). Disorders included Pusher syndrome (26%, 232/893), proprioceptive deficits (1.6%, 14/893) or were categorised as ‘other’ somatosensory disorder (6.4%, 57/893).

Auditory perception

Few studies (8.7%, 7/80) described participants with an auditory perceptual disorder. These were classed as either an auditory processing disorder (3.6%, 32/893) and/or hearing ‘other’ (0.1%, 1/893).

Tactile perception

All participants were reported to experience a form of general tactile dysfunction (12%, 107/893) across six studies (7.5%, 6/80).

Mixed perceptual disorders

Studies that recruited participants with perceptual disorders of two or more senses included participants with mixed tactile-somatosensory disorder (7.6%, 68/893), mixed taste–smell disorder (0.1%, 1/893) or mixed vision–tactile disorder (2.7%, 24/893).

Nature of interventions

Intervention materials included technology based, HCP led and those using specialist equipment.

Mode of delivery

Most interventions were delivered one to one (81.7%, 76/93), with few self-managed (4.3%, 4/93) or delivered to a group (3.2%, 3/93). These details were unavailable for nine interventions (9.7%, 9/93) or were unclear (1.1%, 1/93).

Location

Interventions were delivered in a hospital inpatient (37.6%, 35/93), in or outpatient (21.5%, 20/93), or outpatient basis (1.1%, 1/93). Only three were home-based (3.2%, 3/93). A third of interventions did not report this detail (33.3%, 31/94).

Frequency

Interventions were often delivered over multiple sessions (72%, 67/93). Eight were delivered in a single session (8.5%, 8/93). For 17 interventions (18.3%) these details were unreported or unclear (1.1%, 1/93).

Duration

The included interventions lasted < 1 week (15.1%, 14/93), 1–4 weeks (30.1%, 28/93); 1–3 months (10.8%, 10/93) or > 3 months (4.3%, 4/93). Several studies did not describe the duration (34.4%, 32/93) or it was unclear (5.4%, 5/93).

Modification or tailoring of interventions

Information on tailoring of interventions to participants or any intervention modifications were not reported by approximately half the interventions identified (54.3%, 51/93). Tailoring (usually to initial ability or in relation to progress) was described for 12 interventions (12.8%) with a further 23 interventions (24.5%) stating there was tailoring, but no details were provided. For seven interventions, this information was unclear (7.4%).

Intervention fidelity

Most interventions did not refer to any fidelity measures (96.8%, 90/93) or the information provided was unclear (1%, 1/93). Three interventions described plans to measure intervention fidelity (3%, 3/93).

Some intervention details were rarely reported, and we found no report of the intervention procedure (10.8%, 11/93), provider (52.7%, 49/93) or duration (34.4%, 32/93). Most described an intervention rationale (76.4%, 71/93). Some described the delivery of experimental interventions alongside usual care (22.6%, 21/93), but this was not always reported (73.1%, 68/93).

Visual perception disorder interventions

Of 37 visual perceptual disorder interventions, 31 had a rehabilitation focus (83.8%, 31/37), including restitution (48.4%, 15/31), mixed (25.8.1%, 8/31), compensation (12.9%, 4/31) and substitution interventions (3.2%, 1/31). Other intervention approaches were unclear (9.7%, 3/31) (Table 2).

Restitution approaches used technology (66.7%, 10/15; including interactive computer-based training of visual skills) and HCP-led interventions (33.3%, 5/15) including teaching compensatory skills in real-world simulation tasks, training of eye movements or completion of perceptual-based tasks. Compensatory approaches included intentional blinking, route and strategy training while substitution interventions included covering of one side of glasses with cardboard. One study used NIBS (Repetitive Transcranial Magnetic Stimulation – rTMS) for visual hallucination.

Pharmacological interventions identified (e.g. haloperidol, Librium and risperidone) targeted visual hallucinations and were described in single case reports with limited intervention details. Generally, visual perceptual disorder studies lacked details on the intervention provider, location, intensity and duration.

Five case reports described mixed rehabilitation interventions for children with a visual perceptual disorder including specialist reading software and the development of strategies to overcome difficulties. Two interventions used a compensatory approach (route training and tutorial-based sessions). Where locations were reported they included school, home (using self-delivered tutorials) and inpatient settings.

Somatosensory disorder interventions

Four of the 35 interventions identified in the scoping review targeted somatosensory disorders and were categorised as rehabilitation (restitution 68.6%, 24/35; mixed 17.1%, 6/35), NIBS (8.6%, 3/35) and rehabilitation + NIBS (5.7%, 2/35) (Table 3). Most were HCP-led (48.6%, 17/35) and included postural control or reach training with weight-shifting exercises. Technology-based interventions (25.7%, 9/35) included robot-assisted gait training using the Lokomat device, postural training devices or gaming such as Nintendo Wii^®. Interventions were predominantly delivered on a one-to-one basis (91.4%, 32/35), in an inpatient hospital setting (51.4%, 18/35), for 1 month or less (71.4%, 25/35).

Hearing perception disorder interventions

Seven interventions targeted hearing perceptual disorders, all of which adopted a rehabilitative approach (see Table 4). They were categorised as restitution (28.6%, 2/7), substitution (57.1%, 4/7) or mixed (14.3%, 1/7). Interventions were primarily technology-based (e.g. hearing aids 71.4%, 5/7) but also included HCP-led phonological therapy (14.3%, 1/7). The details of one intervention were unavailable (14.3%, 1/7). Where reported, all interventions were conducted on an individual basis (71.4%, 5/7) in a hospital setting (in/outpatient, 57.1%, 4/7) with one intervention (personal frequency modulated system) self-delivered at home (14.3%, 1/7) (Table 4).

Tactile perception disorder interventions

We identified seven interventions targeting tactile perception disorders which often involved a rehabilitation approach: restitution (57.1%, 4/7) or NIBS (28.6%, 2/7). Interventions included tDCS (28.6%, 2/7) or specialised equipment (57.1%, 4/7) such as texture grids, object recognition or sensory discrimination equipment. One intervention was technology-based (14.3%, 1/7) delivering vibrotactile noise. Physiotherapists provided two interventions (28.6%, 2/7) but most providers were unreported (71.4%, 5/7). All interventions were delivered on an individual basis.

Mixed perception disorder interventions

Studies that recruited stroke survivors with perceptual disorders across more than one sense described seven interventions to address tactile and proprioceptive deficits (71.4%, 5/7) including specialised tactile and proprioceptive training equipment, such as texture grids and proprioceptive stimuli. Interventions were delivered on an individual basis, but few other details were available such as provider, location or duration. One study (14.3%, 1/7) explored the use of NIBS for a visual and tactile perceptual disorder, while another pharmacological study addressed complex taste/smell hallucinations by prescribing carbamazepine and valproate; no other details were provided.

Outcome measures

The outcome measures included in this review reflect those prioritised by our stakeholders [see Stakeholder activities (what happened) and Types of outcome measures]. The full list of outcomes was discussed by the research team to select those of most relevance to the Cochrane Review questions, considering Cochrane reporting guidance, clinical practice and data availability [based on the scoping review (see Outcome measurements)]. We selected a primary outcome measure (see Primary outcome) and five secondary outcome measures (see Secondary outcomes); this did not include any outcomes of feasibility or cost-effectiveness.

