Effectiveness of somatosensory interventions on somatosensory,motor and functional outcomes in the upper limb post-stroke: A systematic review and meta-analysis

Abstract

BACKGROUND:

Research mainly focuses on motor recovery of the upper limb after stroke. Less attention has been paid to somatosensory recovery.

OBJECTIVE:

To review and summarize the effect of upper limb somatosensory interventions on somatosensory impairment, motor impairment, functional activity and participation after stroke.

METHODS:

Biomedical databases Ovid Medline, EMBASE, Web of Science, PEDro, and OTseeker were searched with an update in May 2018. Randomized controlled trials investigating the effect of somatosensory-specific interventions focusing on exteroceptive, proprioceptive or higher cortical somatosensory dysfunction, or any combination were eligible for inclusion. Quality of included studies were assessed using Physiotherapy Evidence Database (PEDro) scale. Standardized Mean Differences and Mean Differences and 95% confidence intervals were calculated and combined in meta-analyses.

RESULTS:

Active somatosensory interventions did not show a significant effect on somatosensation and activity, but demonstrated a significant improvement in motor impairment (SMD = 0.73, 95% CI = 0.14 to 1.32). No study evaluating active somatosensory intervention included participation. Passive somatosensory interventions significantly improved light touch sensation (SMD = 1.13, 95% CI = 0.20 to 2.05). Passive somatosensory interventions did not show significant effects on proprioception and higher cortical somatosensation, motor impairment, activity and participation.

CONCLUSIONS:

To date, there is low quality evidence suggesting active somatosensory interventions having a beneficial effect on upper limb impairment and very low quality evidence suggesting passive somatosensory interventions improving upper limb light touch sensation. There is a need for further well-designed trials of somatosensory rehabilitation post stroke.

Keywords

Stroke Upper Limb Somatosensory Rehabilitation Somatosensation Systematic Review Meta-analysis

1 Introduction

Stroke is one of the leading causes of acquired adult disability (Benjamin et al., 2017). Not only are functionality of the lower limb impaired following stroke, but also functionality of the upper limb. Nearly 70% of patients with stroke suffer motor and/or somatosensory impairments in the upper limb (Meyer, 2015). Upper limb impairments result in non-use and restrict patients’ functionality in daily life activities and participation (Hunter & Crome, 2002).

Somatosensation denotes being aware of body parts by identifying movements, detecting touch and discriminating between stimuli. Proprioception (e.g. position sense and kinesthesia), exteroception (e.g. light touch, pinprick and pain) and higher cortical somatosensation (e.g. two-point discrimination, stereognosis) are the different modalities of somatosensation (Gardner & Martin, 2000).

Somatosensory information is essential for motor function. The integration of different sensations from skin, ligaments, muscles and joints promotes controlled and accurate upper limb movement (Ridding, McKay, Thompson, & Miles, 2001; Stefan, Kunesch, Cohen, Benecke, & Classen, 2000). Thus, regaining functionality of the upper limb after stroke is dependent on sensorimotor recovery (Hunter & Crome, 2002). The systematic review by Meyer et al. (2014) investigated the effects of somatosensory deficits on outcome after stroke and concluded that somatosensory deficits have a negative impact on upper limb function and recovery.

Somatosensory dysfunction is common after stroke (Connell, Lincoln, & Radford, 2008; Findlater & Dukelow, 2017; Tyson, Hanley, Chillala, Selley, & Tallis, 2008) with deficits seen in 41–63% of patients in the acute phase, 21–54% in the subacute phase and 3–50% in the chronic phase having an impairment in at least one of the somatosensory modalities (Meyer et al., 2016a; Meyer et al., 2016b). Somatosensory interventions may have a pivotal role in effective rehabilitation strategies for the upper limb and promoting motor recovery in survivors after stroke (Chen & Shaw, 2014; Kessner, Bingel, & Thomalla, 2016).

Recently, there has been growing interest in the possible impact of somatosensory interventions on upper limb functionality. Over the past few years, four systematic reviews (Doyle, Bennett, Fasoli, & McKenna, 2010; Grant, Gibson, & Shields, 2017; Laufer & Elboim-Gabyzon, 2011; Schabrun & Hillier, 2009) investigated the effectiveness of somatosensory interventions for the upper limb after stroke. Schabrun and Hiller (2009) examined the effects of passive (electrical stimulation applications) and active somatosensory training (specific exercises with clear intention to train somatosensory functions) on impairment and function for both upper and lower limb. They concluded that passive somatosensory training was beneficial for the upper limb. However, active somatosensory training seemed not effective (Schabrun & Hillier, 2009). A Cochrane review (Doyle et al., 2010) from 2010 analyzed all interventions concentrating on somatosensory deficits in the upper limb, but due to scarce evidence (13 trials were included with a total 467 participants), they could not suggest conclusive evidence about the effect of any intervention. Another review in this field from 2011 (Laufer & Elboim-Gabyzon, 2011) investigated the effect of somatosensory transcutaneous electrical nerve stimulation (TENS) on motor function of both upper and lower limb and found that TENS might be beneficial for motor recovery post stroke. And recently, a review explored the effectiveness of somatosensory stimulation of arm and hand on motor function and reported that somatosensory stimulation was not more effective than usual care on motor recovery of the hand (Grant et al., 2017). An updated comprehensive review of this research domain seems warranted.

Overall, considerable research has been devoted to upper limb motor recovery following stroke, and less attention has been paid to somatosensory recovery of the upper limb. We conducted a systematic review with the aim to review and summarize the effect of upper limb somatosensory interventions on upper limb somatosensory impairment as well as motor impairment, functional activity and participation.

2 Method

2.1 Databases and search strategy

We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Liberati et al., 2009) (PRISMA) statement when conducting this review (Appendix 1). Biomedical databases Ovid Medline, EMBASE, Web of Science, PEDro, and OTseeker were searched. Published studies were identified using a search strategy developed in collaboration with an experienced librarian. Key words and MESH terms were upper extremity; stroke; cerebrovascular disorders; hemiplegia; sensation; somatosensation; sensory disorder; and sensory retraining with their synonyms and plurals. Appendix 2 shows the search strategy used for Ovid Medline. For the other databases, the search strategy was adapted accordingly. Reference lists were hand-searched for additional relevant publications. Studies published in English were eligible for inclusion when investigating the effect of an upper limb intervention with a clear emphasis on somatosensation.

2.2 Included study designs

Only randomized controlled trials (RCTs) were considered for inclusion. The first phase of randomized crossover trials was also considered. Single-session studies were not included.

2.3 Participants

Studies that recruited adults (≥18 years) with a confirmed diagnosis of stroke were included. No limitation was applied according to the type (hemorrhagic or ischemic), stage (acute, subacute or chronic), or anatomical location of stroke. Participants with other neurological diseases (e.g. Parkinson’s disease, traumatic brain injury, neuropathy) were excluded. Studies including mixed group of subjects were also excluded, unless they presented results for the stroke subgroup separately. For the stages of stroke, we used the framework determined by The Stroke Recovery and Rehabilitation Roundtable (SRRR). According to this framework, timelines of recovery are distinguished as follows: 0-24 hours: hyperacute, 1 day-7 days: acute, 7 days-3 months: early subacute, 3 months-6 months: late subacute, >6 months: chronic (Bernhardt et al., 2017).

2.4 Outcome measures

The primary outcome of interest was somatosensation. There are various instruments available for measuring somatosensory impairment, but we focused on measurement tools with established psychometric properties, based on adequate referencing. Based on the International Classification of Functioning, Disability and Health (ICF) model (WHO, 2001), our secondary outcomes of interest were impairment, activity and participation. For the impairment domain, all measurements that assessed an upper limb motor performance were included. For the activity domain, tools quantifying independence in (basic) activities of daily living and for the participation domain, measures examining quality of life were included. Trials that included suitable secondary outcomes were only included in our review when they also comprised a somatosensory outcome.

2.5 Interventions

We included studies that compared somatosensory-focused interventions with no therapy, usual therapy, sham, different interventions or a different version of the same intervention. Somatosensory interventions were further divided into active and passive somatosensory interventions. Active somatosensory interventions are defined as somatosensory interventions that require participation of patients to a graded and mostly therapist-delivered somatosensory re-training. Passive somatosensory interventions require patients only to feel afferent stimuli without any active motor or cognitive reaction. Active somatosensory interventions require patients to interpret the afferent stimulation and respond accordingly, whereas passive somatosensory interventions purely stimulate somatosensory receptors. Studies were only included in the review if they focused on upper limb somatosensory interventions. Analyses were performed separately for the results of trials that compared active or passive somatosensory interventions with sham or other therapy as well as active or passive somatosensory interventions with no additional therapy. When analyzing the effect of somatosensory interventions, somatosensation was divided into three categories: exteroception, proprioception, higher cortical somatosensation.

