Non-pharmacological interventions for the improvement of post-stroke activities of daily living and disability amongst older stroke survivors: A systematic review

Globally, stroke remains a leading cause of death and disability, with older adults disproportionately affected. Numerous non-pharmacological stroke rehabilitation approaches are in use to address impairments, but their efficacy in older persons is largely unknown. This systematic review examined the evidence for such interventions as part of the Optimal Evidence-Based Non-Drug Therapies in Older Persons (ONTOP) project conducted under an European Union funded project called the Software Engine for the Assessment and Optimisation of Drug and Non-Drug Therapies in Older Persons (SENATOR) [http://www.senator-project.eu]. A Delphi panel of European geriatric experts agreed activities of daily living and disability to be of critical importance as stroke rehabilitation outcomes. A comprehensive search strategy was developed and five databases (Pubmed, CINAHL, Embase, PsycInfo and Cochrane Database of Systematic Reviews) searched for eligible systematic reviews. Primary studies meeting our criteria (non-pharmacologic interventions, involving stroke survivors aged ≥65 years, assessing activities of daily living and/or disability as outcome) were then identified from these reviews. Eligible papers were double reviewed, and due to heterogeneity, narrative analysis performed. Cochrane risk of bias and GRADE assessment tools were used to assess bias and quality of evidence, allowing us to make recommendations regarding specific non-pharmacologic rehabilitation in older stroke survivors. In total, 72 primary articles were reviewed spanning 14 types of non-pharmacological intervention. Non-pharmacological interventions based on physiotherapy and occupational therapy techniques improved activities of daily living amongst older stroke survivors. However, no evidence was found to support use of any non-pharmacological approach to benefit older stroke survivors’ disability. Evidence was limited by poor study quality and the small number of studies targeting older stroke survivors. We recommend future studies explore such interventions exclusively in older adult populations and improve methodological and outcome reporting.


Introduction
Globally, stroke remains a leading cause of death and disability, with older adults disproportionately affected [1][2]. While effective acute treatment has increased stroke survival within developed nations, increased survival increases the number of those affected by post-stroke impairments [2]. Therefore, effective rehabilitation which can reduce post-stroke impairment and restore a person's functional abilities is imperative.
Stroke guidelines recommend utilising multi-disciplinary stroke rehabilitation teams [3][4][5]. This reflects the diverse physical, psychological and social rehabilitation needs of stroke survivors [3][4][5]. Rehabilitation is primarily non-pharmacologic in nature, and standard approaches include occupational therapy (OT), physiotherapy (PT), and speech therapy [3][4][5]. Several factors contribute towards the overall success of stroke rehabilitation and include stroke severity, the type and location of a stroke, and the patient's general health and pre-stroke health [6]. The patient's age is also generally accepted to be highly influential; older patients are at a higher risk of poorer outcomes following stroke rehabilitation [7].
The evidence base for many of non-pharmacologic stroke rehabilitation interventions is poor. For example, the National Clinical Guidelines for stroke [4], despite recommending that psychological care be offered to all stroke survivors, also describe the evidence behind many psychological therapies (e.g. Cognitive Behavioural Therapy, Counselling) as conflicting and inconclusive. Furthermore, the effectiveness of such interventions within the older stroke population is even less clear. Much of the literature reports upon trials involving participants below 65 years of age [8]. Older adults may differ from younger adults in terms of their rehabilitation needs and preferences [8]. Ageing brings more challenges; older stroke survivors often have higher pre-and post-stroke disability and impairments, some of which can be explained though the natural ageing process [9]. This may make rehabilitation more challenging, limiting benefits from rehabilitation attempts.
While non-pharmacological approaches to treat post-stroke impairments are predominant, they are also preferred for older patients. Older people have an increased risk of adverse drug reactions [10]. Additionally, many drugs commonly prescribed to older people have not been assessed in an older population [8]. Therefore, pharmacologic agents used to treat some poststroke impairment, such as muscle spasticity and movement disorders, are unlikely to have been adequately tested in older patients and so their safety amongst this population group is unknown.
There are compelling reasons behind treating common conditions using non-drug therapies amongst older persons. The aim of this systematic review was to identify and review the evidence for such interventions as applied to older stroke survivors. This systematic review is part of the Optimal Evidence-Based Non-Drug Therapies in Older Persons (ONTOP) project. The ONTOP project aims to systematically review 15 of the most prevalent and difficult to manage conditions in older people and produce a list of recommendations concerning the use of non-drug therapies for these conditions [11][12]. Many of these reviews have been completed, including for pressure ulcer risk reduction and treatment [13] reduction in incidence and treatment of delirium [12] and fall prevention [14]. ONTOP is in turn part of a larger, European Union (EU) funded project called the Software Engine for the Assessment and Primary studies.
• All participants must be !65 years of age, or the mean age of participants must be !65 years of age • All aetiologies, types and severity of stroke/ stroke symptoms included • Involves any non-pharmacological intervention for stroke: a. a single or multi-component non-drug intervention used to improve symptoms post-stroke b. a non-drug intervention being a treatment or therapy that can be performed on or given to a patient, and/or taught to the patient for them to practice themselves.

c. A non -drug intervention which is deliverable in clinical practice
• Treatment for any complications or specific disability of stroke (e.g. urinary incontinence, shoulder subluxation, neglect syndrome etc.) will be included if the study reports !1 relevant outcome • Compares the non-pharmacologic treatment against no treatment, a sham intervention or a treatment considered standard practice at the time of the study.
• A study using Randomised Controlled Trial (RCT) or Quasi RCT methodology • Paper must focus on at least one or more of three Delphi consensus derived outcome variables: ADL, quality of life or disability (global measures only).

• Papers published only in English, Italian and Spanish
Exclusion criteria Primary studies.
• Any therapy for stroke prevention • Any therapy using non-conventional products but administered in a conventional route (e.g. Chinese medicine, herbal supplements) • Observational or before-after studies with historical controls • The inclusion of participants with other neurological conditions • Studies exploring the management of stroke in critical care/ Accident &Emergency • Health services research evaluating the two different stroke units (hospital based, community or home-based), two or more different methods of delivering non-pharmacological therapy (e.g. face to face or telephone rehabilitation), or evaluating different methods of delivering/ co-ordinating discharge care (e.g. named person in charge of discharge/ post-discharge care versus usual care) • Economic evaluations of non-pharmacological therapy • Papers discussing the dose-response relationship (duration, intensity of therapy or time to commence treatment, including early discharge) • Interventions which only involve the provision of education/ stroke information and general sign posting/ liaison with other services where the patient plays a passive role (NB: If these components are included in a broader structured multi-component intervention such as a self-management programme the intervention will be included).