Study selection

Once database searches were updated, duplicates were excluded, as were titles unrelated to stroke and perception (KMcG/KT). Two researchers independently considered the abstracts for remaining records (CH, KMcG/KT), excluding any that did not meet the inclusion criteria. Full texts for all potentially relevant studies were independently assessed by two reviewers (CH, KMcG/KT), as well as full-text reports of all RCTs identified in the scoping review and the historical Cochrane Review6 were appraised. Disagreements relating to abstract or full-text inclusions were resolved through discussion, involving a third reviewer or relevant clinical specialist stakeholder where required. Reasons for exclusion were recorded.

Data extraction and management

Two piloted data extraction forms were used containing the data and categorisation variables (see Data charting and categorising) and detailing comparison(s) and outcomes measurement data. One reviewer extracted the data, which was checked by a second (KMc/KT/CH). For RCTs identified subsequent to the scoping review, the data were extracted directly into Revman (KT) and independently checked by a second reviewer (CH): The variables extracted were:

• Study: country, setting, year, design, number of centres, number of randomised groups classified as (active intervention/no treatment/control).

• Methods: randomisation method, prospective power calculation, use of intention‐to‐treat analysis, recruitment, dropouts.

• Participant: inclusion criteria, exclusion criteria, number, age, sex, stroke details (type, time since stroke, hemisphere affected, severity), perceptual disorder, sense(s) affected and method of diagnosis, severity, presence of other stroke-related impairment.

• Intervention: intervention approach (rehabilitation/NIBS/pharmacological/surgery/assessment and screening). Rehabilitation interventions subclassification (restoration, compensation or restitution, or a combination; see Intervention categorisation). Invention materials, procedures, provider, mode, location, session and duration details, tailoring and modification. 97

• Outcomes: outcome measurement instrument, measurement taken, time point and final value scores. Dichotomous data: participant numbers who experienced the event in each group. Continuous data: means and standard deviations (SDs) by group.

Two review researchers (CH/KMcG/KT) independently classified all intervention approaches, seeking input from a third (DG/SH) where differences were unresolved through discussion.

For RCTs that randomised a mixed stroke/non-stroke or perceptual disorder/non-perceptual disorder subpopulations, we planned to extract the stroke-specific data wherever possible. Where this was unavailable, we used mixed population data if > 80% were stroke survivors with a perceptual disorder. We planned to conduct sensitivity analyses to investigate the effect of including these data.

Where data were unavailable, we calculated it from the available values. 222 For all outcomes, we recorded any significance test, t, f, p values and direction of findings. If a RCT provided data on more than one of the primary outcome measurement instruments, we extracted them in the order described above (see Primary outcome). Where a RCT reported more than one outcome measurement instrument relating to a single outcome previously unidentified, data on the measurement instrument’s validity and reliability were sought and compared, in consultation with a clinical expert in the relevant topic area, to choose the order of preference. This process did not consider the volume of data captured.

Risk of bias

Using the Cochrane ROB-1 tool we considered risk of bias. 223 Two reviewers (CH/KT) independently appraised included RCTs, grading the following risks:

Random sequence generation (selection bias): The method used to generate the random sequence was noted and judged whether it should produce comparable groups.

Allocation concealment (selection bias): The method used to conceal the allocation sequence was noted, and we considered whether intervention allocation could have been foreseen before or during enrolment.

Blinding (performance bias, detection bias): The measures used to mask RCT participants, researchers and outcome assessors from knowledge of allocation was noted, alongside any data on the measure’s effectiveness, and used to determine the degree to which blinding was achieved.

Incomplete outcome data (attrition bias): The completeness of data for each main outcome was noted, including attrition and exclusions from analyses, participant numbers in each intervention group (compared with total randomised), reasons for attrition or exclusions and any re-inclusion in analyses.

Selective reporting (reporting bias): The outcome measurement instruments, and the reported data were compared to identify missing data. Trial protocols were used where possible.

Other bias: We noted any other concerns about bias.

Risks of bias were categorised as high, low or unclear, with the reasons for decisions reported. Judgements for performance bias and detection bias were combined for reporting. If these differed, then the decision protocol was that (1) if low and unclear the overall category was labelled as unclear and (2) if high and low/unclear the overall category was considered as having a high risk of bias.

Data synthesis

The pre-specified comparisons for disorders in each sense (hearing, smell, somatosensation, tactile, taste and vision) were:

• active intervention for perceptual impairment versus no treatment;

• active intervention for perceptual impairment versus control (attention control/standard care/placebo).

We stratified the analysis according to intervention approach category (rehabilitation/NIBS/surgery/pharmacology/assessment and screening).

We also compared active interventions directly, where it was considered meaningful to do so. Decisions to do so were made following discussions among the research team, using tabulated data on the relevant RCTs.

Measures of treatment effect

Review Manager software (RevMan 5.4) supported the statistical analyses to determine treatment effects, using a random-effects model throughout. For dichotomous variables we planned to calculate a Peto odds ratio with 95% confidence intervals (CIs); for continuous data calculated the mean difference (MD) (for measurements in the same scale) and standardised MDs (for measurements in different scales) and 95% CI. We treated ADLs and other ordinal scales as continuous outcomes (an accepted meta‐analytic technique for ordinal outcome data is not yet available). Where a lower value indicated a better outcome on a measurement instrument, we multiplied the reported values by −1, so that a higher value was indicative of a better outcome across all analyses. We extracted final-value scores for analysis. If RCTs reported change-from-baseline values and the baseline value was available, we calculated the final value. If RCTs 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.

Somatosensory perception dysfunction

Population: One hundred and ninety-six participants in seven RCTs. 75,162,163,180,181,187,234 Somatosensory perception disorders included Pusher syndrome (five RCTs), diagnosed using the Burke Lateropulsion Scale75,181 and Scale of Contraversive Pushing,75,163,180 and less specific disorders of somatosensation, diagnosed via practical assessments, such as pinprick and tight touch tests. 187,234 For analysis, we split these into ‘Pusher syndrome’ or ‘not Pusher syndrome’. Further population information is given in Report Supplementary Material 16.

Interventions:

• Rehabilitation (restitution) (10 interventions) – game-based vertical posture training (n = 2),75,162 standard posture training (n = 2),75,162 conventional physiotherapy (for Pusher syndrome) (n = 2),163,234 physiotherapy + sensorimotor training,234 physiotherapy + motor training,234 computerised interactive visual feedback training (WiiFit),180 mirror feedback training. 180

• Rehabilitation (restitution and substitution) (n = 2 studies) – robot-assisted gait training. 163,181

• NIBS (one intervention) – tDCS. 187

Materials and procedures: One NIBS intervention used relevant equipment to deliver tDCS stimulation to the appropriate hemisphere and region. 187 All other interventions used rehabilitative approaches: for Pusher syndrome therapy these could be grouped into either (1) game-based postural training, with supporting equipment, or (2) conventional physiotherapy for Pusher syndrome; for non-pusher disorders one RCT explored sensory-motor training focusing on sense discrimination, and table-top ‘motor therapy’.