2.6 Search process and quality evaluation

Articles were selected between October 2009 (end of search of the Cochrane review) and August 2017, with an update of the search conducted in May 2018. However, all previous papers from the 2010 Cochrane review (Doyle et al., 2010) and the most recent review (Grant et al., 2017) in this field were considered for inclusion as well. A total of 4826 records were identified through database searching. After removal of duplicates, two independent reviewers (C.Y. and L.B.) screened the titles and abstracts. Subsequently, the process of assessment of full-texts was carried out. When disagreement occurred, consensus was searched. If this was not reached, a third reviewer (G.V.) was consulted who decided on inclusion or exclusion.

Quality of the included studies was assessed with the PEDro Scale (de Morton, 2009). Two reviewers (C.Y. and L.B.) independently scored the studies and the scores were checked with the scores in the PEDro database (www.pedro.org.au). In case any RCT was not included in the PEDro database, the reviewers (C.Y. and L.B.) applied their final score in consultation with G.V. RCTs with scores lower than 4 were considered as having high-risk of bias.

2.7 Data extraction and analysis

A data extraction form was generated to gather information regarding patient characteristics (age, stroke severity, stroke type, length of time post stroke), methods (sample size, study design, inclusion and exclusion criteria, intervention characteristics, measurement tools) and results (number of participants, study outcomes, means and standard deviations and adverse events) for included studies. Two reviewers extracted the data independently and results were compared.

Review Manager Software (RevMan5.3) was used to perform meta-analyses and the guidelines in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins & Green, 2011) were followed. Where sufficient information was obtained, standardized mean differences (SMD) were calculated for the outcomes based on the change scores of means (post intervention-pre-intervention) and standard deviations from the included studies. When the same outcome measurements were used, the size of the effect was reported as mean difference (with 95% confidence interval (CI). When different outcome measurements were used, the effect size was reported as standardized mean difference (with 95% CI). In both cases, a random effects model was used due to the heterogeneity observed for several of our comparisons. To deal with missing data, authors were contacted. Where missing data were not provided by the authors, an estimation of the standard deviation scores was calculated following the guidelines of Cochrane Handbook, Chapter 16.1.3. (Higgins & Green, 2011). Since the handbook states that imputation of estimated standard deviations yields approximately accurate results, the study with the most participants in the meta-analysis and the highest standard deviation of an outcome measurement in the same study was chosen to calculate the missing standard deviations. Data heterogeneity is reported using I² test. Following Cohen, the strength of the standardized mean difference was considered as follows: >0.2 small, >0.5 moderate and >0.8 large (Cohen, 1988).

3 Results

3.1 Search process

Database searching identified 4826 studies. According to our inclusion and exclusion criteria, 9 studies were included in this review and 8 studies in the meta-analysis. One of the studies (Byl et al., 2003) was not included in the meta-analysis because of missing data. Figure 1 shows the flowchart of the search process.

Fig.1

Flowchart of selecting identified studies.

3.2 Characteristics of included studies

3.2.1 Quality

The mean PEDro score of the studies was 6.67 (SD 1.94) and quality ranged from 4 (Byl et al., 2003) to 10 (Stein et al., 2010) (Table 1). The most common deficiencies in methods were blinding of therapist (n = 8) and participant (n = 6) and inadequate intention-to-treat analysis (n = 7).

Table 1
PEDro scores of included studies

Study Random allocation Concealed allocation Groups similar at baseline Participant blinding Therapist blinding Assessor blinding <15% dropouts Intention-to-treat analysis Between-group difference reported Point estimate and variability reported Total (0 to 10)

Acerra N 2007 Y N Y N N Y Y N Y Y 6

Byl et al. 2003 Y N Y N N Y Y N N N 4

Cambier et al. 2003 Y N Y N N Y N N Y Y 5

Carey et al. 2011 Y Y Y Y N Y Y N Y Y 8

Chen et al. 2005 Y Y Y N N Y N N Y Y 6

Poole et al. 1990 Y N Y N N N Y N Y Y 5

Salles et al. 2017 Y N Y N N Y Y N Y Y 6

Stein et al. 2010 Y Y Y Y Y Y Y Y Y Y 10

Sullivan et al. 2012 Y N Y Y N Y Y Y Y Y 8

3.2.2 Participants

A total of 257 participants with mean age 62.33 years were included in the review and 128 patients received the somatosensory intervention, with 129 patients in the control group. One study (Acerra, 2007) included acute patients with stroke, two studies (Chen, Liang, & Shaw, 2005; Sallés et al., 2017) included patients in the early subacute phase, one study (Cambier, De Corte, Danneels, & Witvrouw, 2003) included patients in the late subacute phase, four studies (Byl et al., 2003; Carey, Macdonell, & Matyas, 2011; Stein et al., 2010; Sullivan, Hurley, & Hedman, 2012) included patients with chronic stroke and one study (Poole, Whitney, Hangeland, & Baker, 1990) did not provide data about time post stroke. Although the majority of the included studies (Byl et al., 2003; Cambier et al., 2003; Poole et al., 1990; Stein et al., 2010; Sullivan et al., 2012) did not provide information about the types of stroke, two studies (Acerra, 2007; Sallés et al., 2017) included patients with ischemic stroke and two studies (Carey et al., 2011; Chen et al., 2005) included participants with both ischemic and hemorrhagic stroke (Table 2).

Table 2
Summary of included studies

Study Participants Intervention Outcome measures and ICF level

Acerra N 2007 n = 40 Active somatosensory training Body function:

•Frey monofilaments^*

Mean age = 68 years (SD n/a) Exp: motor and sensory discrimination tasks inside mirror box •Temperature hot/cold

•PPT

•Grip strength^*

Stroke phase: acute Con: sham box without mirror 30 min×14 consecutive days Activity:

Type of stroke: ischemia •Motor Assessment Scale*

Adverse events: no adverse events

Byl et al. 2003 n = 21 Active somatosensory training Body structure/function:

**Cross-over study* •Sensory Integration Praxis Test,

Mean age = 63 years (SD 9.4) •Byl-Cheney-Boczai Test

Exp: sensory training for 4 weeks + motor training for 4 weeks •Digital reaction time

•Purdue pegbord

Stroke phase: chronic •Manual muscle test

Type of stroke: not stated Con: motor training for 4 weeks + sensory training for 4 weeks •ROM

•California Functional Evaluation-40

90 min &times 5/wk×8 wk Activity:

•Wolf Motor Function Test

Adverse events: not stated

Cambier et al. 2003 n = 23 Passive somatosensory training Body structure/function:

•Nottingham Sensory

Mean age = 62.5 years (SD n/a) Exp: conventional therapy combined with intermittent pneumatic compression treatment •Assessment*

•Fugl-Meyer Assessment

Stroke phase: late subacute •Ashworth Scale

•Visual Analog Scale^*

Type of stroke: not stated Con: conventional therapy combined with 30 min×5/wk×4 wk

Adverse events: not stated

Carey et al. 2011 n = 50 Active somatosensory training Body structure/function:

•Fabric Matching Test

Mean age = 61.02 years (SD 12.75) Exp: sensory discrimination training •WEST-hand monofilaments

•Tactile Discrimination Test

Stroke phase: chronic Con: sham (repeated exposure to different stimuli) •Wrist Position Sense Test

•Finger Position Sense Test

•functional Tactile Object Recognition Test

Type of stroke: 64% ischemia, 34% hemorrhage, 2% ischemia and hemorrhage 60 min×3/wk×10 sessions ^* For the meta-analyses, the composite score was used.

