836461

research-article2019
CRE0010.1177/0269215519836461Clinical RehabilitationChia et al.

CLINICAL
Original Article REHABILITATION

Clinical Rehabilitation

Sensory retraining of the leg 2019, Vol. 33(6) 964­–979

© The Author(s) 2019

after stroke: systematic review Article reuse guidelines:

and meta-analysis
sagepub.com/journals-permissions
https://doi.org/10.1177/0269215519836461
DOI: 10.1177/0269215519836461
journals.sagepub.com/home/cre

Fenny SF Chia1,2 , Suzanne Kuys1

and Nancy Low Choy1,3

Abstract
Objective: This systematic review aimed to investigate the effects of interventions intended for retraining
leg somatosensory function on somatosensory impairment, and secondary outcomes of balance and gait,
after stroke.
Data sources: Databases searched from inception to 16 January 2019 included Cochrane Library,
PubMed, MEDLINE, CINAHL, EMBASE, PEDro, PsycINFO, and Scopus. Reference lists of relevant
publications were also manually searched.
Review methods: All types of quantitative studies incorporating interventions that intended to
improve somatosensory function in the leg post stroke were retrieved. The Quality Assessment Tool for
Quantitative Studies was used for quality appraisal. Standardised mean differences were calculated and
meta-analyses were performed using preconstructed Microsoft Excel spreadsheets.
Results: The search yielded 16 studies, comprising 430 participants, using a diverse range of interventions.
In total, 10 of the included studies were rated weak in quality, 6 were rated moderate, and none was rated
strong. Study quality was predominantly affected by high risk of selection bias, lack of blinding, and the use
of somatosensory measures that have not been psychometrically evaluated. A significant heterogeneous
positive summary effect size (SES) was found for somatosensory outcomes (SES: 0.52; 95% confidence
interval (CI): 0.04 to 1.01; I2 = 74.48%), which included joint position sense, light touch, and two-point
discrimination. There was also a significant heterogeneous positive SES for Berg Balance Scale scores (SES:
0.62; 95% CI: 0.10 to 1.14; I2 = 59.05%). Gait SES, mainly of gait velocity, was not significant.
Conclusion: This review suggests that interventions used for retraining leg somatosensory impairment
after stroke significantly improved somatosensory function and balance but not gait.

Keywords
Systematic review, somatosensory, retraining, lower limb, stroke

Received: 27 June 2018; accepted: 16 February 2019

1School of Physiotherapy, Faculty of Health Sciences, Corresponding author:

Australian Catholic University, Brisbane, QLD, Australia Fenny SF Chia, Community Rehabilitation Unit, Tasmanian
2Community Rehabilitation Unit, Tasmanian Health Service, Health Service, 90 Davey Street, Hobart, TAS 7000,
Hobart, TAS, Australia Australia.
3Queensland Health Service, Brisbane, QLD, Australia Email: [email protected]; [email protected]
Chia et al. 965