Game-based postural training used one of three named interventions: WiiFit (computer-based exercises and balance board),180 Lokomat (computer-based exercises, supportive harness and treadmill)163,181 and Spine Balance 3D (computer-based exercises and whole-body tilt apparatus). 75,162 Each was used to provide physical therapy, with participants asked to achieve a set body position or movement in response to the computerised exercises, and in relation to any positional change caused by the associated equipment.

Conventional physiotherapy for Pusher syndrome often involved postural training and weight shifting, using visual cues in the room to regulate posture, alongside verbal feedback from the therapist. Materials used were often unclear but included a chair for seated exercises and a mirror for feedback. 75,162,163,180,181

Sensory-motor training used the SENSe approach with three sensory discrimination tasks: texture discrimination, limb position sense and object recognition. 234 Materials in these tasks included different textures (fabric, wallpaper, plastic and sandpaper), different objects of varying shape, size and materials. A range of exercises were used, such as smoothing out fabric while appreciating the texture, moving the limb to a specific position and arranging bottles in order of weight. Motor therapy used tabletop games such as chess (with clear cognitive and attentional demand), with a set programme of upper limb exercises used to improve gross movement and dexterity. 234

Delivery: All interventions were delivered one to one, in a hospital setting. A physiotherapist delivered 5 of 13 interventions (which for 1 RCT specified that they had ‘more than 5 years’ experience’)162,180,181 while the remaining providers were not reported, or unclear.

Schedule and duration: There was similarity in the schedule and duration of the rehabilitation interventions. Sessions typically lasted 30–60 minutes per day, 3–5 times per week, for a duration of 2–4 weeks in total. In contrast, NIBS was delivered for 20 minutes per day for 10 days. 187

Tailoring and modification: Tailoring of interventions varied; seven interventions did not report tailoring. 75,162,163,180,187,234 Others reported that exercises were tailored to the participant’s ability before training began75,162,181,234 or that the difficulty level was altered relative to performance; for example the ‘the speed and range of trunk movement’ was increased or exercises were ‘changing from sitting to standing.’162 Detailed intervention descriptions are given in Report Supplementary Material 14. Intervention modifications were not reported by any RCT.

Tactile perception dysfunction

Population: Seventy participants with tactile perceptual dysfunction, in three RCTs. 201,235,236 Diagnostic tools included the revised Nottingham Sensory Assessment (rNSA)235 and Semmes Weinstein monofilament201,236 (see Report Supplementary Material 16).

Interventions:

• Four rehabilitation (restitution) interventions – pressure sense perception training on stable surface,201 pressure sense perception training on unstable surface,201 hand exercises (without glove)235 and a vibrating glove ‘VTS Glove.’236

• One rehabilitation (restitution and substitution) intervention – robot glove-based hand exercises. 235

Materials and procedures: Interventions were of two main types – pressure sense training involving exercises on a stable or unstable surface,201 and hand exercises, either with or without a glove to assist. 235,236 With pressure sense training, participants stood on either a stable foam block, or on an unstable balance pad. They were asked to shift weight to their affected side, and pressure in the heel was measured to ensure a desired level was reached. 201 Hand exercises included a range of passive range of motion tasks, and task-based activities and games. The addition of a robotic glove was used to detect movement and provide a simultaneous display of performance on a computer screen, as well as providing sensory stimulation. 235 A different, vibrating, glove was used to provide stimulation to skin on the palm and fingers; it did not require any exercises. 236

Delivery: Interventions were delivered by physiotherapists (two interventions201), OTs (two interventions235) in one-to-one sessions in a hospital setting (four interventions201,235), with the vibrating glove used by participants themselves at home (one intervention236).

Schedule and duration: Varied from 30-minute sessions, 3 days per week201 to 3-hour sessions, 7 days per week. 236 Total duration ranged from 4 to 8 weeks.

Tailoring and modification: Tailoring to participants’ ability was not reported for two interventions. 235,236 Pressure sense training was tailored to participants’ heel pressure, and allowed progression to a more difficult stage when a suitable pressure level was achieved,201 and robot-assisted hand exercise settings were adjusted to participants’ ability. 235 Detailed intervention descriptions are given in Report Supplementary Material 14. No modifications of the interventions during the RCT were reported.

Visual perception dysfunction

Population: Two hundred and twenty-five participants, in seven RCTs. 60,66,144,150,152,154,237 Visual perception disorders were diagnosed using specialised tests such as the MVPT or Rivermead Perceptual Assessment Battery,144,237 cognitive tests that include visual perception subsections (Modified Mental State Examination),66,152 and a complex figure-drawing test to test for visual memory deficit. 154

Interventions:

• Eleven rehabilitation (restitution) interventions – image drawing – global processing training,154 image drawing – rote repetition training,154 neurofeedback (NFB) training,66 WiiFit virtual reality training (WVRT),144 general balance training,144 transfer of training perceptual treatment,60 computerised visual perception rehabilitation with motion tracking,237 computer-based cognitive rehabilitation program,237 OT-led perceptual training,150 computer-based cognitive rehabilitation training,152 conventional cognitive rehabilitation. 152

• One rehabilitation (compensation) intervention – functional perceptual treatment. 60

Materials and procedures: All interventions used a rehabilitation approach and grouped into five main types: paper-based tasks, OT-led task-based training, physical interventions, cognitive and perceptual exercises and neurofeedback training.

Three interventions used paper-based tasks: in two, participants traced then reproduced a complex figure (Rey–Osterrieth Complex Figure) either as a whole figure or broken down into its component parts154; the third used ‘conventional cognitive rehabilitation’, the exact nature of which was not clearly stated. 152 Three interventions used an OT approach, training visual perceptual skills using functional and task-based training. 60,150 Although not well described, this training included simple perceptual activities such as stick length sorting, colour matching and parquetry mosaic tasks. One RCT explored physical interventions for visual perceptual disorders alongside balance disturbance. 144 WiiFit with balance board training used a range of activities to stimulate interest and motivation, such as simulated tightrope walking and slalom, and that encouraged multidirectional weight shifting. Balance training used a balance board, with the participant asked to shift weight, using a mirror for feedback. One approach used computerised exercises – these frequently were called ‘cognitive’ in nature, but had a clear focus on improving visual perceptual skills, including object recognition, object constancy, figure-ground organisation, visual discrimination and visual organisation. 152,237

Delivery: Delivery was poorly reported for visual perceptual disorder interventions. Where reported, a physiotherapist (one intervention144) and OT (three interventions235,237) delivered interventions on a one-to-one basis, in hospital settings.

Schedule and duration: Two interventions (involving paper-based repetition training) were delivered in a single 90-minute session. 154 For the other interventions, sessions typically lasted 30 minutes, 3–5 times per week, for 4–6 weeks.

Tailoring and modification: Tailoring was either not reported or was unclear for nine interventions; in three others, the intervention exercises and difficulty were based on the participant’s perceptual ability. 150,152 No intervention modifications were reported.

One RCT (50 participants) explored a mixed perceptual disorder, addressing tactile-somatosensory disorder. Mixed-sensory categories were not included in our analysis plan and were thus excluded.

Randomisation and concealment of allocation

All included RCTs described random sequence generation, but the method used was unclear for five RCTs. 66,75,144,150,236 The concealment of allocation was clearly stated by six RCTs,60,74,154,162,163,234 including sealed, opaque envelopes or a sealed box.