Chen et al. 2005 n = 29 Passive somatosensory training Body structure/function:

Mean age = 59.05 years (SD n/a) Exp: thermal stimulation •Brunnstrom stage

•Wrist extension

•Wrist flexion^*

Stroke phase: early subacute Con: discussion with the physiotherapist about the progress in their treatment. •modified Motor Assessment Scale

Type of stroke: 17 ischemia, 12 hemorrhage 20–30 min×5/wk×6 wk •Modified Ashworth Scale

•Semmes Weinstein monofilaments*

Adverse events: no adverse events

Poole et al. 1990 n = 18 Passive somatosensory training Body function

•Fugl-Meyer Assessment^*

Mean age = 70 years (SD n/a) Exp: conventional occupational therapy + inflatable pressure splint

Adverse events: not stated

Stroke phase: not stated

Con: conventional occupational therapy

Type of stroke: not stated

30 min×5/wk×3 wk

Salles et al. 2017 n = 8 Active somatosensory training Body structure and function:

Mean age = 53.4 years (SD 9.6) Exp: sensory tasks with cognitive activation. •Motor Evaluation Scale for Upper Extremity in Stroke Patients^*

•Motricity Index

Con: sensory tasks without cognitive activation. •Revised Nottingham Sensory Assessment^*

Stroke phase: early subacute 30 min×3/wk×10 wk •Kinesthetic and Visual Imagery Questionnaire

Type of stroke: ischemia Adverse events: not stated

Stein etal. 2010 n = 30 Passive somatosensory training Body structure/function:

Mean age = 63.4 years (SD n/a) Exp: mechanical stimulation (stochastic resonance) was applied to the extensors of fingers, wrist and elbow, at the subsensory threshold level during task-specific activities. •Fugl-Meyer Assessment

•Reaching performance Scale

Stoke phase: chronic •Proprioception^*

Type of stroke: not stated •Vibration

Con: the stochastic resonance device was deactivated after sensory threshold testing. •Pinprick

•Semmes Weinstein monofilaments*

60 min×3/wk×4 wk Activity:

•Wolf Motor Function Test

•Action Research Arm Test

•Motor Activity Log^*

Participation:

•Stroke Impact Scale^*

Adverse events: unserious adverse events

Sullivan et al. 2012 n = 38 Passive somatosensory training Body structure/function:

Mean age = 60.6 years (SD n/a) Exp: home-based task-specific activities using glove electrode at the sensory threshold level. •Fugl-Meyer Assessment

•Arm Motor Ability Test

Stroke phase: chronic Con: Subjects in the sham group wore an electrode glove and electrode attached to the sham stimulator. The sham stimulators’ timer and lights were active; the amplitude was adjustable; however no current was delivered. •Perceptual Threshold Test^*

Type of stroke: not stated

30 min×2/w×4 wk Activity:

•Motor Activity Log^*

Participation:

•Stroke Impact Scale^*

Adverse events: not stated

Study	Participants	Intervention	Outcome measures and ICF level
Acerra N 2007	n = 40	Active somatosensory training	Body function:
			•Frey monofilaments^*
	Mean age = 68 years (SD n/a)	Exp: motor and sensory discrimination tasks inside mirror box	•Temperature hot/cold
			•PPT
			•Grip strength^*
	Stroke phase: acute	Con: sham box without mirror 30 min×14 consecutive days	Activity:
	Type of stroke: ischemia	•Motor Assessment Scale*
			Adverse events: no adverse events
Byl et al. 2003	n = 21	Active somatosensory training	Body structure/function:
		*Cross-over study	•Sensory Integration Praxis Test,
	Mean age = 63 years (SD 9.4)	•Byl-Cheney-Boczai Test
		Exp: sensory training for 4 weeks + motor training for 4 weeks	•Digital reaction time
			•Purdue pegbord
	Stroke phase: chronic	•Manual muscle test
	Type of stroke: not stated	Con: motor training for 4 weeks + sensory training for 4 weeks	•ROM
			•California Functional Evaluation-40
		90 min &times 5/wk×8 wk	Activity:
			•Wolf Motor Function Test
			Adverse events: not stated
Cambier et al. 2003	n = 23	Passive somatosensory training	Body structure/function:
			•Nottingham Sensory
	Mean age = 62.5 years (SD n/a)	Exp: conventional therapy combined with intermittent pneumatic compression treatment	•Assessment*
			•Fugl-Meyer Assessment
	Stroke phase: late subacute		•Ashworth Scale
			•Visual Analog Scale^*
	Type of stroke: not stated	Con: conventional therapy combined with 30 min×5/wk×4 wk
			Adverse events: not stated
Carey et al. 2011	n = 50	Active somatosensory training	Body structure/function:
			•Fabric Matching Test
	Mean age = 61.02 years (SD 12.75)	Exp: sensory discrimination training	•WEST-hand monofilaments
			•Tactile Discrimination Test
	Stroke phase: chronic	Con: sham (repeated exposure to different stimuli)	•Wrist Position Sense Test
			•Finger Position Sense Test
			•functional Tactile Object Recognition Test
	Type of stroke: 64% ischemia, 34% hemorrhage, 2% ischemia and hemorrhage	60 min×3/wk×10 sessions	^* For the meta-analyses, the composite score was used.
Chen et al. 2005	n = 29	Passive somatosensory training	Body structure/function:
	Mean age = 59.05 years (SD n/a)	Exp: thermal stimulation	•Brunnstrom stage
			•Wrist extension
			•Wrist flexion^*
	Stroke phase: early subacute	Con: discussion with the physiotherapist about the progress in their treatment.	•modified Motor Assessment Scale
	Type of stroke: 17 ischemia, 12 hemorrhage	20–30 min×5/wk×6 wk	•Modified Ashworth Scale
			•Semmes Weinstein monofilaments*
			Adverse events: no adverse events
Poole et al. 1990	n = 18	Passive somatosensory training	Body function
			•Fugl-Meyer Assessment^*
	Mean age = 70 years (SD n/a)	Exp: conventional occupational therapy + inflatable pressure splint
			Adverse events: not stated
	Stroke phase: not stated
		Con: conventional occupational therapy
	Type of stroke: not stated
		30 min×5/wk×3 wk
Salles et al. 2017	n = 8	Active somatosensory training	Body structure and function:
	Mean age = 53.4 years (SD 9.6)	Exp: sensory tasks with cognitive activation.	•Motor Evaluation Scale for Upper Extremity in Stroke Patients^*
			•Motricity Index
		Con: sensory tasks without cognitive activation.	•Revised Nottingham Sensory Assessment^*
	Stroke phase: early subacute	30 min×3/wk×10 wk	•Kinesthetic and Visual Imagery Questionnaire
	Type of stroke: ischemia	Adverse events: not stated
Stein etal. 2010	n = 30	Passive somatosensory training	Body structure/function:
	Mean age = 63.4 years (SD n/a)	Exp: mechanical stimulation (stochastic resonance) was applied to the extensors of fingers, wrist and elbow, at the subsensory threshold level during task-specific activities.	•Fugl-Meyer Assessment
			•Reaching performance Scale
	Stoke phase: chronic	•Proprioception^*
	Type of stroke: not stated	•Vibration
		Con: the stochastic resonance device was deactivated after sensory threshold testing.	•Pinprick
			•Semmes Weinstein monofilaments*
		60 min×3/wk×4 wk	Activity:
			•Wolf Motor Function Test
			•Action Research Arm Test
			•Motor Activity Log^*
			Participation:
			•Stroke Impact Scale^*
			Adverse events: unserious adverse events
Sullivan et al. 2012	n = 38	Passive somatosensory training	Body structure/function:
	Mean age = 60.6 years (SD n/a)	Exp: home-based task-specific activities using glove electrode at the sensory threshold level.	•Fugl-Meyer Assessment
			•Arm Motor Ability Test
	Stroke phase: chronic	Con: Subjects in the sham group wore an electrode glove and electrode attached to the sham stimulator. The sham stimulators’ timer and lights were active; the amplitude was adjustable; however no current was delivered.	•Perceptual Threshold Test^*
	Type of stroke: not stated
		30 min×2/w×4 wk	Activity:
			•Motor Activity Log^*
			Participation:
			•Stroke Impact Scale^*
			Adverse events: not stated

n/a: data not available, Exp: experimental group, Con: control group, PPT: pressure pain threshold. ^*Outcome measurements written in bold were used in the meta-analyses.