Introduction effects of interventions for post-stroke leg soma-

Somatosensory impairment is common after tosensory impairment remain unclear. In addition,
stroke, occurring in up to 89% of stroke survivors.1 the first review12 was critiqued for including stud-
Proprioception and tactile somatosensation are ies with participants without somatosensory
more impaired in the leg than in the arm post impairment, and that did not report somatosensory
stroke,2 with the frequency increasing with increas- outcomes.15 Therefore, a targeted systematic
ing level of weakness and stroke severity.2,3 Leg review, addressing the limitations of previous
somatosensory impairment also has a significant reviews, is required to elucidate the effects of inter-
impact on independence in daily activities3 and ventions for post-stroke leg somatosensory
activity participation in stroke survivors,4 as well impairment.
as predicts longer hospital stays and lower fre- It is of interest to clinicians and researchers to
quency of home discharges.5 evaluate the effects of leg somatosensory retraining
Leg somatosensory impairment negatively on factors that may ultimately influence activity
influences balance and gait. Post-stroke plantar and participation, as this could change practice.
tactile deficits correlate with lower balance scores Therefore, this systematic review aimed to exam-
and greater postural sway in standing.6 Tactile and ine the effects of post-stroke leg somatosensory
proprioceptive feedback provide critical informa- retraining on somatosensory impairment, balance,
tion about weight borne through the limb.7 gait, motor impairment, and leg function.
Accordingly, tactile and proprioceptive somatosen-
sory deficits may hinder paretic limb load detection
Methods
ability, potentially leading to reduced weight-bear-
ing and contributing to balance impairment and A protocol was registered in the PROSPERO
falls post stroke.8 Indeed, stroke survivors with International Prospective Register of Systematic
somatosensory impairment have a higher falls inci- Reviews16 (registration no. CRD42017058993)
dence compared to those without somatosensory prior to commencement of this systematic review.
impairment.3 In addition to reduced balance, The PRISMA guidelines17 were utilised in the pro-
impaired load detection may also contribute to gait cess and reporting of this review.
asymmetry, particularly in the push-off phase.8 In For the purpose of this review, the authors
addition, leg proprioception influences variance in defined somatosensory function as the ability to
stride length, gait velocity,9 and walking endurance detect, discriminate, and recognise body sensa-
in stroke survivors.10 In fact, leg somatosensory tions.18 Somatosensory modalities affected by
impairment has been shown to be the third most stroke that have been previously reported were
important independent factor for reduced gait considered, including detection or localisation of
velocity in stroke survivors.11 tactile stimuli, proprioception or kinaesthesia, ste-
Two systematic reviews have previously inves- reognosis or object recognition, pressure or weight
tigated the effects of interventions for retraining discrimination, detection of vibration, texture dis-
somatosensory function after stroke.12,13 In the first crimination, and two-point discrimination.3,13,18–21
review, published more than a decade ago, only Retraining of somatosensory function was defined
four of the 14 included studies targeted the leg,12 as any interventions that addressed the remediation
while the second only included studies of the arm.13 of the above-mentioned somatosensory modalities.
Nevertheless, both reviews reported that there were Intervention methods included elements of educa-
insufficient data to determine the effectiveness of tion; repetitive practice and feedback in detecting,
these interventions. A third systematic review eval- localising, discriminating, or recognising different
uating the effectiveness of proprioceptive train- sensory stimuli, pressure, or objects; proprioceptive
ing14 only included 16 studies with stroke-specific training; and somatosensory stimulation.12,22,23
populations, of which only two specifically Electronic databases including Cochrane
addressed the leg. From these three reviews, the Library, PubMed, MEDLINE, CINAHL, EMBASE,
966 Clinical Rehabilitation 33(6)

PEDro, PsycINFO, and Scopus were searched to Semmes–Weinstein monofilaments),25 global