Blinding

In complex rehabilitation RCTs, adequate blinding of clinicians delivering and participants receiving an intervention is a significant challenge; no RCTs were assessed as having a low risk of bias. Blinding of outcome assessors was reported by 13 RCTs but was unclear for 5 RCTs. 66,75,152,181,201 Six RCTs were judged as having a high risk of bias. 60,74,144,154,234,237

Incomplete outcome data

Three RCTs were judged as unclear66,150,154 in their reporting of attrition bias with the remaining RCTs either including all participants in their analyses or accounting for all participants (with reasons given for dropout) and an intention-to-treat analysis conducted. One RCT234 had a high risk of bias as three participants were excluded from primary and follow-up analysis.

Selective reporting

For 16 RCTs, outcome measures were well reported. Only one RCT provided partial data for three outcomes234 and a further RCT74 failed to provide data for secondary outcomes.

Other potential sources of bias

Two RCTs had a high risk of bias. In one,163 several participants had severe cognitive deficits and the authors stated these deficits might have influenced the participants’ response to the intervention, although there were no data to support this intervention response. In a second,234 the mean age of the experimental group was significantly higher than the control group, and with more right hemispheric lesions.

Comparability of groups at baseline was adequate for 11 RCTs,60,74,144,154,162,180,181,187,201,235,237 while 5 RCTs were judged unclear. This was due to a lack of reporting of the baseline characteristics of participants66,152,236 or due to difficulties ‘securing homogeneity’ of participants, the nature of which was not fully explained. 75 For one,150 there was a change to eligibility criteria part-way through the RCTs to address slow recruitment: the RCT included those left hemisphere strokes, subarachnoid haemorrhage and head injury. It is unclear what interim analyses were undertaken and what the decision-making process was for continuation, adaptation and eventual stopping of the RCT.

Hearing

There were no RCTs of interventions for hearing perceptual disorders.

Smell

There were no RCTs addressing smell perceptual disorders.

Somatosensation

We identified seven relevant RCTs: five recruited people with Pusher syndrome75,162,163,180,181 and two included participants without Pusher syndrome187,234 (Table 7).

Intervention versus no treatment or control

Pusher syndrome

No RCTs.

Not Pusher syndrome

One RCT,187 comparing tDCS to sham treatment.

Activities of daily living

• One RCT (24 participants);187 Figure 11.

• ADLs measured using the Korean Modified Barthel Index.

• Analysis showed no difference between active intervention and control (MD 10.08, 95% CI −2.47 to 22.63, p = 0.12); the evidence was of very low certainty.

FIGURE 11.

Comparison of intervention vs. no treatment or control for somatosensory disorders. Outcome – ADLs. Note: All forest plots in this chapter are taken from the review published in CDSR. 204

Reproduced with permission from Hazelton et al. 204 This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. The text below includes minor additions and formatting changes to the original text.

Mobility

• One RCT (24 participants);187 Figure 12.

• Mobility assessed using the Functional Ambulation Category.

• Analysis showed no difference between active intervention and control (MD 0.50, 95% CI −0.38 to 1.38, p = 0.27); the evidence was judged to be of very low certainty.

FIGURE 12.

Comparison of intervention vs. no treatment or control for somatosensory disorders. Outcome – mobility and navigation.

Perception

• One RCT (24 participants). 187

• Measured somatosensory perception (modified Nottingham Sensory Assessment).

• Summary data were unreported (subscales only) and could not be included in analysis.

Adverse events

• One RCT187 (24 participants).

• Stated that ‘all the participants completed the stimulation sessions successfully without complaining about any discomfort during the procedure’.

Intervention 1 versus intervention 2

Pusher syndrome

Five RCTs addressed this comparison. 75,162,163,180,181 All RCTs compared an intervention training posture or/and movement using equipment such as balance boards, treadmills or harnesses alongside computerised to conventional physiotherapy for Pusher syndrome. Interventions in both groups were considered similar and the data were combined in meta-analyses.

Activities of daily living

• Three RCTs (80 participants);75,162,181 Figure 13.

• ADLs assessed using the Korean Modified Barthel index.

• Intervention 1 (computerised balance and movement training) was more beneficial than Intervention 2 (‘standard’ Pusher syndrome physiotherapy) [MD 10.19 (4.94 to 15.44), p = 0.0001]; there was no heterogeneity (I² = 0%) and evidence was of very low certainty.

FIGURE 13.

Comparison of active intervention 1 vs. active intervention 2 for somatosensory perception disorder. Outcome – ADLs.

Mobility

• One RCT (30 participants);163 Figure 14.

• Mobility assessed using the Performance Oriented Mobility Assessment – Balance (POMA-B) outcome measure.

• Analysis showed no evidence of a difference between Interventions 1 and 2 [MD 1.00 (−1.51 to 3.51)]. Evidence was judged to be of very low certainty.

FIGURE 14.

Comparison of active intervention 1 vs. active intervention 2 for somatosensory perception disorder. Outcome – mobility and navigation.

Perception

• One RCT (30 participants);163 Figure 15.

• Assessed perception via subjective visual vertical test.

• There was no evidence of a difference between intervention 1 and 2 [standardized mean difference – SMD 0.52 (−0.21 to 1.25)]. Evidence was judged to be of very low certainty.

FIGURE 15.

Comparison of active intervention 1 vs. active intervention 2 for somatosensory perception disorder. Outcome – perception.

Adverse events

Three studies reported on adverse events;75,162,181 in each case none had occurred.

Pusher syndrome outcomes

• Four RCTs (86 participants); Figure 16.

• Pusher syndrome severity assessed using the Burke Lateropulsion Scale75,162 or the Scale of Contraversive Pushing. 163,180

• Analysis showed a tendency for intervention 1 (computerised balance and movement training) to be more beneficial than intervention 2 (standard Pusher syndrome physiotherapy) (SMD 1.03, 95% CI 0.33 to 1.73), p = 0.004). Results may suggest substantial heterogeneity (I² = 50%), and the evidence was judged to be of very low certainty.

FIGURE 16.

Comparison of active intervention 1 vs. active intervention 2 for somatosensory perception disorder. Outcome – Pusher syndrome outcomes.

Not Pusher syndrome

One RCT234 addressed this comparison, comparing the SENSe intervention, consisting of sensorimotor therapy and sensory discrimination tasks, with cognitive table-top games and motor exercises.

Perception

• One RCT (36 participants);234 Figure 15.

• Perception assessed using the Nottingham Sensory Assessment.

• Analysis showed no evidence of a difference between intervention 1 and intervention 2 (SMD −0.38, 95% CI −1.04 to 0.28); the evidence was very low certainty.

Tactile

We identified three relevant RCTs. 201,235,236 Data contributed to analysis by Kim (2015)201 and Lee (2021)235 (Table 8).

Intervention versus no treatment or control

One RCT compared two active interventions with no treatment. 201

Quality of life – navigation and mobility

• One RCT with two relevant comparisons (30 participants);201 Figure 17.

• Mobility and navigation were measured using the timed-up-and-go measure.

• Analysis showed no difference between active intervention and no treatment (MD 6.50, 95% CI −4.81 to 17.81, p = 0.26). Results may suggest substantial heterogeneity (I² = 50%). Due to a number of methodological concerns, it was judged there was insufficient evidence to support a conclusion based on these data.