3.2.3 Outcome measures

Somatosensory modalities analyzed in the studies were: light touch, pinprick, temperature, position sense, kinesthesia, vibration, two-point discrimination, stereognosis and graphesthesia. The most commonly used sensory outcome measurements were the Nottingham Sensory Assessment (Cambier et al., 2003; Sullivan et al., 2012) and monofilament testing (Semmes Weinstein and von Frey) (Acerra, 2007; Carey et al., 2011; Chen et al., 2005; Stein et al., 2010). The most commonly used motor outcome measurement was the upper limb motor section of the Fugl-Meyer Assessment (Cambier et al., 2003; Poole et al., 1990; Stein et al., 2010; Sullivan et al., 2012). Body structure and function was the most commonly evaluated ICF domain. While 6 studies (Acerra, 2007; Cambier et al., 2003; Chen et al., 2005; Sallés et al., 2017; Stein et al., 2010; Sullivan et al., 2012) evaluated the activity domain, the participation domain was only addressed in 3 studies (Acerra, 2007; Stein et al., 2010; Sullivan et al., 2012). The majority of the included studies (Byl et al., 2003; Cambier et al., 2003; Carey et al., 2011; Poole et al., 1990; Sallés et al., 2017; Sullivan et al., 2012) did not give information about adverse events. Two studies (Acerra, 2007; Chen et al., 2005) reported no adverse events and one study (Stein et al., 2010) reported 3 unserious adverse events, including skin irritation (1 patient), transient pain (1 patient) and shoulder pain (1 patient) (Table 2).

3.2.4 Interventions

Four studies used an active somatosensory training (Acerra, 2007; Byl et al., 2003; Carey et al., 2011; Sallés et al., 2017) and five studies (Cambier et al., 2003; Chen et al., 2005; Poole et al., 1990; Stein et al., 2010; Sullivan et al., 2012) used passive somatosensory training. Active somatosensory interventions included mirror therapy with sensorimotor tasks, sensory discrimination training and sensory discrimination tasks combined with Perfetti’s method. Passive somatosensory interventions consisted of therapy with intermittent pneumatic compression, thermal stimulation, electrical stimulation and stochastic resonance therapy. The mean duration of somatosensory intervention sessions was 38.89 minutes (SD = 22.05) (minimum: 20, maximum: 90 minutes). Active somatosensory interventions were delivered during minimum 10 and maximum 30 sessions. Passive somatosensory intervention sessions were provided with minimum 8 and maximum 30 sessions. The length of the active somatosensory interventions was between 2 to 10 weeks and the length of the passive somatosensory intervention was between 4 to 6 weeks (Table 2).

3.3 Effect of active somatosensory interventions

3.3.1 Somatosensation

When compared with sham or other therapy, there is a non-significant improvement in light touch (3 trials; SMD = 1.52, 95% CI = – 0.45 to 3.48) (Fig. 2A). When compared with sham or other therapy, there is a non-significant improvement in proprioception (2 trials; SMD = 0.20, 95% CI = – 0.83 to 1.23) (Fig. 2B). Finally, when compared with sham or other therapy, there is a non-significant improvement in higher cortical somatosensation (1 trial; SMD = 0.55, 95% CI = – 0.01 to 1.12) (Fig. 2C). We downgraded the quality of evidence for light touch and proprioception as very low since there was high heterogeneity between studies and one of the pooled studies had a very small sample size. The quality of evidence for higher cortical somatosensation was low (Table 3).

Fig.2

Effect of active somatosensory interventions on A) light touch, B) proprioception, C) higher cortical somatosensation.

Table 3

Active sensory training compared to sham therapy for somatosensory deficits in the upper limb following stroke

Outcomes	Anticipated absolute effects^* (95% CI)		No of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with sham therapy	Risk with active sensory training
Exteroception: Light touch assessed with: different instruments. (RNSA, Composite index of functional somatosensory discrimination capacity, Frey monofilaments) ³	The mean light touch impairment at the end of intervention phase for the control group was N/A	SMD 1.52 higher (0.45 lower to 3.48 higher)	87 (3 RCTs)	⊕⊖⊖ ⊖ VERY LOW ¹²	We are uncertain about the effect of active sensory training on light touch impairment.⁴
Proprioception assessed with: different instruments. (RNSA, Composite index of functional somatosensory discrimination capacity) ³	The mean proprioception impairment at the end of intervention phase for the control group was N/A	SMD 0.2 higher (0.83 lower to 1.23 higher)	58 (2 RCTs)	⊕⊖⊖ ⊖ VERY LOW ⁵⁶	We are uncertain about the effect of active sensory training on proprioception.⁷
Higher cortical somatosensation: Sensory discrimination assessed with: composite index of functional somatosensory discrimination capacity.	The mean stereognosis impairment at the end of intervention phase for the control group was N/A	SMD 0.55 higher (0.01 lower to 1.12 higher)	50 (1 RCT)	⊕⊖⊖ ⊖ LOW ⁸	Active sensory training does not result in an important reduction in sensory discrimination. ⁹

¹Heterogeneity between studies is large (I² = 91%, p = 0.0008). There is variation in size of effects of the pooled studies. ²One of the pooled studies is with small sample size. ³RNSA: Revised Nottingham Sensory Assessment. ⁴Active sensory training for this comparison: Motor and sensory discrimination tasks inside mirror box, sensory discrimination training, sensory tasks with cognitive activation. ⁵Large variation in size of effects. ⁶One of the pooled studies is with small sample size. ⁷Active sensory training for this comparison: Sensory tasks with cognitive activation and sensory discrimination training. ⁸This comparison includes only one study. ⁹Active sensory training for this comparison: sensory discrimination training.

3.3.2 Motor impairment

When compared with sham or other therapy, active somatosensory interventions showed a moderate significant and positive effect on upper limb motor impairment (2 trials; SMD = 0.73, 95% CI = 0.14 to 1.32) (Fig. 3A). We rated the evidence for this outcome as low quality (Table 4).

Fig.3

Effect of active somatosensory interventions on A) motor impairment and B) activity.

Table 4

Active sensory training compared to sham therapy for motor and functional outcomes in the upper limb following stroke

Outcomes	Anticipated absolute effects^* (95% CI)		No of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with sham therapy	Risk with active sensory training
Motor impairment assessed with: different instruments. (Grip strength, MI) ²	The mean motor impairment at the end of intervention phase for the control group was N/A	SMD 0.73 higher (0.14 higher to 1.32 higher)	48 (2 RCTs)	⊕⊖⊖ ⊖ LOW ¹	Active sensory training likely reduces motor impairment.³
Activity assessed with: different instruments. (MAS-Item 7, MESUPES-Hand) ⁴	The mean activity impairment at the end of intervention phase for the control group was N/A	SMD 0.16 higher (0.41 lower to 0.72 higher)	48 (2 RCTs)	⊕⊖⊖ ⊖ LOW ¹	Active sensory training is unlikely to improve activity.³

¹One of the pooled studies is with small sample size. ²MI: Motricity Index. ³Active sensory trainings for this comparison: Motor and sensory discrimination tasks inside mirror box, sensory tasks with cognitive activation. ⁴MAS: Motor Assessment Scale, MESUPES: Motor Evaluation Scale for Upper Extremity in Stroke Patients.

3.3.3 Activity

When compared with sham or other therapy, active somatosensory interventions did not significantly improve activity of daily living (ADL) (2 trials; SMD = 0.16, 95% CI = – 0.41 to 0.72) (Fig. 3B). The evidence for this outcome was rated as low quality (Table 4).

3.3.4 Participation

There was no study evaluating the effect on participation.

3.4 Effect of passive somatosensory interventions

3.4.1 Somatosensation

When compared with sham therapy, passive somatosensory interventions did not significantly improve light touch sensation (3 trials; SMD = 0.29, 95% CI = – 0.43 to 1.01) (Fig. 4A). Pooling of two trials using passive somatosensory interventions did not result in a significant improvement on proprioception (2 trials; SMD = 0.39, 95% CI = – 0.26 to 1.04) (Fig. 4B). There was also no significant improvement in stereognosis (2 trials; MD = 1.17, 95% CI = – 1.04 to 3.37) (Fig. 4C). We rated evidence for these outcomes as very low quality (Table 5). When compared to no additional therapy, there was a large, significant and positive effect for passive somatosensory interventions improving light touch sensation (2 trials; SMD = 1.13, 95% CI = 0.20 to 2.05) (Fig. 5). The evidence for this outcome was downgraded as very low since there was a variation in effect and one of the pooled studies had a small sample size (Table 6).

Fig.4

Effect of passive somatosensory interventions compared with sham therapy on A) light touch, B) proprioception, C) stereognosis.

Fig.5

Effect of passive somatosensory interventions compared with no additional therapy on light touch.