identify relevant publications, from inception to 16 measures of sensation (e.g. Nottingham Sensory
January 2019. The search strategy (Supplemental Assessment),26 and sensory subscales of impair-
Table S1) was developed in collaboration with a ment-based measures (e.g. Fugl-Meyer Assessment
librarian by breaking down the review question into (FMA)).27
components: population, interventions, compara- Data were extracted by one author (F.S.F.C.)
tors, outcomes, and study design (PICOS).24 using a purpose-designed spreadsheet. A second
Identification of key search terms was followed, author (S.K.) checked the data for accuracy. The
using synonyms and variants of the search terms. following data were extracted from each study:
The search strategy was trialled on several data- study information (author(s), year of publication,
bases and adjusted accordingly to maximise the location of study, study design), participant base-
sensitivity of the search. Two reviewers (F.S.F.C., line information (demographics and characteris-
S.K.) independently screened titles and abstracts of tics), details of training intervention and dosage,
the studies yielded from the searches to assess for details of control conditions (if any), follow-up
eligibility. Full-text publications of potentially eli- period (if any), adverse effects, primary and sec-
gible studies were retrieved and further evaluated ondary outcomes, and study results. Missing infor-
by the same two reviewers. In addition, reference mation required for data analysis was requested
lists of relevant publications, including available from the study authors.
systematic reviews, the included studies, and narra- Each included study was assessed for quality
tive reviews, were manually searched for eligible using the Quality Assessment Tool for Quantitative
articles. Studies,28 which is valid and reliable.29 Two review-
Literature search was restricted to humans and ers (F.C.F.S., S.K.) assessed the quality of the articles
adults (19 years and above), and only studies pub- independently, and discrepancies were discussed
lished in English were included. There were no until consensus was reached. The Quality Assessment
restrictions to publication date and study setting. Tool for Quantitative Studies Dictionary30 was uti-
Studies were included if participants had leg soma- lised to guide ratings. Assessment components were
tosensory impairment following a stroke, with no selection bias, study design, confounders, blinding,
restrictions to the stage (acute, subacute, or data collection methods, withdrawals and drop-outs,
chronic), category (ischaemic or haemorrhagic), or intervention integrity, and analyses. All components
anatomical location of stroke. All types of quantita- except intervention integrity and analyses were given
tive studies incorporating interventions that aimed a rating of strong, moderate, or weak. A global rating
to improve leg somatosensory function after stroke was awarded based on ratings of the six components
were included. Studies that did not measure soma- – strong for no weak ratings, moderate for one weak
tosensory impairment or employed somatosensory rating, and weak for two or more weak ratings.
stimulation that produced muscle contraction were Descriptive analyses of the included studies
excluded, as muscle contraction could have been a were summarized. Effect sizes, pooled standard
confounding factor. Other reasons for exclusion deviations (SDs), and P-values (two-tailed) of con-
were studies evaluating assessment tools, observa- trolled clinical trials were calculated using a pre-
tional studies not investigating outcome of inter- constructed Microsoft Excel spreadsheet called
ventions, descriptive studies, expert opinions, Effect Size Calculator.31 Standardised mean differ-
qualitative studies, systematic reviews, conference ence (SMD) was used as the effect size to enable
abstracts, and unpublished studies. analysis of similar outcome measures with differ-
The primary outcome was somatosensory ent scales.32 Calculations for each study were based
impairment. Secondary outcomes were balance, on post-intervention outcomes, at the latest time
gait, motor impairment, and leg function. Any points, as recommended in the Cochrane handbook
measure of somatosensory impairment was consid- for systematic reviews of interventions.33 Effect
ered, including modality-specific measures (e.g. size bias was corrected using Hedges’ g,34 from
Chia et al. 967

which the 95% confidence intervals (CIs) were stimulation41,48 and transcutaneous electrical nerve
derived. The null hypothesis was rejected if the stimulation (TENS).41,45,49 Intervention dosages
P-value was less than 0.05. Heterogeneity between ranged from a single dose lasting up to approxi-
studies was assessed by calculating the I2 statistic35 mately two hours49 to 30 minutes daily for six
using another preconstructed Microsoft Excel months.39
spreadsheet.36 An I2 value greater than 50% was Outcome measures used in the included studies
considered of substantial heterogeneity.35 A meta- are described in Table 1. Common somatosensory
analysis was conducted by pooling the Hedges’ g modalities were light touch, measured with
values to calculate the summary effect size (SES).24 Semmes–Weinstein monofilaments in three stud-
A random-effects model of meta-analysis was ies, and joint position sense (JPS), measured in five
applied as it was expected that heterogeneity studies with the Biodex equipment,50 and in three
between studies would be relatively high. Subgroup other studies with the distal proprioception test
analyses were conducted for studies using similar (DPT). The most common measure for balance
outcome measures. In studies that had more than was the Berg Balance Scale (BBS), used in five
one outcome measure using the same assessment studies. Gait velocity was the main outcome meas-
tool within the same sample (e.g. light touch meas- ure for gait. Motor impairment measures included
ured at multiple sites of the limb), a hierarchy of the modified Motor Assessment Scale and ankle
the preferred measure was set up a priori and only strength. Leg function measures included Timed
the SMD based on the measure highest on the list Up and Go (TUG), Barthel Index, and the Iowa
was calculated.37 A narrative summary was pro- Level of Assistance Scale.
vided for data not statistically analysed (e.g. stud-
ies without a control group or used a paired design,
Quality appraisal
data not available, discrete data).
A summary of the quality appraisal of included stud-
ies is presented in Supplemental Table S2. Of the 16
Results included studies, 10 were rated as weak and six rated
A flow diagram of the study selection process is as moderate. None of the studies had a strong rating.
presented in Figure 1. A total of 16 studies were In the selection bias component, four studies scored
included at the end of the selection process. somewhat likely to be representative of the target
population,22,38,44,46 and none of the studies scored
very likely, due to incomplete reporting of recruit-
Description of included studies ment processes. A total of 14 studies were rated
Characteristics of the 16 included studies are dis- strong in study design for being controlled trials.
played in Table 1. There were a total of 430 partici- However, only five of these trials described the ran-
pants ranging in age from 18 to 82 years. Time domisation method.22,44,47,49,51 The four studies that
since stroke ranged from seven days38 to nearly were rated as weak in the confounders component
16 years.39 Eight studies were set in inpatient reha- (i.e. confounders not accounted for) either did not
bilitation,22,38,40–45 and the included studies were provide sufficient information to ascertain whether
executed in nine different countries. or not there were important between-group baseline
Three studies used a similar retraining approach, differences,23,46,48 or there were important differ-
which included a combination of education, detec- ences that were unaccounted for including sex and
tion, localization, discrimination, recognition, and age.45 Only three studies had blinding of both out-
proprioception of the hemiparetic leg.22,23,41 There come assessors and study participants.22,44,47 In total,
was a range of proprioceptive training strategies, 10 studies were rated as strong for data collection
including treadmill training with visual depriva- methods and one received a weak rating.44 Five stud-
tion,43 compelled body weight shift,38,46 and aquatic ies were rated separately for somatosensory meas-
gait training.47 Other interventions included vibration ures and secondary outcomes due to the use of
968 Clinical Rehabilitation 33(6)