FIGURE 17.

Intervention vs. no treatment or control for tactile perception disorder. Outcome – mobility and navigation.

Perception

• One RCT with two relevant comparisons (30 participants);201 Figure 18.

• Perception was assessed using a dynamometer-based measure of proprioception.

• Analysis showed that a tendency for active intervention was more beneficial than no treatment (MD 4.64, 95% CI 3.06 to 6.22, p ≤ 0.00001). Results showed no heterogeneity (I² = 0%), and the evidence was judged to be of very low certainty.

FIGURE 18.

Intervention vs. no treatment or control for tactile perception disorder. Outcome – perception.

Active intervention 1 versus active intervention 2

Two RCTs addressed this comparison: one235 compared robot gloved-based hand exercises to hand exercises alone, and another201 compared pressure sense perception training on a stable surface to an unstable one (balance board). As the interventions were quite different, data were not combined in meta-analysis.

Activities of daily living

• One RCT (24 participants);235 Figure 19.

• ADLs were assessed using the Modified Barthel Index.

• This RCT demonstrated no difference between the interventions (MD −0.41, 95% CI −12.31 to 11.49). The evidence was considered to be of very low certainty.

FIGURE 19.

Active intervention 1 vs. active intervention for tactile perception disorder. Outcome – ADLs.

Quality of life – navigation and mobility

• One RCT (20 participants);201 Figure 20.

• Mobility and navigation were assessed using the timed-up-and-go measure.

• This RCT provided evidence that intervention 2 (training on the unstable balance board) was more beneficial than intervention 1 (training on the stable balance board) (MD −11.60, 95% CI −19.50 to −3.70). The evidence was judged to be of very low certainty.

FIGURE 20.

Active intervention 1 vs. active intervention for tactile perception disorder. Outcome – mobility and navigation.

Perception

• Two RCTs (44 participants);201 Figure 21.

• One assessed proprioception using a dynamometer and the other used the kinaesthetic subtest of the rNSA.

• Analysis did not combine data for these two RCTs: individually neither RCT provided evidence of a difference between interventions 1 and 2.

FIGURE 21.

Active intervention 1 vs. active intervention for tactile perception disorder. Outcome – perception.

Adverse events

One RCT235 collected data on adverse events, reporting ‘no safety concerns or adverse events’.

Taste

No RCTs of interventions addressing taste perception disorder were identified.

Vision

There were seven relevant RCTs60,66,144,150,152,154,236 (Table 9).

Intervention versus no treatment or control

One RCT compared intervention (neurofeedback training) with no treatment,66 and one RCT compared intervention (perceptual training) with control (conventional (not perceptual) therapy). 150

EADLs

• One RCT (33 participants);150 Figure 22.

• ADLs were assessed using the Rivermead ADLs scale.

• Analysis showed there was no evidence of a difference between active intervention and control (MD 0.94, 95% CI –1.60 to 3.48, p = 0.47). The evidence was judged to be of very low certainty.

FIGURE 22.

Intervention vs. no treatment or control for visual perception impairment. Outcome – EADLs.

Perception

• One RCT (27 participants);66 Figure 23.

• Perceptual outcomes assessed using the MVPT. Another RCT150 measured perception using the Rivermead Perceptual Assessment Battery: as only subscale scores for the Rivermead Perceptual Assessment Battery were reported, the data were not included in the analysis.

• Analysis showed no difference between active intervention and no treatment (MD −1.75, 95% CI −5.39 to 1.89, p = 0.35). The evidence was judged to be of very low certainty.

FIGURE 23.

Intervention vs. no treatment or control for visual perception impairment. Outcome – perception.

Active intervention 1 versus active intervention 2

Five RCTs addressed this comparison. 60,144,152,154,237 The interventions were dissimilar, including OT-led training in practical tasks, paper-based repetition exercises and computer-based games; statistical pooling of data was inappropriate.

Activities of daily living

• Two RCTs (96 participants);60,237 Figure 24.

• ADLs were assessed using the Modified Barthel Index.

• Data were not combined due to differences in the interventions. The evidence from each included RCT was judged to be of very low certainty.

FIGURE 24.

Active intervention 1 vs. active intervention for visual perception impairment. Outcome – ADLs.

Quality of life – mobility and navigation

• One RCT (28 participants);144 Figure 25.

• 10-m walking test used to assess mobility. This compared with the WVRT versus general balance training for visual perceptual disorders.

• This RCT showed no evidence of a difference between the two interventions (MD −0.12, 95% CI −13.62 to 13.38); evidence was judged to be of very low certainty.

FIGURE 25.

Active intervention 1 vs. active intervention for visual perception impairment. Outcome – mobility and navigation.

Perception

• Five RCTs (163 participants);60,144,152,154,237 Figure 26.

• A range of perception outcomes were used: Modified Taylor Complex Figure,154 MVPT,60,144,152,237 and Rivermead Perceptual Assessment Battery. 60

• Data were not pooled due to differences in the interventions but were displayed as standardised MDs. The evidence from each included RCT was judged to be of very low certainty.

FIGURE 26.

Active intervention 1 vs. active intervention for visual perception impairment. Outcome – perception.

Definitions of key terms (level of involvement: controlling)

Lived Experience Group members described the terms used to define perception as ‘complicated jargon’, saying that ‘it was like a complex vocabulary … that’s not really something I’ve come across…’.

Outputs included an agreed definition of perception (‘specific mental functions of recognizing and interpreting sensory stimuli’), lay definition of perception (‘processing and understanding information from the senses’), definitions of included senses, inclusion criteria in relation to specific (complex) disorders and detailed list of perceptual disorders.

Impact on the project: Definitions agreed by the stakeholders were applied throughout the review, including in the search strategy, selection of studies, data synthesis and interpretation of findings. Lists of disorders were used to inform the search strategy.

Despite the development of a lay definition of perception, definitions of each individual sense, provision of a glossary for each project activity, and a concerted effort by the research team to use simple language throughout, Lived Experience Group members repeatedly found it challenging to recall and to understand many of the terms when they were used throughout the project.

Outcome measure prioritisation (level of involvement: influencing and controlling)

There was general agreement between Lived Experience Group, Clinical Expert Group and researchers on the ‘Top 10’ most important outcome measures. However, there were variations in the specific rankings, with stakeholders placing greater priority on social activity and participation. The inclusion of outcome measures within the scoping review and the selection of outcomes for the Cochrane systematic review provided clear evidence that the stakeholders had an impact on the reviews; the use of specific outcomes within the review is summarised in Table 12.

Outputs: List of impact of perceptual impairment on daily life, prioritised list of outcome measures.

Impact of activities: A prioritised list of outcome measures was agreed. This directly informed the selection of outcome measures for inclusion in the scoping and Cochrane Reviews.

Stakeholders were pleased with their contribution to outcome measurement prioritisation ‘It’s good news that what we’re seeing is embedded in what’s been done’. They queried whether the reviews sufficiently addressed the emotional impact of living with perceptual problems after stroke: ‘I’m not sure that we explored the emotional impact, as much as perhaps we needed to, because it is very important….’.