Study selection
For this review, 18,932 potentially relevant articles were identified from database searches ( Fig  2). After removing duplicates, 13,627 unique records were screened by title and abstract by two reviewers. Only 363 full texts of systematic reviews were deemed eligible based on their abstracts. Of these, 173 reviews matched the eligibility criteria and were read in full, and their references were hand searched to identify potentially relevant primary studies. The initial searches were conducted in December 2015, with no restrictions on publication date, and resulted in 83 primary articles for inclusion. The review was updated as above in April 2018 and a further six papers were added to the findings.

Data collection
The results of the database searches were amalgamated using Refworks 6.0 software (ProQuest LLC, USA). A list of the titles and abstracts of systematic reviews were screened by two independent assessors (EG, CS). Any disagreements over eligibility were resolved through discussion with other members of the research team (RS and PKM). The full-text articles of potentially eligible reviews and meta-analyses were then retrieved and assessed for eligibility, again by two independent assessors (EG and CS). The references of the included studies in eligible systematic reviews were hand-searched to identify primary studies relevant to this review. A list of the titles and abstracts of potentially eligible primary studies was screened (EG, CS, SS, RS and PKM). Thereafter, the full-text articles of potentially relevant primary studies were retrieved and screened by EG and CS.

Data extraction
A data extraction form was designed by adapting the Cochrane Collaboration's Data Extraction and Assessment Template. The information contained on the data extraction forms (study methodology, participant characteristics, and outcome data) was then transferred to an Excel spreadsheet for narrative analysis. Results were also transferred to RevMan 5.3 [Cochrane Collaboration, UK, http://community.cochrane.org/help/tools-and-software/ revman-5] to facilitate risk of bias tables. Results were also transferred to the GRADE Pro online system [http://www.gradeworkinggroup.org] for the development of recommendations for each type of non-pharmacological intervention. Types, or categories, for non-pharmacological interventions were developed and applied to organise the included studies into meaningful categories of interventions for the analysis. Data extraction was performed by two independent assessors (CS & EG).

Risk of bias
Risk of bias was assessed using the Cochrane Collaboration's Risk of Bias tool [18]. This tool assesses: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other biases. A decision was made as to whether the risk of bias for each category should be described as low, unclear or high risk. The overall risk of bias for the study was then judged by taking account of the scores for each individual category. Results from the risk of bias assessment were entered into RevMan 5.3 software to enable the production of risk of bias graphs and summary tables.

Development of PICO questions
Clinical questions were formulated using the PICO (Population, Intervention, Comparator, and Outcome) framework for each intervention type and outcome assessed. Due to the small number of papers in each category of intervention, the PICO questions chosen were considered to be the most pertinent and inclusive questions. For most categories of intervention one question assessing the efficacy of intervention types upon each outcome was chosen. As physiotherapy and occupational therapy are often standard care in stroke rehabilitation, studies investigating these therapies did not have a no intervention control. Therefore, we split physiotherapy and occupational therapy studies depending upon whether they compared a more intensive (increased time and duration) of therapy against usual intensity, or if they compared two or more different forms of therapy.

Narrative analysis
All primary studies were included in a narrative assessment. The effects reported in each study were described as favouring the intervention, favouring the control, or as showing no significant difference. The overall findings of the studies were assessed qualitatively considering methodological quality and risk of bias. Patterns of effect across the studies were described and possible reasons for effect differences between studies explored, as per guidance offered by the ESRC [19]. Due to substantial clinical heterogeneity between studies and poor study reporting, meta-analysis of results was not considered appropriate. Clinical heterogeneity was assessed qualitatively by all authors and focused upon intervention content, target (e.g. upper or lower limb impairment), delivery, duration.

Assessing quality of evidence
After the completion of analysis, evidence for each non-pharmacological category was assessed using the GRADE method [16]. The GRADE approach assesses the evidence across all studies analysed for a given outcome, rather than assessing the evidence from each study individually. The GRADE framework allows the quality of the body of evidence, and consequentially any recommendations to be made from this evidence, to be judged across five criterions known to limit the quality of evidence. Further details regarding each of these criteria can be found on the GRADE website [http://www.gradeworkinggroup.org]. The quality of the evidence was assigned an overall rating of quality, as described below in Table 1.

Development of recommendations
After the quality of evidence had been determined, concise recommendations were designed regarding the use of non-pharmacological therapies after stroke in older persons. These recommendations were written taking account of the quantity, quality and GRADE score of the available evidence.

Acupuncture
Studies. Eleven randomised controlled trials (RCT) were included in this category. Five were conducted in China, two in Sweden, two in the UK and one each from Taiwan and Germany.
Participants. In total, 1064 participants were involved. Mean ages ranged from 65.5 (SD 9.71) years [20] to 78.3 (SD 5.9) years [21]. Time between stroke onset and commencing intervention ranged from a mean of 14.2 (SD 19.2) hours [22] to 14.4 (SD 7.8) days [20]. Participant's characteristics across the included studies are presented in Table 2. Interventions. Interventions varied in their design (for example number of points used or whether manual or electrical stimulation was applied) and in their duration. Intervention descriptions are summarised in Table 2.
Risk of Bias. Eight (72.7%) studies were at risk of bias arising from unblinded or inadequately blinded participants. Many of the studies were unclear regarding their methods of randomisation and allocation concealment. What is the effectiveness of acupuncture (traditional or electro) upon older stroke survivors ADL recovery in comparison to either usual rehabilitation care without acupuncture or sham treatment?. Of the 11 trials investigating the impact of acupuncture upon older stroke survivors ADL recovery (see Table 3), three studies demonstrated statistically significant benefit upon ADL scores favouring acupuncture [25][26][27]. Pei et.al. (2001) also found limited short term support for the use of acupuncture for ADL recovery, but any significant benefit had disappeared by week 2 post-intervention [29].
It should be noted that those reporting significant findings in relation to ADL recovery following acupuncture tended to be smaller trials. Quality assessment, using the GRADE approach identified the overall body of to be low (see Table 3) due to risks imposed by various biases and the heterogeneity between intervention content, delivery and duration. This means that further studies are very likely to have an important impact on the estimate of effect. Overall, the evidence does not support the use of acupuncture after stroke.
What is the effectiveness of acupuncture upon older stroke survivors' disability recovery in comparison to usual rehabilitation care without acupuncture or sham treatment?. Liu et al (2016) was the only trial to assess the effect of acupuncture upon stroke survivors' disability recovery [22]. In their small trial (n = 38) no significant differences (p = .15) in mean Systematic review of non-drug therapy for older people with stroke modified Rankin scale scores were identified between those receiving acupuncture (1.13, SD 1.25) and those receiving usual care only (1.27, SD 1.16) at 12 weeks post-intervention. A GRADE assessment found the evidence to be moderate but due to the limited number of studies, and the small sample size, further studies are likely to impact upon the expected effect identified in this review. We therefore cannot recommend acupuncture to improve older stroke survivors post-stroke disability.