Figure 1.  PRISMA flow diagram.17

measures with a range of psychometric properties, scored a strong rating in withdrawals and drop-outs.
all receiving a strong rating for balance or gait Only two studies reported percentage of compli-
measures, and a weak rating for somatosensory ance with treatment protocol,39,47 and only two of
measures.22,23,39,41,45 The majority of studies reported the randomised trials performed an intention-to-
80%–100% of participants completing the study and treat analysis.22,47
 Chia et al.

Table 1.  Characteristics of included studies.

Primary author Study design and Participants Intervention and dosage Control and dosage Outcome measures Global quality
(year) country/setting rating

Aruin (2012)46 RCT; USA/NR •• n = 18 (9 exp/9 con) Compelled body weight shift CPT •• FMA-LE Weak
•• Age = 35–75 years (0.6 cm shoe lift in unaffected 1 hour, 1×/week •• BBS
•• Time since stroke = 1.1–14.1 years foot) + CPT •• %BW on affected side
All days, 6 weeks •• 10MWT gait velocity
Follow-up = 3 months
Cordo (2009)39 Pretest–posttest; •• n = 12a Assisted Movement with Enhanced NA •• Days to 90% recovery of Weak
USA/community •• Age = 38–75 years Sensation robotic treatment JPS and ankle strength
•• Time since stroke = 1.9–15.6 years (cycled at 5°/second through •• %BW on affected side
±17.5°) + electromyography •• Gait velocity, cadence, and
feedback on affected ankle stride length
30 minutes/day, 6 months Follow-up = 6 months
Da Silva Ribeiro RCT; Brazil/NR •• n = 30 (15 exp/15 con) Nintendo Wii virtual rehab CPT •• FMA-LE (sensation, Moderate
(2015)51 •• Age = 18–60 years 60 minutes, 2×/week, 2 months 60 minutes, 2×/week, balance, leg motor, and leg
•• Time since stroke = mean 42.1 months 2 months coordination subscores)
(exp), 60.4 months (con) No follow-up
Han (2013)40 CCT; Korea/ •• n = 62 (31 exp/31 con) Unstable surface (proprioceptive Unstable surface •• JPS error (Biodex Weak
inpatient rehab •• Mean age = 56.1 years (exp), 56.6 years sense) exercise underwater exercise on land electrogoniometer)
(con) 40 minutes, 3×/week, 6 weeks 40 minutes, 3×/week, •• BBS
•• Time since stroke = mean 15.2 months 6 weeks •• Postural sway area EO
(exp), 16.1 months (con) and EC
No follow-up
Hillier (2006)23 Single-case, •• n = 3 Education, detection, localisation, NA •• SWM light touch Weak
repeated •• Age = 63–73 years discrimination, recognition, and •• DPT
measures; •• Time since stroke = >2 years proprioception training on affected •• Postural sway path length
Australia/NR foot No follow-up
45 minutes/session 3×/week,
2 weeks
Huzmeli (2017)41 CCT; Turkey/ •• n = 26 (13 exp/13 con) Proprioception, localisation, Neurodevelopmental •• SWM light touch Weak
inpatient rehab •• Age = 45–65 years vibration, pressure discrimination, rehab •• 2-point discrimination
•• Time since stroke = mean 40.2 months and TENS (parameters not 45 minutes/session 10 •• DPT
(exp), 20.1 months (con) described) on affected posterior sessions, 2 weeks •• Heat/cold senses
thigh + neurodevelopmental rehab •• BBS
45 minutes/session 10 sessions, •• FRT
2 weeks •• Barthel Index
Follow-up = 2 weeks
969
 970