Interpretation of review findings, implications and research recommendations (level of involvement: contributing, influencing and controlling)

Stakeholders expressed disappointment at the small number of studies, evaluating interventions which did not reflect ‘real-world’ experiences. They urged collaboration between clinicians and researchers to identify and evaluate clinical practice for patients with perception problems.

Outputs: Interpretation of the meaning of the scoping review and Cochrane Review findings, list of clinical implications, prioritised list of research gaps.

Impact of activities: The stakeholders raised several key issues relating to the meaning of the scoping review and Cochrane Review results. These points influenced the writing of the review findings and discussion. The stakeholders agreed lists of implications which have been incorporated into the findings from this project and reached a shared consensus on a prioritised list of research gaps (see Chapter 8).

Few stakeholder feedback forms were returned after these activities, perhaps due to frequency of project meetings and time commitments which may have been exacerbated by pandemic-related challenges (see Limitations). Despite the lack of feedback forms, stakeholders spoke passionately about the importance and potential impact of their input:

I really would like to think that something would come out of this study, in terms of just getting basic things at the beginning when somebody has a stroke …

Members of the Lived Experience Group emphasised that getting these ‘basic things’ right for people with perceptual disorders after stroke could have a big impact on an individual’s life, such as early assessment and diagnosis facilitating access to the right support, for example a bus pass for someone with visual issues struggling with mobility and transport.

Interventions for somatosensation perception disorders

We found low-certainty evidence of no difference between interventions and control on measures of ADLs, navigation and mobility. We found low-certainty evidence of no difference between two interventions for somatosensory perception dysfunction (not Pusher syndrome) on perception outcomes. For Pusher syndrome, we found low-certainty evidence of no difference between game-based posture training and standard physiotherapy for measures of mobility and navigation and perception. We also found low-certainty evidence suggesting that game-based posture training for Pusher syndrome may be more beneficial than standard physiotherapy for improving ADLs and measures of Pusher syndrome severity (see Table 13).

Interventions for touch perception disorders

We found low-certainty evidence that there is no difference between intervention and no treatment for navigation and mobility outcomes, but there may be a beneficial effect of active intervention on perceptual function. Evidence relating to one intervention versus another was varied, and insufficient to draw generalisable conclusions (see Table 13).

Interventions for visual perception disorders

We found low-certainty evidence of no difference between intervention and no treatment on measures of perception; there was no difference between active intervention and control for measures of EADLs. We identified some data for outcomes of ADLs, navigation and mobility and perception from RCTs comparing one intervention to another. Due to differences in the interventions and comparisons, the data were not statistically pooled, and it was not possible to draw generalisable conclusions (see Table 13).

Overall, there is insufficient evidence to determine the effectiveness of any one intervention for any sensory modality, nor the effect of one intervention relative to another.

Study eligibility criteria – selection biases

The PIONEER sequential review process had two research questions and different eligibility criteria. Conducted consecutively, the scoping review used very broad studies inclusion criteria to provide an overview of the topic area (see Chapter 4) while the Cochrane systematic review narrowed these selection criteria to focus on the RCT data (see Chapter 6). Both reviews were based on a thorough and up-to-date systematic search strategy developed by an information specialist and informed by research and clinical experts. Contact was made with leaders in the field and forward citation tracking was also used. Between the scoping and Cochrane Review, the search was updated to ensure that no emerging trials were omitted between the scoping review’s last search date and the start of the Cochrane Review process.

In the complex context of perceptual disorder research, we drew on existing definitions of perception (WHO)7,8 and working with our stakeholder groups, and these were expanded and operationalised to create subcategories by sense (visual, auditory, tactile, somatosensory, smell and taste) to support the planned reviews (see Chapter 3). We acknowledge that alternative definitions and categorisations could have emerged and in turn could have impacted on the identification of studies and the inclusion of additional or alternative research in the reviews. However, our consensus use of WHO definition7,8 was agreed by people with lived experience of perceptual disorders after stroke, and by multidisciplinary healthcare and research groups and thus were ideally suited to address our research questions. This ensured that all study eligibility decisions were consistently made across reviews, and thus unlikely to have introduced systematic limitations.

Both reviews used unambiguous, appropriate a priori eligibility criteria with no restrictions on the language, date or location of the study. The scoping review applied no restrictions to study design, perceptual disorders, interventions or outcome data considered eligible, reflecting the nature of the scoping review objectives. Qualitative study designs were eligible, but none were identified.

The Cochrane systematic review added three additional restrictions to the eligibility criteria used by the scoping review; only RCT data were included reflecting the Cochrane Review research questions, only adult participants were included, and a smaller number of outcome measures were included.

Both reviews focused on participants with stroke-related perceptual disorders; studies with a population where at least 80% had stroke-related perceptual disorders were also included. The scoping review included eight such mixed-sample studies [24 participants (2.6%) without a perceptual disorder or non-stroke-related perceptual problems]. Almost all (98.9%) of Cochrane Review participants experienced stroke-related perceptual disorders – one study included six participants with perceptual disorders following head injury. Throughout, we accepted the primary research diagnosis of perceptual disorders following stroke.

Data identification and selection – availability bias

In conjunction with a trained information specialist (JC) and informed by the stakeholder groups, we undertook detailed and exhaustive search of published and grey literature electronic databases (including hand searching) and forward/backward citation tracking (see Appendix 1 and Report Supplementary Material 9) to identify all relevant research to inform the scoping and subsequent Cochrane Review. The strength of this approach was the initial broad, inclusive, comprehensive search strategy supporting the scoping review. This was later narrowed using a RCT methodological filter to inform the Cochrane systematic review (see Appendix 3 and Report Supplementary Material 13). Application of methodological filters is known to increase the risk of omitting relevant studies,254,255 but our initial broader scoping review search guarded against this risk. All searches used free-text and subject index terms, reflecting the population, disorder and intervention. While it is possible that relevant research remained unidentified, we made every effort to identify it and include it in our reviews.

The data identified research conducted across the world, across disciplines and languages, over publication time points and stroke chronicity and across various subject topic areas and countries. Data eligibility criteria were applied by independent members of the research team at abstract and full-text stage. Grey literature was screened by a single member of the research team with decisions carefully checked by a second team member. Where disagreements were unresolved through discussion, a third team member or a Clinical Expert Group member was consulted.

Data collection and appraisal

Our structured Excel-based data extraction tool, underpinning the scoping review and Cochrane systematic reviews was independently piloted by two reviewers; the data extracted were compared and tool discussed and refined before it was used. All potentially relevant data were extracted. Data were extracted by one researcher and carefully checked by a second. Where clinical details were ambiguous, we involved a member of the Clinical Expert Group in decisions. Data synthesis, extraction items and categorisation used reflected published definitions. For RCTs, the data were extracted and directly entered into RevMan but were carefully checked by a second reviewer. The Cochrane Review risk of bias appraisal was undertaken by two independent researchers.