Caregiver training
Studies. Only one study investigated the effect of caregiver training upon older stroke survivors ADL recovery. The RCT [31] was conducted within an inpatient rehabilitation unit in one UK hospital.
Intervention. Caregiver training consisted of three to five 30-45 minute sessions of instruction. Sessions covered common stroke related problems, their prevention and management, and included hands on training in moving and handling, mobility encouragement, transfers, and speech/ communication. Sessions were conducted in the hospital whilst the participant was an inpatient. One final session was delivered to the caregiver in the participant's home environment following participants discharge. The control group participants received usual care only.
Risk of bias. A lack of blinding represents the most significant risk of bias for this study Can pre-discharge caregiver training influence post-discharge stroke survivor ADL recovery in comparison to those who receive no caregiver training?. One study, by Kalra et. al. (2004), investigated if caregiver training could influence ADL recovery [31]. Twelve months post intervention, the intervention group scored a median Frenchay Activities Index (FAI) score of 15 (IQR 9-23) versus the control participants median of [16][17][18][19][20][21][22]. The difference between groups on this measure was not significant. Using the Barthel Index (BI), presented as the number of persons considered 'improved', classified by BI score >18 at the measurement point, 51% (n = 77) of the intervention participants at three months were considered 'improved' versus 35% (n = 52) of the control participants. The difference in the number of persons 'improved' was found to be significant (p = .007). At 12 months, 62% (n = 93) of the intervention participants were considered improved, versus 50% (n = 75) of the control participants. However, the difference between groups at 12 months was not significant (p = .074); improvement in ADL recovery was not sustained over time. Pre-discharge caregiver training may offer some benefit to older stroke survivors ADL recovery in the short term. However, only one study investigated this type of intervention and actual BI scores were not presented. GRADE quality assessment was low. Insufficient evidence means we cannot make a recommendation for this type of intervention. There is, however, evidence warrants further investigation as to the potential benefits of such an intervention.
Can pre-discharge caregiver training effect post-discharge stroke survivor Disability scores in comparison to those who receive routine post-discharge advice with no caregiver training?. Only one study [31] investigated if caregiver training could influence participant disability scores. Disability was assessed with the modified Rankin score, presented as the number of persons considered 'improved' as classified by a Rankin score of 0-2. At three months, 53% (n = 80) of intervention participants, versus 42% (n = 63) of control participants, were considered improved. At 12 months, 66% (n = 100) of intervention participants, versus 58% (n = 87) of control participants, were considered improved. However, differences between groups were not significant at either time point (p = .054, p = .18). Very limited evidence from one RCT, with a GRADE assessment of low, suggests no benefit to older stroke survivors' disability arising from caregiver training.

Constraint induced movement therapy
Studies. Two RCTs exploring the benefits of Constraint induced movement therapy (CIMT) upon older stroke survivors were identified [32][33]. Both were conducted using rehabilitative outpatient departments, but one was conducted in Taiwan [33] and the other in China [32].
Interventions. Wu et. al. (2007) employed a modified CIMT technique (n = 13) where subjects placed unaffected hands in self-adhesive strapping for six hours per day while at home [33]. In addition, participants received two-hour sessions of CIMT with a therapist for five days per week. Sessions, provided after participants' regular OT appointments, focused on carrying out basic ADLs with the affected arm while the unaffected hand was constrained. This was compared against a traditional rehabilitation programme (n = 13) that focussed on practising ADLs but without constraining the affected limb.
(2016) compared self-regulated modified CIMT (n = 30), modified CIMT (n = 30), and a control group (n = 30) employing conventional rehabilitation [32]. Self-regulated CIMT participants were encouraged to reflect on their abilities and come up with their own solutions to problems, rather than follow instructions. Modified CIMT participants received therapist feedback and improvement suggestions during arm restraint training sessions. Each group practised 10 tasks daily, through graded and progressive exercises, over two weeks, for one hour per day, five days per week.
Risk of bias. Risk of bias from both studies is possible due to unblinded participants and lack of clarity surrounding allocation concealment.
What is the effectiveness of CIMT upon older stroke survivors ADL recovery in comparison to those receiving conventional rehabilitation only?. Wu et. al. (2007) reports there to have been no significant difference between the groups (p = .018) at 3 weeks follow-up [33]. Lui et. al. (2016) report that participants receiving self-regulated modified CIMT had significantly higher Instrumental Activities of Daily Living (IADL) scores at week 2 (p < .001) than those receiving modified CIMT alone or conventional therapy [32]. However, this difference had disappeared at 1 month follow up (p = .51). The limited number of studies, combined with a low GRADE quality assessment means that the evidence base is inadequate to conclude CIMT is effective for older stroke survivors ADL.

Device assisted physiotherapy
Studies. Eight RCTs exploring device assisted physiotherapy were identified. Two each were conducted in France and Italy, and one each from the USA, China, Sweden and the UK.
Interventions. Interventions varied in their content and duration and are described in Table 4. We subdivided trials into those using either robotic or non-robotic devices.
Risk of bias. All studies were at risk of bias resulting from unblinded or inadequately blinded participants, although most had blinded assessors.
What is the effectiveness of a robotic physiotherapy device upon older stroke survivors ADL recovery in comparison to those receiving conventional physiotherapy?. Two studies The control occupational therapy group received group therapy led by a certified occupational therapist.
One 40-minute session per day, 5 days per week, up to 12 sessions.

NR (Continued)
Systematic review of non-drug therapy for older people with stroke [38,40] assessed effectiveness of robot-assisted therapy versus usual care (presented in Table 5). Maseiro et. al. (2007) compared a sensorimotor robotic training programme against usual care [38]. Although intervention participants had a significantly better Functional Systematic review of non-drug therapy for older people with stroke Independence Measure (FIM) score at 6 weeks than control participants, its benefit was not sustained at three or eight month follow-up. The findings of Rabadi et. al. (2008), in their comparison of arm ergometer training, the MIT-MANUS robotic trainer, and usual care, also suggest that additional physiotherapy devices offered no further benefit than usual rehabilitative care [40]. The substantial heterogeneity, combined with a serious risk of bias due to lack of methodological reporting, resulted in a downgrading of the quality of the evidence to very low (see Table 5).
What is the effectiveness of a non-robotic physiotherapy device upon older stroke survivors ADL recovery in comparison to those receiving conventional physiotherapy?. Six studies investigated the impact of non-robotic physiotherapy devices upon stroke survivors ADL recovery (presented in Table 6). Only one of the six studies, Wiart et. al. (1997), reported a moderate significant benefit favouring the non-robotic device at day 30, versus usual care control (p < .03) [41]. That said, there appears to be an imbalance between baseline scores of the two groups and there is no report if this was significant, nor any reporting of the difference between the change in means between groups. The quality of evidence of these studies was rated as very low (see Table 6), owing to a serious risk of bias and small sample size.