Table 5

Passive sensory training compared to sham therapy for somatosensory deficits in the upper limb following stroke

Outcomes	Anticipated absolute effects^* (95% CI)		No of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with sham therapy	Risk with passive sensory training
Exteroception: light touch assessed with: different instruments. (NSA, SWM, PTT) ⁴⁵⁶	The mean light touch impairment at the end of intervention phase for the control group was N/A	SMD 0.29 higher (0.43 lower to 1.01 higher)	91 (3 RCTs)	⊕⊖⊖ ⊖ VERY LOW ¹²	We are uncertain about the effect of passive sensory training on light touch.³
Proprioception assessed with: different instruments. (NSA, Proprioception)⁴	The mean proprioception impairment at the end of intervention phase for the control group was N/A	SMD 0.39 higher (0.26 lower to 1.04 higher)	53 (2 RCTs)	⊕⊖⊖ ⊖ VERY LOW ⁷⁸	We are uncertain about the effect of passive sensory training on proprioception.⁹
Higher cortical somatosensation: stereognosis assessed with: NSA. ⁴	The mean stereognosis. impairment at the end of intervention phase for the control group was 5.26.	MD 1.17 higher (1.04 lower to 3.37 higher)	61 (2 RCTs)	⊕⊖⊖ ⊖ VERY LOW ⁸¹⁰	We are uncertain about the effect of passive sensory training on stereognosis.¹¹

¹Heterogeneity is large. ²Small sample size. CI is wide. ³Passive sensory training for this comparison: conventional therapy combined with intermittent pneumatic compression treatment, mechanical stimulation at the sub-sensory threshold level during task-specific activities, task specific activities with glove electrode at the sensory threshold level. ⁴NSA: Nottingham Sensory Assessment ⁵SWM: Semmes Weinstein Monofilaments. ⁶PTT: Perceptual Threshold Test. ⁷Small sample size. Wide CI. ⁸Large variation in effect. ⁹Passive sensory training for this comparison: Conventional therapy combined with intermittent pneumatic compression treatment, mechanical stimulation at the sub-sensory threshold level during task-specific activities. ¹⁰Small sample size. CI is wide. ¹¹Passive sensory training for this comparison: Conventional therapy combined with intermittent pneumatic compression treatment, task-specific activities with glove electrode at the sensory threshold level.

Table 6

Passive sensory intervention compared to no additional therapy for somatosensory deficits in the upper limb following stroke

Outcomes	Anticipated absolute effects^* (95% CI)		No of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with no additional therapy	Risk with passive sensory intervention
Exteroception: light touch assessed with: different outcome measurements. (SWM, FMA-Sensation) ¹	The mean light touch impairment at the end of intervention phase for the control group was N/A	SMD 1.13 higher (0.2 higher to 2.05 higher)	47 (2 RCTs)	⊕⊖⊖ ⊖ VERY LOW ²³	Passive sensory intervention may reduce light touch impairment, but we are very uncertain.⁴

¹SWM: Semmes Weinstein Monofilaments, FMA: Fugl-Meyer Assessment. ²Large variation in effect. CIs do not overlap. ³Small sample size. Wide CI. ⁴Passive sensory training for this comparison: Thermal stimulation, inflatable pressure splint.

3.4.2 Motor impairment

There was no significant improvement of passive somatosensory interventions on upper limb motor impairment when compared with sham therapy (3 trials; SMD = 0.29, 95% CI = – 0.43 to 1.01) (Fig. 6A), or no additional therapy (2 trials; SMD = 0.39, 95% CI = – 0.45 to 1.23) (Fig. 7), respectively. We rated the evidence for these outcomes as very low (Tables 7and 8).

Fig.6

Effect of passive somatosensory interventions on A) motor impairment, B) activity, C) participation.

Fig.7

Effect of passive somatosensory interventions compared with no additional therapy on motor impairment.

Table 7

Passive sensory training compared to sham therapy for motor and functional outcomes in the upper limb following stroke

Outcomes	Anticipated absolute effects^* (95% CI)		No of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with sham therapy	Risk with passive sensory training
Body functions/structure: motor impairment assessed with: different outcome measurements	The mean motor impairment at the end of intervention phase for the control group was N/A	SMD 0.29 higher (0.43 lower to 1.01 higher)	91 (3 RCTs)	⊕⊖⊖ ⊖ VERY LOW ¹²	We are uncertain about the effect of passive sensory training on motor impairment.³
Activity assessed with: different outcome measurements.	The mean activity impairment at the end of intervention phase for the control group was N/A	SMD 0.22 higher (0.19 lower to 0.64 higher)	91 (3 RCTs)	⊕⊖⊖ ⊖ LOW ⁴	Passive sensory training probably does not reduce activity.³
Participation assessed with: different outcome measurements.	The mean participation impairment at the end of intervention phase for the control group was N/A	SMD 0.14 lower (0.61 lower to 0.34 higher)	68 (2 RCTs)	⊕⊖⊖ ⊖ LOW ⁵	Passive sensory training likely results in no difference in participation.⁶

¹CIs do not overlap. Large variation in effect. ²Small sample size. Wide CI. ³Passive sensory training for this comparison: Conventional therapy combined with intermittent pneumatic compression treatment, mechanical stimulation at the sub-sensory threshold level during task-specific activities, task-specific activities using glove electrode at the sensory threshold level. ⁴Small sample size. Wide CI. ⁵Small sample size. Wide CI. ⁶Passive sensory training for this comparison: Mechanical stimulation at the sub-sensory level during task-specific activities, task-specific activities using glove electrode at the sensory threshold level.

Table 8

Passive sensory training compared to no additional therapy for motor outcomes in the upper limb following stroke

Outcomes	Anticipated absolute effects^* (95% CI)		No of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with no additional therapy	Risk with passive sensory training
Body functions/structure: motor impairment assessed with: different outcome measurements	The mean motor impairment at the end of intervention phase for the control group was N/A	SMD 0.39 higher (0.45 lower to 1.23 higher)	47 (2 RCTs)	⊕⊖⊖ ⊖ VERY LOW ¹²	When compared to no additional therapy, we are very uncertain about the effect of passive sensory trainings on motor impairment.³

¹Large variation in effect. CIs do not overlap. ²Small sample size. Wide CI. ³Passive sensory training for this comparison: Thermal stimulation, inflatable pressure splint.

3.4.3 Activity

When compared with sham therapy, passive somatosensory interventions did not result in a significant improvement in activity (3 trials; SMD = 0.22, 95% CI = – 0.19 to 0.64) (Fig. 6B). The quality of evidence for this outcome was downgraded as low because of small sample size (Table 7).

3.4.4 Participation

When compared with sham therapy, passive somatosensory interventions did not significantly affect participation (2 trials; SMD = – 0.14, 95% CI = – 0.61 to 0.34) (Fig. 6C). We rated the evidence for this outcome as low quality (Table 7).

4 Discussion

The aim of this study was to update and summarize the effectiveness of somatosensory interventions on upper limb somatosensory impairments, motor impairments and upper limb activity limitations and participation in daily life following stroke. We identified nine RCTs in this review and included eight in our meta-analysis and found significant and beneficial effects of active somatosensory interventions on motor impairment and of passive somatosensory interventions on light touch. Although our results corroborate benefits of somatosensory interventions, there is currently insufficient evidence for conclusive results.

We found that active somatosensory interventions have beneficial effects on motor impairment (low quality of evidence). Although this analysis included two trials, we believe that the improvement in motor impairment is largely influenced by the study from Acerra (2007) evaluating mirror therapy since this type of therapy also includes a motor component (Michielsen et al., 2011). On the other hand, this might also suggest that approaches combining active somatosensory interventions and motor therapy are further to be investigated when evaluating the effect of somatosensory upper limb therapy. We also found that passive somatosensory interventions improved light touch in the upper limb (very low quality of evidence). Passive therapy in this analysis comprised of thermal stimulation (Chen et al., 2005) and inflatable pressure splinting (Poole et al., 1990) and we believe that thermal stimulation therapy largely affects the improvement in light touch. Information from the activated thermoreceptors is transmitted through the spinal cord until the spinothalamic tract reaches the thalamus and primary somatosensory cortex. This increased activation in the somatosensory cortex might affect the improvement in light touch since temperature and light touch information is carried through the same pathway.