Data restrictions

The scoping review and Cochrane systematic review methodology supported a broad and inclusive review of the stroke-related perceptual disorder literature, but the data extraction was restricted to that reported in the literature. Research teams were only contacted where it was not possible to determine inclusion based on the published data. The subsequent Cochrane Review methods included some data gathered via direct contact with the included trialists (2 of the 18 included trials data reflect information gathered from unpublished communications) (see Chapter 7). One trial reported only partial outcome data234 (MOCA, nine-hole peg test and functional magnetic resonance imaging) and a second trial described collection of secondary outcomes, but data were not available. 74

Eleven trials were classified as awaiting classification. Due to a lack of clarity of reporting it was not possible to confirm that studies met inclusion criteria in relation to the disorder (n = 6), method (including randomisation process) (n = 2) or more than one inclusion criteria (n = 3). In each case we attempted to contact the author but received no reply.

Publication bias

Funnel plot analysis to assess the risk of publication bias is not typical in scoping review methodology, and there were insufficient data within the Cochrane Review to support this analysis (see Chapter 7). Generally, identified data sets were based on small sample sizes; predominately single-participant within the scoping review and only two RCTs randomising ≥ 50 participants, raising some questions about the quality of the data available.

Age of data sets

Of the data sets included within the scoping review, 63.7% (51/80) were published since 2010 with 34 (42.5%) of these published since 2015. Most trials (13 RCTs) included in the Cochrane Review were published in the last 5 years (2015–21); the remaining 5 were published between 2012 and 1985 and included the 2 largest trials in the review (n > 50).

Origin of data sets

The data sets included in the scoping and Cochrane systematic reviews reflect international research efforts. Research identified in the scoping review emerged from Asia, Europe, North America, South America and Australia. Within the Cochrane systematic review, the trials were conducted in Australia, Belgium, Germany, South Korea, Taiwan, UK and USA.

Data synthesis

Scoping review data were classified and organised by the perceptual disorder addressed and intervention assessed (subclassified by approaches used). Categories were pre-specified and agreed with the stakeholder groups as were the outcome measures of interest in the review. This was a complex task – difficulties in categorising interventions (especially where there was lack of reporting) were addressed by having a third reviewer, who was provided with clear definitions of intervention approaches, categorise all interventions. Further specialist input relating to Pusher syndrome interventions was provided by a somatosensory expert. Data were tabulated and presented using standard Word or Excel tables and an interactive visual map which were positively received by the Lived Experience Group members. The Cochrane Review interventions were organised in a similar manner, by perceptual disorder and intervention approach (with rehabilitation approaches subclassified).

In the Cochrane Review we followed pre-specified analyses as described in our protocol, registered with PROSPERO (CRD42019160270) and NIHR’s website [https://fundingawards.nihr.ac.uk/award/NIHR128829 (accessed November 2022)] and pooled trial data that compared active interventions with no treatment/control/alternative intervention. We used a random effects model (RevMan 5.4) to support statistical analysis. Our planned sensitivity analyses included exploration of the use of a fixed-effects model, but the lack of data prevented this and other planned sensitivity and subgroup analyses.

The populations represented in the scoping and Cochrane Reviews spanned the clinical trajectory of stroke-related perceptual disorders (acute to chronic) and we sought (but did not always identify) a wide range of perceptual disorders, interventions and outcomes. A small number of participants included in both the scoping and the Cochrane Reviews were recruited to mixed study populations; they did not have perceptual disorders or their perceptual disorder was not stroke-related. Given the overall lack of data availability and overlap of the available data, it is highly unlikely that an alternative approach to data synthesis would alter our findings.

Across the included trials, however, biases were evident in the RCTs, particularly in relation to the randomisation process, blinding and inadequate outcome reporting (described above). Most biases were related to inadequate reporting, though a third of trials were considered at high risk of detection bias due to an absence of outcome assessor blinding.

Equality, diversity and inclusion in the context of this study

Key factors known to be associated with underrepresentation in stroke research (especially trials) include female sex,272 ethnic minority background, age (> 80, < 18)273 and having stroke-related impairments in cognition274 and language. 275 In our reviews we sought to extract (and categorise) data on age, sex and concurrent impairments to both describe the population and identify areas of limited inclusion. Age and sex data were generally well described (see Participants included). We grouped age data into < 18, 18–65 and > 65 years old; 31.3% (25/80) studies had participants aged > 65 years suggested good representation of older stroke survivors; however, as this was not specific to those aged > 80 we cannot comment on any lack of representation in this group. We noted a lack of studies relating to children and under-representation of women. We did not collect data on race/ethnic background; on consideration this may have been a lost opportunity to explore this important issue.

We also collected data on whether studies reported any involvement of stroke survivors or carers in the design and conduct of their study (PPI): no studies reporting any PPI (see Stroke survivor or family involvement in research), possibly suggesting of a lack of consideration of stroke survivor’s experiences and priorities within perception interventions research.

Accessibility and Inclusivity when presenting results

We were aware of potential barriers to access arising from the complexity and unfamiliarity of perception and perceptual impairment language. We simplified wording in documents, provided glossaries of research and topic-specific terms and provided reminders of the meaning of key terms each time we met. Members of the Lived Experience Group helped to create a lay definition of perception that would be widely understood, and reviewed and edited all Plain Language Summaries for project publications, alongside input from a speech and language therapist research team member familiar with creating accessible versions of research information, to support accessibility. As some members of the Lived Experience Group had visual problems, we used clear print guidance throughout to guide the choice of font layout, use of colour and contrast, etc.

Research team

Our multidisciplinary research team reflected a range of topic and research methodology expertise to support the conduct of the study to a high standard (see Equality, diversity and inclusion in the context of this study).

Our involvement of stroke survivors, carers and clinicians was extensive, and is discussed in Stakeholder input and relevance.

Positionality

The core research team included multidisciplinary researchers with clinical expertise in optometry, audiology, physiotherapy and speech and language therapy, and neuropsychology (adult and paediatric) many of whom are stroke rehabilitation specialists and clinical academics in stroke care. All but one (SH) were based in Scotland. The core group’s methodological expertise includes extensive systematic review and data synthesis methodologies (Cochrane, mixed methods, network meta-analyses), complex interventions, priority setting, reducing research waste and maximising collaborative research efforts to the benefit of people after stroke. The core research team worked closely with the Lived Experience and Clinical Expert Groups [see Stakeholder activities (what happened)] who contributed their expertise from the broader perspectives of England (four members), international (one member) and clinical areas of audiology, ENT and visual neuropsychology.

Methodological rigour

We worked to the highest methodological standards. We built upon existing definitions to develop a clear, specific, working definition of perception at project outset and agreed our inclusion criteria in relation to the senses and specific disorders: given the complexity of this topic, this gave clear scope and practical guidance for succeeding activities. The WHO ICF8 definition underpinned our working definitions and was applied to all senses. This helped us address the challenge of perceptual terminology and reporting, with inconsistent terminology for similar conditions within and between senses, participant populations and disciplinary fields as well as lack of clarity on the precise nature of disorders. Our definition also informed our broad search strategy to identify published and non-published research, from both key electronic databases and a range of other sources including the grey literature. While we recognise that others may develop alternative approaches, our definitions and framework were feasible and were consistently applied across our reviews and supported the reporting of our findings. Thus, these definitions offer a transparent and replicable approach to support future perceptual disorder rehabilitation research.