Music therapy
Studies. One RCT, conducted in Italy, investigated the role of music therapy in the treatment of older stroke survivors ADL.
Participants. The study [42] involved 38 participants, 42.1%(n = 16) of whom were male. The mean age of participants in the intervention arm was 70.4 (SD 8.9) years versus 75.4 (SD 7.6) of control participants. All participants commended intervention within six to eight weeks of stroke onset.
Intervention. Intervention participants received Relational Active Music Therapy, conducted by trained musical therapists. Participants were encouraged to use rhythmical instruments during these sessions, which were provided three times per week and lasted for around 30 minutes per session. Participants received up to 20 sessions in total. Control participants received no additional intervention.
Risk of bias. For most bias types, this study was rated as being unclear due to insufficient reporting Can music therapy effect stroke survivors ADL recovery against usual care alone?. Both intervention and control participants in this small study improved over time (p < .001) but no significant difference between groups final scores or change in scores from baseline were identified [42]. Music therapy participants improved their mean FIM scores from 76.58 (20.35) at baseline, to 110.47 (9.9) at follow up. Similarly, control participants improved their FIM scores from 71.26 (19.33) to 106.89 (16.83). The inclusion of only one small study, with an unclear risk of bias, which demonstrated no improvement in ADL, means that we are not able to recommend the use of music therapy to improve ADL recovery amongst older stroke survivors. The evidence has been given a GRADE quality assessment of low, meaning that further studies are very likely to change the effect estimate.

Nerve stimulation
Studies. Three RCTs presented findings in relation to nerve stimulation devices and ADL recovery amongst older stroke survivors. Two studies were undertaken in Sweden, and one study undertaken in the United States of America (USA).

Participants.
In total, 232 participants were randomised, of which 62.5% (n = 145) were male. Table 7 presents a summary of participant characteristics.
Interventions. The three studies varied in their type of nerve stimulation, location of bodily impairment targeted and duration of treatment. Table 7 presents a summary of each intervention.
Risk of bias. Of the three studies, two studies had a lack of or inadequate blinding procedures. Insufficient reporting to clarify risk of several other bias sources resulted in a number of bias assessments being unclear. Can the use of nerve stimulation devices influence older stroke survivors ADL recovery in comparison to those who receive usual rehabilitation care or sham treatment only?. Three studies explored the efficacy of nerve stimulation device use upon older stroke survivors ADL recovery (see Table 8 for study results). Johansson et. al. (2001) and MacDonell et. al. (1994), who both studies participants in the acute phase of stroke recovery, found no significant difference in BI scores between those receiving nerve stimulation and those receiving usual care [43][44]. Conversely, Sonde et. al. (1998;, in their 3-year follow up study involving chronic post-stroke survivors, suggests that Trans-electrical Nerve Stimulation (TENS) may not result in significant improvements in ADL immediately following intervention but may allow older stroke survivors to better maintain ADL scores in the 3 years following stroke [45][46]. While both the intervention and control groups ADL scores declined from 3 months to the 3 year follow up, the control groups reduction in ADL was significantly greater than those who received the TENS intervention [45][46].
The evidence for the use of nerve stimulation for ADL recovery in older stroke survivors is limited (see Table 8). Nerve stimulation does not appear to improve ADL in the immediate term, but may offer protective benefits over future decline in ADL in the years following stroke. Quality assessment suggests the evidence to be very low which means that further Systematic review of non-drug therapy for older people with stroke studies will likely have an impact on findings. At present, it is not possible to recommend the use of nerve stimulation in the treatment of older stroke survivors to enhance ADL recovery.

Occupational therapy
Studies. Twelve RCTs were included. Eight were conducted in the UK, with one each from China, Holland, Canada and Italy.
Participants. In total, 1632 participants were randomised, of which 49.6% (n = 810) were male. Mean ages ranged from 65.9 (SD 8.16) years [47] to 88.6 (SD 6.5) years [48]. Most studies did not report time between stroke onset and commencement of intervention. Those who did are presented in Table 9 alongside other participant characteristics.
Interventions. Interventions varied widely in their content and duration and are summarised in Table 9.
Risk of bias. Ten of the twelve studies involved unblinded or inadequately blinded participants, and two studies had inadequate assessor blinding.
Does increased occupational therapy intensity influence older stroke survivors ADL recovery against no occupational therapy or usual occupational therapy care?. As summarised (Table 10), only four of the ten RCTs reported any positive impact upon ADL scores from an occupational therapy (OT) intervention [49,52,54,58]. In the study by Chiu et. al. (2004) both control and intervention participants mean FIM scores improved and the difference in mean change scores was significant (p = .001), favouring the additional occupational therapy [49]. However, no follow up beyond the end of intervention was conducted. Gilbertson et. al. (2000) found significant improvement in mean ADL scores between their control and intervention participants at eight weeks post-intervention, however these differences were not maintained at six months [52]. Conversely, Walker et. al. (1999) found significant improvement in ADL scores at six months post-intervention, favouring the home-based occupational therapy programme [58]. For the studies which were unable to demonstrate significant benefit, there was little, if any, change in ADL scores, regardless of outcome measure used.
In consideration of several risks of bias, heterogeneity between trials, and mixed findings, the quality of these findings is considered low using the GRADE rating system (see Table 10). This review proposes that increased OT may be beneficial regarding ADL and so should therefore be considered for older stroke survivors. Systematic review of non-drug therapy for older people with stroke Systematic review of non-drug therapy for older people with stroke

What is the effectiveness of alternative occupational therapy techniques upon older stroke survivors ADL recovery against traditional ADL based occupational therapy?.
Regarding comparing alternative OT approaches, two studies presented relevant ADL data (Table 11). Jongbloed et. al. (1989) found no statistically significant differences between sensorimotor occupational therapy and ADL-based occupational therapy [53]. Donkervoort et. al. (2001) found a statistically significant difference between groups at eight-week follow-up (p<0.01) favouring strategy training over ADL training [51]. However, by five months the different no longer reached significance (p = .11). Very limited evidence and a low GRADE assessment score (see Table 11), means that we are unable to recommend one OT approach above another in relation to older stroke survivors ADL recovery. What is the effectiveness of additional task-specific occupational therapy versus usual occupational therapy upon older stroke survivors disability scores?. Only one RCT, Parker et.al. (2001), explored the impact of additional OT upon disability scores [57]. They compared usual OT rehabilitation against participants receiving additional task specific training in either leisure activity engagement or self-care activities. Using the Oxford Handicap Scale, measured at six and twelve months post-intervention, the authors found no significant differences between the groups at any assessment (p-values not reported) [57]. Due to unblinded Systematic review of non-drug therapy for older people with stroke participants, the evidence was downgraded to moderate using the GRADE approach. At present, we are unable to recommend the use of additional OT to improve older stroke survivors post-stroke disability scores