Our findings vary from those of previous reviews in this field. Schabrun and Hillier (Schabrun & Hillier, 2009) included non-randomized trials, with several focusing on providing electrical stimulation, as well as patients with sensory deficits in the lower limb. They concluded that passive somatosensory training might be more beneficial than active training. However, our results showed that active somatosensory interventions might have a beneficial effect whereas passive somatosensory interventions showed no significant effect on motor impairment. In addition, our trials including passive therapy do not only include electrical stimulation studies. We identified only one electrical stimulation trial (Sullivan et al., 2012) which actually assessed somatosensation. Thus, the information from randomized controlled trials evaluating specifically somatosensory impairment is currently limited. In agreement with the findings in the Cochrane review (Doyle et al., 2010), we also found that active training including mirror therapy with sensorimotor tasks may be beneficial for improving light touch. Mirror therapy can be considered a feasible and effective method to support upper limb recovery post-stroke. Our reason for including this study was not related to the mirror therapy component but in this study, the experimental group performed sensory discrimination tasks in the mirror box. We therefore included this study, as it has a somatosensory component that met our inclusion criteria.

The analysis of the effect of active somatosensory interventions on participation seems lacking, since none of the included studies with an active therapy component evaluated this ICF domain. Measures of all ICF domains are key elements for patients with stroke. The impairments of the upper limb result in limited activity and participation restrictions following stroke. Therefore, RCTs including measurement tools in all three domains are warranted. In addition, very few studies mentioned adverse events which warrant attention when applying interventions in clinical settings.

Some limitations regarding our inclusion criteria should be considered. First, it should be noted that this review examined only published trials in English language, with three non-English studies identified but excluded. Second, our results rely on a limited number of trials, with a limited number of participants. In addition, although some studies identified in the search process did include an intervention with a somatosensory focus, we had to exclude those studies (Au-Yeung & Hui-Chan, 2014; Cameirao, Badia, Duarte, Frisoli, & Verschure, 2012; Carrico, Chelette, Westgate, Powell, et al., 2016; Carrico, Chelette, Westgate, Salmon-Powell, et al., 2016; Chanubol et al., 2012; de Diego, Puig, & Navarro, 2013; Dos Santos-Fontes, Ferreiro de Andrade, Sterr, & Conforto, 2013; Feys et al., 1998; Fleming et al., 2015; Hunter et al., 2011; Ikuno et al., 2012; Lee, Bae, Jeon, & Kim, 2015; Lin, Huang, Chen, Wu, & Huang, 2014; Wilson et al., 2016; Wolny, Saulicz, Gnat, & Kokosz, 2010; Wu et al., 2010), since they did not comprise a somatosensory outcome. Yet, the latter was our primary focus. The number of participants in the trials ranged from only eight to a maximum of 50, with a certain amount of heterogeneity in intervention and outcome measures. Meta-analyses with a high number of comparable studies are of course preferred and give the most robust and precise estimate of effect of an intervention. Standards in at least outcome measures used should be agreed upon and then applied in future research. Another drawback is the lack of data of included trials. One study (Byl et al., 2003) was excluded from the meta-analysis because of missing baseline and post intervention mean and SD values. Byl et al. (2003) focused on sensory re-educating in a crossover trial, which consisted of mainly sensory discrimination tasks (Table 2) and they reported that both groups had significant improvements in somatosensation of the upper limb. In addition, the order of the treatments (motor or sensory) had no effect on the results. Yet, we were not able to include data from the first phase of this study. Finally, although imputation of missing standard deviations is statistically valid, it gives only an estimate of results. We chose the largest SDs to impute results to be as strict as possible in our meta-analyses. Large SDs widen the confidence interval and may have yielded results towards a lack of effect.

Notwithstanding its limitations, it is important to emphasize that our review applied inclusion and exclusion criteria to critically and primarily focus on the effect of somatosensory interventions on somatosensation, as well as to differentiate effects according to active or passive somatosensory interventions provided and trials comparing an intervention to sham/other therapy or no additional therapy. As a consequence, this approach yielded a different number of studies from the Cochrane review; we included nine studies, whereas the Cochrane review comprised 13. Our review included 5 of their studies and, in addition, we identified 4 more studies (Carey et al., 2011; Sallés et al., 2017; Stein et al., 2010; Sullivan et al., 2012) adhering to our inclusion criteria, after the publication of the Cochrane review. Thus, we feel confident in providing the most up-to-date summary of the published literature in this emerging domain.

In conclusion, this systematic review and meta-analysis demonstrated significant and beneficial effects of active somatosensory interventions on upper limb motor impairment and passive somatosensory interventions on light touch, but with low and very low level of evidence respectively. Active somatosensory therapy may thus have a positive effect on upper limb motor recovery. These results emphasize the persistent need for further well-designed and high-quality studies investigating the effect of active somatosensory interventions on somatosensation, motor, functional and participation outcomes after stroke.

Conflict of interest

The authors declare that there are no conflicts of interest or funding regarding publication of this paper.

Footnotes

Appendix 1

PRISMA Checklist

Section /topic	#	Checklist item	Reported on page #
TITLE			1
Title	1	Identify the report as a systematic review, meta-analysis, or both.	1
ABSTRACT	1
Structured summary	2	Provide a structured summary including, as applicable: background; objectives; data sources; study eligibility criteria, participants, and interventions; study appraisal and synthesis methods; results; limitations; conclusions and implications of key findings; systematic review registration number.	1, 2
INTRODUCTION	3
Rationale	3	Describe the rationale for the review in the context of what is already known.	3, 4
Objectives	4	Provide an explicit statement of questions being addressed with reference to participants, interventions, comparisons, outcomes, and study design (PICOS).	5
METHODS	5
Protocol and registration	5	Indicate if a review protocol exists, if and where it can be accessed (e.g., Web address), and, if available, provide registration information including registration number.	–
Eligibility criteria	6	Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., years considered, language, publication status) used as criteria for eligibility, giving rationale.	5–7
Information sources	7	Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify additional studies) in the search and date last searched.	5
Search	8	Present full electronic search strategy for at least one database, including any limits used, such that it could be repeated.	5, 6, 7
Study selection	9	State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, included in the meta-analysis).	7, 8
Data collection process	10	Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators.	7, 8
Data items	11	List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made.	8
Risk of bias in individual studies	12	Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis.	8
Summary measures	13	State the principal summary measures (e.g., risk ratio, difference in means).	8, 9
Synthesis of results	14	Describe the methods of handling data and combining results of studies, if done, including measures of consistency (e.g., I²) for each meta-analysis.	8, 9
Risk of bias across studies	15	Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective reporting within studies).	–
Additional analyses	16	Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, meta-regression), if done, indicating which were pre-specified.	–
RESULTS	9
Study selection	17	Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally with a flow diagram.	9
Study characteristics	18	For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow-up period) and provide the citations.	9–11
Risk of bias within studies	19	Present data on risk of bias of each study and, if available, any outcome level assessment (see item 12).	–
Results of individual studies	20	For all outcomes considered (benefits or harms), present, for each study: (a) simple summary data for each intervention group (b) effect estimates and confidence intervals, ideally with a forest plot.	12–14
Synthesis of results	21	Present results of each meta-analysis done, including confidence intervals and measures of consistency.	12–14
Risk of bias across studies	22	Present results of any assessment of risk of bias across studies (see Item 15).	–
Additional analysis	23	Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, meta-regression [see Item 16]).	–
DISCUSSION	14
Summary of evidence	24	Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (e.g., healthcare providers, users, and policy makers).	14–16
Limitations	25	Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level (e.g., incomplete retrieval of identified research, reporting bias).	16–17
Conclusions	26	Provide a general interpretation of the results in the context of other evidence, and implications for future research.	17–18
FUNDING	–
Funding	27	Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the systematic review.	–

Appendix 2

Search strategy used for Ovid Medline and adapted for other databases

Searches	Results
1-exp cerebrovascular disorders/or exp basal ganglia cerebrovascular disease/or exp brain ischemia/or exp carotid artery diseases/or exp cerebral small vessel	332827
diseases/or exp cerebrovascular trauma/or exp intracranial arterial diseases/or exp intracranial arteriovenous malformations/or exp “intracranial embolism and thrombosis”/or exp intracranial hemorrhages/or exp vasospasm, intracranial/
2-exp stroke/or exp brain infarction/or exp stroke, lacunar/	112375
3-(stroke or post-stroke or post stroke or cerebrovasc$ or brain vasc$ or cerebral vasc$ or cva$ or apoplex$ or SAH).tw.	224777
4-exp hemiplegia/or paresis/	17595
5-(hemipleg$ or hemipar$ or paresis or paretic).tw.	29262
6-1 or 2 or 3 or 4 or 5	447125
7-exp Upper Extremity/	154927
8-(upper extremit$ or upper limb$).tw.	35140
9-(arm or shoulder or elbow or forearm or wrist or finger$).tw.	302361
10-7 or 8 or 9	406983
11-sensation/or exp proprioception/or exp thermosensing/or exp touch/	59103
12-sensation disorders/or exp somatosensory disorders/	24251
13-form perception/or exp stereognosis/or size perception/or touch perception/or weight perception/	23220
14-(stereognosis or touch or tactile or propriocept$ or kinesthesi$ or two point discriminitation or position sense).tw.	37467
15-(‘sensory relearn$ or ‘sensory learn$’ or ‘sensory train$’ or ‘sensory retrain$’ or ‘sensory educat$’ or somatosensation or sensation or ‘sensory re-educat$’ or ‘sensory stimulation’ or ‘somatosensory stimulation’).tw.	30551
16-11 or 12 or 13 or 14 or 15	146100
17-6 and 10 and 16	1084

References

Acerra

(2007). Sensorimotor dysfunction in CRPS1 and stroke: Characterisation, prediction and intervention. PhD thesis, School of Health and Rehabilitation Sciences, The University of Queensland, Australia.