Table 1. (Continued)

Primary author Study design and Participants Intervention and dosage Control and dosage Outcome measures Global quality
(year) country/setting rating

Kwon (2013)52 RCT; Korea/NR •• n = 45 RPM or RAM exercises on affected No exercise •• Knee JPS error – passive Weak
(RPM 15/RAM 15/Con 15) leg; angle speed 120°/second; angle repositioning
• Mean age = 58.7 years (RPM), 61.2 years 10°–100° •• Knee JPS error
(RAM), 63.3 years (con) 15 minutes in total, 3 sets of 60 – active angle
•• Time since stroke = mean 10.1 months repetitions repositioning (Biodex
(RPM), 10.6 months (RAM), 10.9 months electrogoniometer)
(con) No follow-up
Lee (2015)42 RCT; Korea/ •• n = 36 (18 exp/18 con) Motor imagery Proprioceptive training •• JPS error (Dualer IQ Weak
inpatient rehab •• Age = <65 years (n = 28), ≥65 years (n = 8) (5 minutes) + proprioceptive (30 minutes) inclinometer)
•• Time since stroke = mean 11.5 months training (25 minutes) 30 minutes/session •• BBS
(exp), 11.6 months (con) 30 minutes/session 5×/week, 5×/week, 8 weeks •• Affected/unaffected WB
8 weeks ratio
•• TUG
No follow-up
Lynch (2007)22 RCT; Australia/ •• n = 21 (10 exp/11 con) Education, detection, localisation, Relaxation + standing •• SWM light touch Moderate
inpatient rehab •• Age = 21–82 years discrimination, and proprioception with EC •• DPT
•• Time since stroke = 13–122 days on affected big toe and ankle 30 minutes/session 10 •• BBS
30 minutes/session 10 sessions, sessions, 2 weeks •• 10MWT gait velocity
2 weeks •• ILAS
Follow-up = 2 weeks
Mazuchi (2018)47 RCT; Brazil/NR •• n = 12 Aerobic deep water walking Aerobic treadmill •• Knee JPS absolute Moderate
•• Age = 40–65 years training (stationary gait) walking training error (Biodex isokinetic
•• Time since stroke = mean 67.6 months 30 minutes/session 1st week (50% 30 minutes/session 1st dynamometer)
(exp), 56.6 months (con) heart rate), 40 minutes/session week (50% heart rate), No follow-up
2nd-9th week (60% heart rate); 40 minutes/session
5-minute warm-up and 5-minute 2nd-9th week (60%
cool-down; 3×/week; 9 weeks heart rate); 5-minute
warm-up and 5-minute
cool-down; 3×/week;
9 weeks
Mohapatra RCT; USA/ •• n = 11 (5 exp/6 con) Compelled body weight shift CPT •• FMA-LE Moderate
(2012)38 inpatient rehab •• Mean age = 49.2 years (0.6 cm shoe lift in unaffected 90 minutes/session •• BBS
•• Time since stroke = 8–13 days (exp), foot) + CPT weekdays, 30 minutes/ •• %BW on affected side
7–45 days (con) 90 minutes/session weekdays, session Saturdays, 6×/ •• Gait velocity
30 minutes/session Saturdays, 6×/ week, 2 weeks No follow-up
week, 2 weeks

(Continued)
Clinical Rehabilitation 33(6)
Table 1. (Continued)
Chia et al.