Optical
Studies. Three RCTs present findings in relation to the use of interventions designed to address visual neglect experienced by older stroke survivors. Studies were undertaken in the U.K. (n = 1), Japan (n = 1) and China (n = 1).
Risk of bias. Two of the three studies had no or inadequate participant blinding, although all three had adequate assessor blinding. Additionally, each of the three studies methods of allocation concealment were inadequately described.
Can optical interventions which target stroke related visual neglect influence stroke survivors ADL recovery in comparison to those receiving conventional rehabilitation only?. All three optical intervention studies measured ADL as an outcome (results presented in Table 13) [59][60][61]. However, not one of the studies reported significant benefit favouring the optical intervention. Overall, there is no evidence to support the use of interventions targeting visual neglect to improve ADL recovery amongst older stroke survivors. Concerns regarding bias, and the heterogeneity of the trials, results in the quality of evidence being considered very low (see Table 13). According to GRADE, this means that further studies are very likely to change the estimated effect. At present, the use of these approaches cannot be recommended.

Physiotherapy
Studies. Seventeen RCTs presented findings in relation to physiotherapy (PT) interventions designed to improve ADL and/or disability recovery of older stroke survivors. Seven  [63] did not present participants sex information). Participant characteristics are presented in Table 14 . Systematic review of non-drug therapy for older people with stroke Systematic review of non-drug therapy for older people with stroke  Table 14 .

Interventions. Intervention content, delivery and duration varied widely between studies and each intervention is summarised in
Risk of bias. Almost all studies (n = 16) were at risk of bias from unblinded or inadequately blinded participants. This said, most studies (n = 15) had adequate outcome assessor blinding. Several studies were at potential risk from biases resulting from randomisation or allocation methods.
Does increased physiotherapy influence ADL recovery of older stroke survivors in comparison to those who receive usual rehabilitation care only?. As summarised in Table 15, of the 10 studies addressing this question, only three reported a significant benefit upon ADL Systematic review of non-drug therapy for older people with stroke Systematic review of non-drug therapy for older people with stroke  [65,67,77]. Each of these three studies identified this benefit only at an intermediate timepoint. Duncan et.al. (2003) found a significant difference in BI scores at three months favouring PT, but not at six months [65]. Similarly, Galvin et.al. (2011) and Sivenius et. al., (1985) also reported results favouring increased PT at in the intermediate term (8 weeks and 3 month respectively), but again differences were not maintained in the long term [67,77]. The degree of heterogeneity between interventions, combined with a number of sources of bias across included studies, resulted in the quality of evidence for this intervention being downgraded to low (see Table 15). According to GRADE further studies are very likely to change the effect estimate. In view of the limited evidence, this review proposes that increased PT may be beneficial regarding ADL recovery in the short term and so should be considered for older stroke survivors.
What is the effectiveness of specific physiotherapy approaches versus alternative physiotherapy approaches or usual care upon older stroke survivors ADL recovery?. Nine studies provided results to address this question, summarised in Table 16. Only one study [71]  6mths: 14  (Continued) Systematic review of non-drug therapy for older people with stroke Systematic review of non-drug therapy for older people with stroke reported any benefit favouring an alternative PT approach, and the benefit was limited. In their three-arm trial, Kwakkel et. al. (1999) reported a statistically significant difference between additional arm training and leg training, favouring leg training, at the 20 week followup (p<0.05). However, no statistically significant differences between the three groups at 26-week or one-year follow-up were found [6]. There is little evidence to suggest that one PT technique benefits stroke survivors ADL recovery more than an alternative technique. GRADE assessment of quality suggests the overall evidence base as being low (see Table 16), largely as a result of substantial heterogeneity between intervention content. Therefore, it is not possible for this review to make any recommendations regarding specific PT approaches to enhance older stroke survivors ADL recovery.

Psychological therapies
Studies. Six RCTs explored the use of psychological therapies amongst older stroke survivors, with three conducted in the UK, one in Holland, one in Australia and one in the USA.

Risk of bias.
All studies had a risk of bias arising from either unblinded participants. However, all reported adequate assessor blinding, and all but one had a low risk of bias arising from randomisation or allocation methods.
Can psychological therapies influence older stroke survivors ADL recovery in comparison to those who receive usual rehabilitation care or sham treatment only?. Only one of the six studies addressing this question reported a significant improvement in ADL (see Table 18). In the study by Clark et. al. (2003), intervention participants who received counselling sessions from a trained social worker had a significantly greater improvement in ADL score than control participants at six month follow up [84]. It should, however, be noted that although statistical significance was reached, the difference between groups is arguably small and clinical significance questionable.
At present, the evidence was assessed by GRADE to be of low quality and does not support a recommendation for the use of psychological therapies to improve ADL recovery in older stroke survivors.
Can psychological therapies affect post-stroke disability in older stroke survivors in comparison to those who receive usual rehabilitation care or sham treatment only?. Only one study, Lincoln et. al. (2003), explored the use of CBT upon post stroke disability amongst older stroke survivors [81]. The three arm trial compared CBT, a sham talking treatment and usual care [81]. Using the London Handicap Scale (LHS), the study identified no significant difference in LHS scores between the groups at baseline, 3 or 6 months. The evidence was assessed by GRADE to be of low quality and therefore at this time, there is no evidence to recommend the use of CBT in older stroke survivor

Self-management education
Studies. Six RCTs presented findings in relation to self-management education interventions targeting either older stroke survivors ADL recovery or disability. Four were conducted in the UK and one each from Sweden and Israel.
Participants. In total, 1012 older stroke survivors participated in these trials, of which 531 (52.5%) were male. Participant characteristics are summarised in Table 19. Systematic review of non-drug therapy for older people with stroke Interventions. Each intervention focused upon providing education and developing selfmanagement skills and plans, but their content and mode of delivery varied, as described in Table 19.
Risk of bias. Each RCT had at least one significant risk of bias, most commonly from unblinded or inadequately blinded participants. Several studies also were at high risk of bias arising from their randomisation and allocation methods. Systematic review of non-drug therapy for older people with stroke

NR
Begins with a small group education session for inpatients and their carers and is followed by six sessions at home. Programme aims to improve knowledge of stroke, treatments, and services, provide advice and an opportunity to ask questions and gain support. Led by one member of multidisciplinary team with input from nursing, PT, OT, speech/language therapy, psychology, social work, carers, and a stroke club.  Table 20. Only one of the six studies, Nir et. al. 2004, identified a significant improvement in FIM scores following the intervention against a control group receiving usual care [92]. However, this study also measured ADL using the IADL questionnaire and found no significant difference between groups at any follow-up.
In summary, while several studies have explored the impact of self-management education interventions upon post-stroke ADL recovery, there is very little evidence to support their use. The evidence was considered to be of low quality and therefore we cannot recommend these interventions to benefit older stroke survivors.
Do self-management education interventions influence post-stroke disability scores of older stroke survivors in comparison to those who receive usual rehabilitation care?. Two studies explored the impact of self-management education interventions upon post-stroke disability score. One study, Rodgers et al 1999, did not report any original data but we are informed that there were no significant differences between the disability scores of the intervention and control participants [93]. In the study by Smith et al (2004), both intervention and control groups were found to improve their level of disability post-stroke, as measured by the London Handicap Score, but the difference between those who participated in the self-management education intervention, and those who did not, was not significant [94]. There is presently no evidence to support the use of self-management education programmes for older stroke survivors to improve post-stroke disability. An assessment of the quality of these studies, using the GRADE approach, rates the overall quality as very low. Principally, this is due to small sample size, heterogeneity between the RCTs and risk of bias.