Au-Yeung

S.S.

, & Hui-Chan

C.W.

(2014). Electrical acupoint stimulation of the affected arm in acute stroke: A placebo-controlled randomized clinical trial. Clin Rehabil, 28(2), 149–158. doi:10.1177/0269215513494875

Benjamin

E.J.

, Blaha

M.J.

, Chiuve

S.E.

, Cushman

, Das

S.R.

, Deo

, et al. (2017). Heart Disease and Stroke Statistics-2017 Update: A Report From the American Heart Association. Circulation, 135(10), e146–e603. doi:10.1161/CIR.0000000000000485

Bernhardt

, Hayward

K.S.

, Kwakkel

, Ward

N.S.

, Wolf

S.L.

, Borschmann

, Cramer

S.C.

(2017). Agreed definitions and a shared vision for new standards in stroke recovery research: The Stroke Recovery and Rehabilitation Roundtable taskforce. Int J Stroke, 12(5), 444–450. doi:10.1177/1747493017711816

Byl

, Roderick

, Mohamed

, Hanny

, Kotler

, Smith

, Abrams

(2003). Effectiveness of sensory and motor rehabilitation of the upper limb following the principles of neuroplasticity: Patients stable poststroke. Neurorehabil Neural Repair, 17(3), 176–191. doi:10.1177/0888439003257137

Cambier

D.C.

, De Corte

, Danneels

L.A.

, & Witvrouw

E.E.

(2003). Treating sensory impairments in the post-stroke upper limb with intermittent pneumatic compression. Results of a preliminary trial. Clin Rehabil, 17(1), 14–20. doi:10.1191/0269215503cr580oa

Cameirao

M.S.

, Badia

S.B.

, Duarte

, Frisoli

, & Verschure

P.F.

(2012). The combined impact of virtual reality neurorehabilitation and its interfaces on upper extremity functional recovery in patients with chronic stroke. Stroke, 43(10), 2720–2728. doi:10.1161/STROKEAHA.112.653196

Carey

, Macdonell

, & Matyas

T.A.

(2011). SENSe: Study of the effectiveness of neurorehabilitation on sensation: A randomized controlled trial. Neurorehabil Neural Repair, 25(4), 304–313. doi:10.1177/1545968310397705

Carrico

, Chelette

K.C.

, 2nd, Westgate

P.M.

, Powell

, Nichols

, Fleischer

, & Sawaki

(2016). Nerve Stimulation Enhances Task-Oriented Training in Chronic, Severe Motor Deficit After Stroke: A Randomized Trial. Stroke, 47(7), 1879–1884. doi:10.1161/STROKEAHA.116.012671

10.

Carrico

, Chelette

K.C.

, 2nd, Westgate

P.M.

, Salmon-Powell

, Nichols

, & Sawaki

(2016). Randomized Trial of Peripheral Nerve Stimulation to Enhance Modified Constraint-Induced Therapy After Stroke. Am J PhysMed Rehabil, 95(6), 397–406. doi:10.1097/PHM.0000000000000476

11.

Chanubol

, Wongphaet

, Chavanich

, Werner

, Hesse

, Bardeleben

, & Merholz

(2012). A randomized controlled trial of Cognitive Sensory Motor Training Therapy on the recovery of arm function in acute stroke patients. Clin Rehabil, 26(12), 1096–1104. doi:10.1177/0269215512444631

12.

Chen

J.-C.

, Liang , C.-C. , & Shaw

F.-Z.

(2005). Facilitation of Sensory and Motor Recovery by Thermal Intervention for the Hemiplegic Upper Limb in Acute Stroke Patients: A Single-Blind Randomized Clinical Trial. Stroke, 36(12), 2665–2669. doi:10.1161/01.STR.0000189992.06654.ab

13.

Chen

J.-C.

, & Shaw

F.-Z.

(2014). Progress in sensorimotor rehabilitative physical therapy programs for stroke patients. World Journal of Clinical Cases: WJCC, 2(8), 316–326. doi:10.12998/wjcc.v2.i8.316

14.

Cohen

(1988). Statistical power analysis for the behavioral sciences 2nd edn. Erlbaum Associates, Hillsdale, USA.

15.

Connell

L.A.

, Lincoln

N.B.

, & Radford

K.A.

(2008). Somatosensory impairment after stroke: Frequency of different deficits and their recovery. Clin Rehabil, 22(8), 758–767. doi:10.1177/0269215508090674

16.

de Diego

, Puig

, & Navarro

(2013). A sensorimotor stimulation program for rehabilitation of chronic stroke patients. Restor Neurol Neurosci, 31(4), 361–371. doi:10.3233/RNN-120250

17.

de Morton

N.A.

(2009). The PEDro scale is a valid measure of the methodological quality of clinical trials: A demographic study. Aust J Physiother, 55(2), 129–133. doi:10.1016/s0004-9514(09)70043-1

18.

Dos Santos-Fontes

R.L.

, Ferreiro de Andrade

K.N.

, Sterr

, & Conforto

A.B.

(2013). Home-based nerve stimulation to enhance effects of motor training in patients in the chronic phase after stroke: A proof-of-principle study. Neurorehabil Neural Repair, 27(6), 483–490. doi:10.1177/1545968313478488

19.

Doyle

, Bennett

, Fasoli

S.E.

, & McKenna

K.T.

(2010). Interventions for sensory impairment in the upper limb after stroke, Cochrane Database Syst Rev (6), CD006331. doi:10.1002/14651858.CD006331.pub2

20.

Feys

H.M.

, De Weerdt

W.J.

, Selz

B.E.

, Cox Steck

G.A.

, Spichiger

, Vereeck

L.E.

, Van Hoydonck

G.A.

(1998). Effect of a therapeutic intervention for the hemiplegic upper limb in the acute phase after stroke: A single-blind, randomized, controlled multicenter trial. Stroke, 29(4), 785–792.

21.

Findlater

S.E.

, & Dukelow

S.P.

(2017). Upper Extremity Proprioception After Stroke: Bridging the Gap Between Neuroscience and Rehabilitation. J Mot Behav, 49(1), 27–34. doi:10.1080/00222895.2016.1219303

22.

Fleming

M.K.

, Sorinola

I.O.

, Roberts-Lewis

S.F.

, Wolfe

C.D.

, Wellwood

, & Newham

D.J.

(2015). The effect of combined somatosensory stimulation and task-specific training on upper limb function in chronic stroke: A double-blind randomized controlled trial. Neurorehabil Neural Repair, 29(2), 143–152. doi:10.1177/1545968314533613

23.

Gardner

E.P.

, & Martin

J.H.

(2000). Coding of sensory information. In: Principles of Neural Science. McGraw-Hill, New York, USA.

24.

Grant

V.M.

, Gibson

, & Shields

(2017). Somatosensory stimulation to improve hand and upper limb function after stroke — a systematic review with meta-analyses. Top Stroke Rehabil, 25(2), 150–160. doi: 10.1080/10749357.2017.1389054

25.

Higgins

J.P.

, & Green

(2011). Cochrane handbook for systematic reviews of interventions (Vol. 4). The Cochrane Collaboration. Available from http://handbook.cochrane.org

26.

Hunter

S.M.

, & Crome

(2002). Hand function and stroke. Rev Clin Gerontol, 12(1), 68–81. doi:10.1017/S0959259802012194

27.

Hunter

S.M.

, Hammett

, Ball

, Smith

, Anderson

, Clark

, Pomeroy

V.M.

(2011). Dose-response study of mobilisation and tactile stimulation therapy for the upper extremity early after stroke: A phase I trial. Neurorehabil Neural Repair, 25(4), 314–322. doi:10.1177/1545968310390223

28.