We reported to the best practice and methodological and reporting guidelines wherever possible including TiDIER,97 Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA),128 PRISMA-A, PRISMA-ScR,117 GRIPP 2,108 Cochrane Review quality appraisal process for both studies (risk of bias)223 and comparisons (GRADE),226,227 ROBIS,253 and Cochrane methodological guidance for data synthesis. 222

As noted, prior research has often explored perception in conjunction with sensory, cognitive or attentional deficits. Our very clear focus on perceptual disorders alone is therefore unique. It has offered the opportunity to closely examine the research on this specific set of disorders, and we hope this provides a distinctive contribution to clinical care, enabling HCPs to better focus on and address perception-specific disorders.

Stakeholder involvement

We involved stakeholders from project inception and throughout the study (see Chapter 3), using current UK guidance and an established framework to plan and describe involvement. 103

We acknowledge that our stakeholder discussions were typically based on a small number of participants and somewhat subjective, meaning that discussions (and rankings) could have been biased by the experiences of participants and thus not reflective of a national or international viewpoint. Our involvement of stakeholders in the clinical and research implication sections brings substantial strength, reflecting the views of many people with lived experience of perceptual problems, and of HCPs addressing their care.

Lack of evidence

Stroke is the third leading cause of disability worldwide, and a growing number of stroke survivors are living with long-term disabilities, including perceptual disorders. The limited research activity relating to perceptual disorder interventions is important, as is the limited quality of the research identified (see Chapter 7). Consequently, there is a significant absence of high-quality evidence informing the provision of stroke care.

The reason for the lack of research in this area is not clear. The Lived Experience Group described how the considerable impact of their perceptual disorders was, for some, only appreciated months after their return home, when access to stroke rehabilitation was limited. An alternative interpretation may be that insight into such disorders may only emerge as other rehabilitation gains are made. Clinical experts suggested that people with severe stroke may have perceptual impairments masked by other conditions such as poor cognition or aphasia which in turn preclude formal assessments and limit self-report. Thus, clinicians may be unaware that a perceptual disorder is present, and why few research studies address the topic. Timely assessment, information provision and treatment should be important.

Our scoping review employed a broad and rigorous search of electronic databases and grey literature, adopting a comprehensive definition of perceptual disorders. Despite these efforts, due to the complex nature of the topic and terminology, some relevant papers may have been missed. In the absence of a universally accepted intervention categorisation, we utilised an existing method to support categorisation consistency, relevant to perceptual disorder research. 125 but which may not necessarily directly align with other categorisation approaches. 276 As a scoping review, we did not conduct quality appraisal and thus comment on quality or generalisability of study findings was not possible. Similarly, we failed to identify qualitative studies that focused on perceptual disorders after stroke; we recognise that some relevant data may have been available in more generic reports of stroke impact reports which may not have been identified in our comprehensive search of the literature. We are aware of one potentially relevant qualitative report published after our scoping review was completed,277 and of ongoing qualitative work in this field. 278

Study methodology

The scoping review identified a predominance of single subject and RCT designs, the former describing highly personalised treatments, making clinical relevance and validity difficult to establish. While RCTs can be a high-quality design approach, we found risk of bias concern in most RCTs included in our review. Small RCTs are typically insufficiently powered to generate confidence in conclusions and overlap in the small trials we identified was rare. A co-ordinated and systematic approach to enquiry in this research field is required and our scoping and Cochrane systematic review offer an important foundation on which to build this evidence base.

Reporting limitations

The TiDIER checklist,97 available since 2014, aims to support the replication of treatment, as reporting gaps have clear consequences for intervention implementation. We sought to extract the relevant intervention data using the TiDIER checklist, where possible describing the rationale, materials, procedures, intervention provider, mode of delivery, location, dosage and any modification of the intervention. Such details were not commonly available (see Interventions). Forty per cent (34/80) of the perceptual disorder intervention studies included in our scoping review were published after 2014 and thus intervention reporting could have benefited from use of this checklist.

Representation of senses

Evidence to inform interventions for some senses was lacking; interventions for hearing, taste and smell perceptual disorders and descriptions of their impact on life after stroke were not commonly reported. 28,43–45 Training and guidelines on this topic are limited3; however, recent care pathway guidance279,280 on the assessment and management of visual disorders after stroke gives greater impetus to improving care and the research needed to underpin this.

Paediatric populations

The paediatric population are particularly under-researched in relation to stroke-related perceptual disorders (see Visual perception disorder interventions), which reflects the generally limited paediatric stroke rehabilitation evidence base. 281 The paediatric and adult stroke populations are different. We lack agreement about the nature and extent282 of natural recovery following stroke among the paediatric population (due to neurodevelopmental plasticity);283 and where extensive recovery is expected, the need for management interventions is reduced. Greater evidence on the persistence of perceptual disorders and factors affected with this is a key area to inform further research for a paediatric population. Similarly, interventions designed for adult populations may not engage or suit paediatric populations leading to adherence challenges. Thus, paediatric perceptual disorder studies should be developed and conducted to inform their care and rehabilitation.

Under-researched populations

Most of the research identified in the scoping review related to adults and thus there is a need to address perceptual disorders experienced by children and young people. Similarly, further research on hearing, taste and smell perception disorders is required, particularly to establish the incidence, natural history, prevalence and impact of these.

In developing this research field, guidance on the development of stroke rehabilitation interventions and trials of their effectiveness are available to support the exploration of the mechanisms of action, and questions relating to dosage and target population,276,284. In addition, full consideration should be given to the context of care, and feasibility, sustainability and economic factors affecting intervention delivery; pragmatic designs should be considered in order to maximise the clinical relevance and applicability of emerging findings.

Involving stakeholders’ priorities

It is important that future research considers the needs of key stakeholders: stroke survivors, carers and families affected by perceptual disorders, and clinicians providing care. Researchers should aim to use a structured process to identify and engage stakeholders as fully as possible, ensuring that their priorities are considered and outcomes of importance addressed with interventions reflecting current practice in community-based settings.

Clarity and completeness of reporting

We suggest that researchers use relevant international reporting standards97,108,128 for populations, interventions and specific study designs, etc. and fully report key aspects of their research including

• Sufficiently detailed theoretical rationale for, and description of, the interventions to allow implementation into clinical practice and research replication.

• Detailed diagnostic information on individuals’ perceptual problems, given the heterogeneity in perceptual problems in terms of type, severity and likely impact on everyday function.

• Recording whether participants have concurrent impairments, including more than one perceptual impairment and/or other stroke and non-stroke relating impairments.

We further suggest the development of standardised terminology for perceptual disorders, to aid clarity or reporting and understanding for researchers, clinicians and stroke survivors, across all the senses. This could further help improve awareness, multidisciplinary identification and intervention for those affected.

Recommendations relating to randomised controlled trial design

Additional recommendations arising from the Cochrane Review and relating specifically to RCTs are:

1. Provide a standard care control group, carefully documenting the content and amount of standard care, which can be highly variable.

2. Endeavour to reduce risk of bias through rigorous methodology and reporting, for example ensure allocation concealment, attempt and report masking of outcome assessors, report all loss to follow‐up and results from all outcome measures, control for other biases.

3. Have the statistical power to answer clinically important questions about long‐term functional outcomes.

4. Include relevant outcomes, including economic analysis.

5. Adopt an intention-to-treat approach to measurement of outcomes in all individuals as well as to analysis of measured outcomes by treatment group.

6. Investigate real-world interventions, rather than solely novel technologies.