Videogames
Studies. One RCT investigating the role of videogames in the treatment of older stroke survivors was identified. This study, reported by Lee et. al. (2013) [95], was conducted in Korea, but is considered at high risk of bias due to a lack of reporting study methodological information.
Participants. The trial randomised 14 participants, 9 (64%) of whom were male. The mean age of the intervention group was 71.71 (SD 9.14) years, and the control group 76.43 (SD 5.8) years. The intervention group had a mean time between stroke onset and commencement of intervention of 7.29 months (SD 1.38), in comparison to the control group mean of 8.29 months (SD 3.4).
Intervention. The 6-week video gaming intervention involved participants being asked to choose two games on an Xbox Kinect games console to play while sitting or standing. The games were played in a separate room without distractions. The games were designed to complement conventional OT therapy and participants asked to participate in these 60-minute sessions three times per week. Control participants received usual rehabilitation care only.
Risk of bias. For almost all bias types, this study was rated as being unclear due to insufficient reporting.
Can the use of video games in addition to conventional occupational therapy delivered within inpatient rehabilitation care influence stroke survivors ADL recovery against usual occupational therapy alone?. The small study (n = 14) by Lee et. al. (2013) found that both intervention and control groups significantly improved their FIM scores at the post-intervention assessment (Intervention group post-test: 71.42, SD 15 v control group post-test: 61.24, SD 11.9), but no significant difference between the groups in relation to their degree of Systematic review of non-drug therapy for older people with stroke improvement or final scores were identified (p values not given) [95]. Based upon one small study (n = 14), with a high risk of bias, which demonstrated no benefit, means that we are unable to recommend the use of videogames to improve ADL recovery amongst older stroke survivors. Using the GRADE system the results suggest the evidence is of low quality, meaning that further studies are very likely to change the effect estimate.

Wheelchair
Studies. Only one RCT, conducted by Barrett et. al. 2001, was identified which investigated if self-propulsion of a wheelchair improved ADL recovery against non-self-propulsion [96]. It was conducted across two inpatient rehabilitation units within one UK hospital had a high risk of bias.
Participants. This study involved forty participants, 24 (60%) of whom were male. The mean age of the intervention group was 67.5 years (SD 10.4) versus the control group of 66.7 years (SD 12.0). Intervention participants had a mean time between stroke and intervention of 16.1 days (SD 8.8) versus control participants 15.6 days (SD 8.1).
Intervention. Intervention participants were encouraged to self-propel a wheelchair following instruction by a physiotherapist on how to self-propel. Intervention participants were encouraged by ward staff members to self-propel as much as they could, with weekly reminders from the study team. This encouragement continued until discharge or for a maximum of 8 weeks. Control participants were also provided a wheelchair but were actively discouraged from self-propulsion.
Risk of bias. Lack of blinding and insufficient reporting resulted in this study being considered to be of unclear bias risk.
Can encouragement to self-propel a wheelchair influence ADL outcome amongst inpatient stroke survivors against discouragement to self-propel a wheelchair?. Only one paper, Barrett et. al. (2001), addresses this question [96]. Both groups improved their BI scores at 3 months (11.4 (4.0 SD) v 9.8 (5.0 SD)) and 12 months (11.9 (5.2 SD) v 11.9 (4.1 SD) but no significant difference between the groups was identified (no p value reported). Similarly, both groups improved their NeADL scores at 3 months (5.8 (5.2 SD) v 5.3 (4.0 SD)) and at 12 months (7.1 (4.7 SD) v 8.0 (5.3 SD) but again the difference between groups was not significant. Therefore, there is no evidence at present to recommend the encouragement of wheelchair propulsion as part of stroke rehabilitation care amongst older stroke survivors. With the bias risks imposed (no participant blinding, concerns about selection bias and small sample size) the evidence has been rated as low using the GRADE system meaning that further studies are very likely to have an important impact on the estimate of benefit. Table 21 presents a summary of the recommendations this study proposes based on the identified evidence.

Acupuncture
Acupuncture is rarely mentioned as a therapy for stroke survivors within the guidelines. The Royal College of Physicians [4] refer to the limited evidence for acupuncture in the treatment of post-stroke dysphagia. The Scottish Intercollegiate Guidelines Network [5] state that they do not recommend acupuncture for the treatment of post-stroke pain syndromes due to insufficient evidence. Teasel et. al. (2003) report that the evidence linking acupuncture to post- Table 21. Summary of recommendations for non-pharmacological interventions for older stroke survivors rehabilitation.

Acupuncture
There is very limited evidence to show that acupuncture can benefit older stroke survivors ADL and no evidence to support its benefit upon Disability. A low GRADE quality assessment score, combined with this limited evidence, means that we cannot recommend acupuncture for older stroke survivors.

Caregiver Training
There was very limited evidence to show that caregiver training can benefit older stroke survivors ADL, and no evidence to support any benefit upon disability scores. Only one study was considered in this category and was given a GRADE quality assessment score of low. Therefore, we are unable to recommend caregiver training to benefit older stroke survivors.

CIMT
There was no evidence to show that CIMT improves ADL performance and the study was assessed by GRADE to be of low quality. Therefore, we cannot recommend the use of CIMT for older stroke survivors.

Device assisted PT
There is very limited evidence to support the use of device assisted PT to enhance older stroke survivors ADL. The quality of the evidence was assessed by GRADE and found to be low. Therefore, we cannot recommend device assisted PT for older stroke survivors

Music Therapy
Evidence from one study does not allow us to support the use of music therapy to enhance older stroke survivors ADL. The quality of the evidence was assessed by GRADE and found to be low and therefore we cannot recommend the use of music therapy to benefit older stroke survivors.

Nerve Stimulation
There was very limited evidence to show that nerve stimulation can benefit older stroke survivors ADL. The quality assessment score was very low. Therefore, we cannot recommend nerve stimulation to benefit older stroke survivors ADL.