Ikuno

, Kawaguchi

, Kitabeppu

, Kitaura

, Tokuhisa

, Morimoto

, Shomoto

(2012). Effects of peripheral sensory nerve stimulation plus task-oriented training on upper extremity function in patients with subacute stroke: A pilot randomized crossover trial. Clin Rehabil, 26(11), 999–1009. doi:10.1177/0269215512441476

29.

Kessner

S.S.

, Bingel

, & Thomalla

(2016). Somatosensory deficits after stroke: A scoping review. Top Stroke Rehabil, 23(2), 136–146. doi:10.1080/10749357.2015.1116822

30.

Laufer

, & Elboim-Gabyzon

(2011). Does Sensory Transcutaneous Electrical Stimulation Enhance Motor Recovery Following a Stroke? A Systematic Review. Neurorehabil Neural Repair, 25(9), 799–809. doi:10.1177/1545968310397205

31.

Lee

, Bae

, Jeon

, & Kim

(2015). The effects of cognitive exercise therapy on chronic stroke patients’ upper limb functions, activities of daily living and quality of life. J Phys Ther Sci, 27(9), 2787–2791. doi:10.1589/jpts.27.2787

32.

Liberati

, Altman

, Tetzlaff

, Mulrow

, Gøtzsche

, Ioannidis

, et al., (2009). The PRISMA Statement for Reporting Systematic Reviews and Meta-Analyses of Studies That Evaluate Health Care Interventions: Explanation and Elaboration, Ann Intern Med, 151(W65–W94). doi: 10.7326/0003-4819-151-4-200908180-00136

33.

Lin

K.C.

, Huang

P.C.

, Chen

Y.T.

, Wu

C.Y.

, & Huang

W.L.

(2014). Combining afferent stimulation and mirror therapy for rehabilitating motor function, motor control, ambulation, and daily functions after stroke. Neurorehabil Neural Repair, 28(2), 153–162. doi:10.1177/1545968313508468

34.

Michielsen

, Selles

, van der Geest

, Eckhardt

, Yavuzer

, Stam

, Smits

, Ribbers

, Bussmann

. (2011). Motor recovery and cortical reorganization after mirror therapy in chronic stroke patients: a phase II randomized controlled trial. Neurorehabil Neural Repair, 25(3), 223–33. doi: 10.1177/1545968310385127

35.

Meyer

, Karttunen

, Thijs

, Feys

, Verheyden

. (2014). How do somatosensory deficits in the arm and hand relate to upper limb impairment, activity, and participation problems after stroke? A systematic review. Phys Ther, 94(9), 1220–31. doi: 10.2522/ptj.20130271

36.

Meyer

(2015) Recovery after stroke: Long term outcome and upper limb somatosensory function. PhD thesis, Doctoral School Biomedical Sciences, KU Leuven, Belgium.

37.

Meyer

, De Bruyn

, Krumlinde-Sundholm

, Peeters

, Feys

, Thijs

, Verheyden

(2016a). Associations Between Sensorimotor Impairments in the Upper Limb at 1 Week and 6 Months After Stroke. J Neurol Phys Ther, 40(3), 186–95. doi: 10.1097/NPT.0000000000000138

38.

Meyer

, De Bruyn

, Lafosse

, Van Dijk

, Michielsen

, Thijs

, Truyens

, Oostra

, Krumlinde-Sundholm

, Peeters

, Thijs

, Feys

, Verheyden

. (2016b). Somatosensory Impairments in the Upper Limb Poststroke: Distribution and Association With Motor Function and Visuospatial Neglect. Neurorehabil Neural Repair, 30(8), 731–42. doi:10.1177/1545968315624779

39.

World Health Organization (2001). International Classification of Functioning, Disability and Health: ICF. Geneva, Switzerland.

40.

Poole

J.L.

, Whitney

S.L.

, Hangeland

, & Baker

(1990). The Effectiveness of Inflatable Pressure Splints on Motor Function in Stroke Patients. OTJR: Occupation, Participation and Health, 10(6), 360–366. doi:10.1177/153944929001000605

41.

Ridding

M.C.

, McKay

D.R.

, Thompson

P.D.

, & Miles

T.S.

(2001). Changes in corticomotor representations induced by prolonged peripheral nerve stimulation in humans. Clin Neurophysiol, 112(8), 1461–1469. doi:10.1016/S1388-2457(01)00592-2

42.

Sacco

R.L.

, Kasner

S.E.

, Broderick

J.P.

, Caplan

L.R.

, Culebras

, Elkind

M.S.

, Hoh

B.L.

(2013). An updated definition of stroke for the 21st century. Stroke, 44(7), 2064–2089.

43.

Sallés

, Martin-Casas

, Gironés

, Dura

M.J.

, Lafuente

J.V.

, & Perfetti

(2017). A neurocognitive approach for recovering upper extremity movement following subacute stroke: A randomized controlled pilot study. J Phys Ther Sci, 29(4), 665–672. doi:10.1589/jpts.29.665

44.

Schabrun

, & Hillier

(2009). Evidence for the retraining of sensation after stroke: A systematic review. Clin Rehabil, 23(1), 27–39. doi:10.1177/0269215508098897

45.

Stefan

, Kunesch

, Cohen

L.G.

, Benecke

, & Classen

(2000). Induction of plasticity in the human motor cortex by paired associative stimulation. Brain, 123(3), 572–584. doi:10.1093/brain/123.3.572

46.

Stein

, Hughes

, D’Andrea

, Therrien

, Niemi

, Krebs

, Harry

(2010). Stochastic resonance stimulation for upper limb rehabilitation poststroke. Am J Phys Med Rehabil, 89(9), 697–705. doi:10.1097/PHM.0b013e3181ec9aa8

47.

Sullivan

J.E.

, Hurley

, & Hedman

L.D.

(2012). Afferent stimulation provided by glove electrode during task-specific arm exercise following stroke. Clin Rehabil, 26(11), 1010–1020. doi:10.1177/0269215512442915

48.

Tyson

S.F.

, Hanley

, Chillala

, Selley

A.B.

, & Tallis

R.C.

(2008). Sensory loss in hospital-admitted people with stroke: Characteristics, associated factors, and relationship with function. Neurorehabil Neural Repair, 22(2), 166–172. doi:10.1177/1545968307305523

49.

Wilson

R.D.

, Page

S.J.

, Delahanty

, Knutson

J.S.

, Gunzler

D.D.

, Sheffler

L.R.

, & Chae

(2016). Upper-Limb Recovery After Stroke: A Randomized Controlled Trial Comparing EMG-Triggered, Cyclic, and Sensory Electrical Stimulation. Neurorehabil Neural Repair, 30(10), 978–doi:10.1177/1545968316650278

50.

Wolny

, Saulicz

, Gnat

, & Kokosz

M.Ç.

(2010). Butler’s neuromobilizations combined with proprioceptive neuromuscular facilitation are effective in reducing of upper limb sensory in late-stage stroke subjects: A three-group randomized trial. Clin Rehabil, 24(9), 810. doi:10.1177/0269215510367561

51.

H.C.

, Lin

Y.C.

, Hsu

M.J.

, Liu

S.M.

, Hsieh

C.L.

, & Lin

J.H.

(2010). Effect of thermal stimulation on upper extremity motor recovery 3 months after stroke. Stroke, 41(10), 2378–2380. doi:10.1161/STROKEAHA.110.593673

Study	Random allocation	Concealed allocation	Groups similar at baseline	Participant blinding	Therapist blinding	Assessor blinding	<15% dropouts	Intention-to-treat analysis	Between-group difference reported	Point estimate and variability reported	Total (0 to 10)
Acerra N 2007	Y	N	Y	N	N	Y	Y	N	Y	Y	6
Byl et al. 2003	Y	N	Y	N	N	Y	Y	N	N	N	4
Cambier et al. 2003	Y	N	Y	N	N	Y	N	N	Y	Y	5
Carey et al. 2011	Y	Y	Y	Y	N	Y	Y	N	Y	Y	8
Chen et al. 2005	Y	Y	Y	N	N	Y	N	N	Y	Y	6
Poole et al. 1990	Y	N	Y	N	N	N	Y	N	Y	Y	5
Salles et al. 2017	Y	N	Y	N	N	Y	Y	N	Y	Y	6
Stein et al. 2010	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	10
Sullivan et al. 2012	Y	N	Y	Y	N	Y	Y	Y	Y	Y	8