OT
There is evidence to show that additional OT can benefit older stroke survivors ADL but no evidence to show that alternative OT approaches can benefit older stroke survivors ADL. GRADE quality assessment suggests the evidence for to be of a low score. Therefore, the use of additional OT can be recommended as it may benefit older stroke survivors ADL. However, alternative OT approaches cannot be recommended for older stroke survivors ADL. In relation to disability, there was no evidence to suggest that additional occupational therapy can improve post-stroke disability. The evidence for disability was given a GRADE quality assessment score of moderate. Therefore, we are unable to recommend additional occupational therapy to benefit older stroke survivors' disability.

Optical
There was no evidence to show that optical interventions can benefit older stroke survivors ADL and quality assessment of included studies was awarded a GRADE score of very low. Therefore, we are unable to recommend optical therapies to benefit older stroke survivors.
Psychological Therapies There is very limited evidence to show that psychological therapies can benefit older stroke survivors ADL and no evidence to support its benefit upon disability. GRADE quality assessment was scored as low. Therefore, we cannot recommend psychological therapies to benefit older stroke survivors.

PT
There is some evidence to show that additional PT can benefit older stroke survivors ADL but no evidence to show that alternative PT approaches can benefit older stroke survivors ADL. GRADE quality assessment suggests the evidence for ADL to be of a low score. Therefore, the use of additional PT can be recommended as it may benefit older stroke survivors ADL. However alternative approaches to PT cannot be recommended for older stroke survivors.

Self-management Education
There is very limited evidence to show that self-management education programmes can benefit older stroke survivors ADL and no evidence to support any benefit upon disability. GRADE quality assessment suggests the evidence for ADL to be of low quality, and for disability, very-low, quality. Therefore we cannot recommend self-management education programmes to benefit older stroke survivors.

Videogames
There was no evidence to show that videogames can benefit older stroke survivors ADL and quality assessment of the one study in this category was given a GRADE score of very low. Therefore, we are unable to recommend videogames to benefit older stroke survivors.

(Continued)
Systematic review of non-drug therapy for older people with stroke stroke ADL recovery is conflicting [97]. Our current review corresponds with previous uncertainties. There was limited evidence to show that acupuncture can benefit older stroke survivors and further research is required.

Caregiver training
Stroke guidelines acknowledge the insufficient evidence behind the benefits of caregiver training as part of stroke rehabilitation, but do promote carer involvement in patient rehabilitation as good practice [4][5]. Our current review identified only one study exploring the impact of caregiver training upon older stroke survivors ADL. This study was sufficiently large, and demonstrated benefits in ADL recovery, but these were short term [31]. Therefore, caregiver training may be beneficial, but further research is required to examine this intervention further.

CIMT
Our current review found little evidence to support the use of CIMT with older stroke survivors, similar to the findings of Veerbeek et. al. (2014) [98]. Several stroke rehabilitation guidelines recommend CIMT to improve function of impaired upper limbs [3][4][5]99]. In view of the evidence, SIGN (2010) specifically state that "Constraint induced movement therapy may be considered for carefully selected individuals with at least 10 degrees of finger extension, intact balance and cognition" (p20) [5]. RCP (2016) also explain that the benefits of CIMT often relate to arm function only and within the confines of the activities used within the intervention [4]. Similar to other stroke rehabilitation interventions, CIMT appears most effective when effectiveness is measured in terms of its immediate effect on physiological variables, such as muscle strength. But such benefits do not appear to be associated with improvements in more comprehensive or global outcomes of ADL or disability.

Device assisted physiotherapy
Our current review identified limited evidence to support the use of device assisted physiotherapy to enhance older stroke survivors stroke rehabilitation. Our findings are in line with other reviews and guidelines [4,98]. The use of robotic devices has been recommended by Teasel et. al. (2003) as they considered this approach beneficial for those with impaired arm function [97], but this recommendation was based on achieving improved arm function, not improved ADL. Conversely, as a result of the overall low quality of evidence behind robot assisted movement therapies the RCP (2016) guidelines stipulate that this type of therapy should only be offered as an adjunct to conventional therapy and within the context of a clinical trial [4]. Systematic review of non-drug therapy for older people with stroke

Music therapy
Only one study explored music therapy in relation to older stroke survivors ADL [42] and no evidence was found to support use of this intervention. Music therapy has been explored previously within neuro-rehabilitation and reviews have identified several benefits such as improved motor function, language and mood [100][101][102]. Nevertheless, their efficacy with older stroke survivors, and/or impact on global ADL or disability, remains unknown.

Nerve stimulation
Current stroke guidelines have all noted uncertainties surrounding the efficacy of nerve stimulation [4-5, 97, 99]. Evidence has shown that while nerve stimulation techniques can improve specific impairments, such as muscle strength or gait, these improvements do not lead to significant improvements in ADL or disability [4]. Within this review, limited evidence was identified supporting the efficacy of nerve stimulation upon older stroke survivors ADL. Therefore nerve stimulation may benefit older stroke survivors ADL, but the quality of evidence is weak.
The number of included studies focusing exclusively upon older stroke survivors is small, making it difficult to sub-divide studies into those focusing on specific types of stimulation or use of stimulation in different locations (e.g. upper or lower body). Reviews which have included adult participants of all ages suggest the best evidence behind nerve stimulation may be found in its use for upper limb impairments [98,103]. However, due to inconsistencies, current evidence remains insufficient to make any recommendations [5].

Occupational therapy
There was some evidence to show that additional OT can benefit older stroke survivors ADL. This is consistent with the reporting of uncertainties regarding the effects of increased intensity or frequency of OT [4][5]. This review found no evidence to suggest that one OT approach is more beneficial than anther, consistent with the review by Teasel et. al. (2003) [97]. All guidelines recommend ADL focused OT as an important feature of stroke rehabilitation, but acknowledge that optimal intensity and/or duration is yet to be determined [4][5].

Optical
Within the Royal College of Physicians (2016) stroke guidelines, optical interventions such as prism glasses are recommended for stroke survivors with visual neglect [4]. This said, it is noted that the evidence is very limited and that patient participation in such interventions would be most beneficial within the context of a clinical trial [4], a finding echoed by Langhorne, Bernhardt & Kwakkel (2011) [99]. A Cochrane review of interventions targeted at spatial neglect following stroke concluded that there was insufficient evidence to support the use of these interventions to improve disability and ADL [104]. However, the evidence was more promising for specific visual neglect measures [104]. This current review identified no evidence to support the use of these interventions. However, a combination of few studies and small sample sizes may obscure any potential positive impact from these interventions. As suggested by Bowen, Lincoln & Dewey (2007) [104], we also recommend further research involving larger high quality RCTs.

Physiotherapy
It has been reported that many PT interventions such as balance exercises, gait training, and fitness training do lead to benefits in their respective objectives i.e. improved balance, gait, cardiovascular fitness, but rarely lead to improvement in more global measures such as ADL and disability [97]. Veerbeek et. al. (2014) systematically reviewed 467 RCTS involving PT interventions to examine their efficacy in stroke rehabilitation [98]. The strongest evidence supported task specific functions and activities which are repeated at high intensity [98]. However, few global outcomes were reported; outcomes included muscle, joint, bone and sensory function, gait pattern, balance and walking [98]. Benefits in terms of basic ADL were reported to arise from interventions involving activity-based balance training, assisted gait training and VR training for paretic arm [98]. This current review identified limited evidence to demonstrate that additional PT can benefit older stroke survivors' ADL in comparison to usual care. When restricted to older stroke survivors, and those reporting global measure of ADL or disability, the resulting number of included studies is considerably smaller than those cited in reviews such as that by Veerbeek et. al. (2014) [98]. Nevertheless, we found some evidence suggesting older stroke survivors may benefit from increased PT.

Psychological therapies
Within the guidelines for stroke rehabilitation it is recommended that all stroke survivors be considered and offered psychological care, and not just offer to those with an identified mental health disorder [4]. This recommendation is based upon good practice, possibly due to the frequent development of post-stroke depression [4][5]. In this review limited evidence was identified to show that psychological therapies benefit older stroke survivors. Therefore, psychological therapies may benefit older stroke survivors ADL but the evidence for this is weak and requires further investigation.

Self-management education
Self-management is reported to be capable of influencing function and social participation, and recommended for stroke survivors [4]. This review identified limited evidence to show that such interventions can benefit older stroke survivors. The quality of included studies varied from low to very low and therefore the evidence for this is weak. Further research focusing exclusively upon older stroke survivors and global outcomes is recommended.

Videogames
Current guidelines report that the evidence behind virtual reality as a stroke rehabilitation approach is weak to moderate [4][5]. A Cochrane review by Laver et. al. (2015) suggests that virtual reality can benefit upper limb impairments and ADL, but that evidence is limited to younger stroke survivors and those who are more than one-year post-stroke [105]. In addition to larger high quality RCTs, it has been recommended that research focus upon identifying what the important elements of virtual reality are, and if benefits can be sustained in the long term [105]. Another review suggests that the best evidence lies behind the efficacy of virtual reality upon gait improvement [97]. Based on the evidence generated in this current review it is not possible to recommend virtual reality for older stroke survivors' rehabilitation. However, based on evidence identified by studies involving slightly younger stroke survivors, further research involving appropriately sized high quality RCTs is warranted.

Wheelchair use
Only one study was identified which investigated if self-propulsion of a wheelchair improved ADL recovery against non-self-propulsion, and there was no evidence of the efficacy of this approach. Current recommendations do not include such an intervention but do recommend the use of wheelchairs in those with impaired mobility to promote independence [4][5].

Limitations
This review has several limitations which must be considered alongside our findings. Firstly, we did not involve patients or carers in the Delphi process, and our identified critical outcomes may not reflect patient and carer preferences. Due to study heterogeneity and insufficient data this review has been limited to narrative analysis only. While describing comparisons between studies is important, it has potential for researcher bias through the imposition of the researchers own subjective ideas about the findings and lacks the rigour of qualitative and objective analysis. Although we used the GRADE criteria recommended by the Cochrane Collaboration this also introduced a degree of subjectivity. This means our results should be interpreted cautiously. We also cannot exclude the possibility that this review has omitted important studies. We have not searched the grey literature and our search strategy focused exclusively on identifying systematic reviews which may have resulted in omission of some trials, particularly those more recently published. However, our comprehensive strategy and the checking of reference lists and published clinical guidelines does go some way in reducing this risk. Categories of non-pharmacological interventions were developed in a somewhat arbitrary fashion. It could be argued that interventions exploring nerve stimulation devices, often delivered by trained physiotherapists, could be considered an alternative physiotherapy approach, as opposed to a category in its own right. Our decisions regarding categorising interventions were largely pragmatic and aimed to organise and present findings in a meaningful way. However, the findings should be interpreted with caution since the interventions lack specificity. We also do not consider the preferences of patients and their carers regarding intervention types. Little work has been done in this area and acceptability of these non-pharmacological approaches are unknown.
Finally, our review is restricted by the significant lack of published studies which met our age criteria (mean age! 65 years) and presented results using a global measure of ADL and/or disability. Age-based criteria allow us to examine the evidence as it specifically relates to older adults, but it risks excluding interventions which may be beneficial but have not been adequately tested in an older population. The impact of age as a modifier of treatment effect for many of the interventions examined is unknown. Similarly, the exclusion of so many studies due to lack of global outcome measures again risks excluding worthy interventions which may have demonstrated efficacy had a global outcome been assessed. Additionally, this review uncovered a number of methodological and reporting problems, making the ascertainment of the evidence challenging. Small sample sizes and failure to adequately report details regarding participant selection, randomisation, allocation concealment and data analysis, especially the management of missing data, led to many studies being deemed high risk of bias. One important challenge regarding RCTs involving non-pharmacological treatments is the lack of participant blinding. Although blinding of non-pharmacological treatments is challenging, reviews do highlight many creative approaches to doing so [106]. However, opinions regarding the importance of this are divided. Lack of patient blinding in RCTs presents opportunity for bias, particularly for subjective outcomes [106] such as those explored in the present manuscript. However, concerns have been raised about false negative results arising from RCTs involving non-pharmacological treatments as a result of blinded participants [107]. It is argued that what factors blinding controls for may be an integral component of non-pharmacological therapy [107]. For example, the additional care an intervention participant may receive as part of their acupuncture treatment may contribute towards overall benefit of the treatment [107]. In pharmacological RCTs this additional care would be considered incidental and would be controlled for through provision of similar care to control participants [106][107]. However, it has been argued that this takes away from some of the benefits non-pharmacological treatments bring, and therefore leads to findings of no-benefit [107]. It may be prudent for future work to explore the role of incidental and placebo effects in non-pharmacological treatments for stroke survivors to enhance our confidence in future results.

Conclusion
Due to the substantial heterogeneity, moderate to high risk of biases, and insufficient data provided, this review has had to make recommendations based on narrative analysis only.
Limited evidence suggests additional physiotherapy or occupational therapy may benefit older stroke survivors ADL. Very limited evidence also suggests acupuncture, self-management education, psychological therapies, nerve stimulation, CIMT, and caregiver training may benefit older stroke survivors ADL.
However, the current evidence base is limited by the low number and quality of studies. This review revealed a distinct lack of evidence behind the use of non-pharmacological interventions for stroke survivors aged 65 years and older. Of studies which did involve those aged 65 and older, evidence is limited by poor study designs and inadequate study reporting. Therefore, we also recommend that future studies explore these interventions exclusively in older adult populations, and ensure studies are adequately reported both in terms of methodological detail but also in terms of their outcomes.