Primary author Study design and Participants Intervention and dosage Control and dosage Outcome measures Global quality
(year) country/setting rating

Moon (2015)43 CCT; Korea/ •• n = 30 (15 exp/15 con) Treadmill training with EC Treadmill training •• Knee JPS error (Biodex Weak
inpatient rehab •• Mean age = 55.7 (exp), 50.1 (con) 3 × 10 minutes/session 3×/week, with EO isokinetic dynamometer)
•• Time since stroke = mean 33.3 months 4 weeks 3 × 10 minutes/session No follow-up
(exp), 36.0 months (con) 3×/week, 4 weeks
Morioka RCT; Japan/ •• n = 26 (12 exp/14 con) Hardness discrimination CPT •• Hardness discrimination Moderate
(2003)44 inpatient rehab •• Age = 51–79 years perceptual learning exercises NR •• 2-point discrimination
•• Time since stroke = 31–111 days (3 different levels of rubber sponge •• Postural sway locus length
hardness) + CPT EO and EC
10 trials/session 10 sessions, No follow-up
2 weeks
Peurala (2002)45 CCTb; Finland/ •• n = 59c TENS (monophasic constant, NA (no subjects for •• Limb skin sensitivity (VAS) Weak
inpatient rehab (foot 19 exp/0 con; hand 32 exp/8 con) frequency 50 Hz) using sock foot con) •• SEP to posterior tibial nerve
•• Mean age = 54.4 years electrode on affected foot + CPT •• 10MWT gait velocity
•• Time since stroke = 7 months–14 years 20 minutes/session 2×/day, •• MMAS
3 weeks No follow-up
Tyson (2013)49 Paired-sample •• n = 29 TENS (biphasic symmetrical, Placebo TENS •• Ankle proprioception Moderate
randomised •• Age = 28–82 years phase duration 50 μs, frequency Not specifiedd detection threshold
cross-over trial; •• Time since stroke = NR 70–130 Hz, 5 seconds/cycle) using (Biodex isokinetic
UK/research sock electrode on affected foot; dynamometer)
facility Not specifiedd •• FRT
•• 10MWT gait velocity
•• Ankle strength
No follow-up
Zankel (1969)48 CCT; Columbia/ •• n = 40e (10 exp/10 con) Vibration (120 cycles/second) on No vibration •• SWM light touch Weak
NR •• Age = 39–79 years affected sole NA •• FMA-LE
•• Time since stroke = NR 30 minutes, 10 stimuli/minute, •• BBS
1 month •• 10MWT gait velocity
No follow-up

RCT: randomized controlled trial; CCT: controlled clinical trial; NR: not reported; exp: rehab: rehabilitation; experimental group; con: control group; RPM: repeated passive movement; RAM:
repeated active movement; CPT: conventional physiotherapy treatment; TENS: transcutaneous electrical nerve stimulation; NA: not applicable; EC: eyes closed; EO: eyes open; FMA-LE: Fugl-Meyer
Assessment for lower extremity; BBS: Berg Balance Scale; %BW: percentage of body weight; 10MWT: 10-metre walk test; JPS: joint position sense; SWM: Semmes–Weinstein monofilaments; DPT:
distal proprioception test; FRT: Functional Reach Test; TUG: Timed Up and Go Test; ILAS: Iowa Level of Assistance Scale; VAS: visual analogue scale; SEP: somatosensory evoked potentials; MMAS:
Modified Motor Assessment Scale.
aOnly outcomes of subjects receiving lower limb treatment were extracted.
bNote that there was no control group for the foot.
cOnly data of the foot were analysed in this review.
dUntil subjects felt comfortable with the sensation and outcome measures were obtained while stimulated.
eCharacteristics of 40 recruited subjects were reported in the article, although only 20 subjects were selected for the study.
971
972 Clinical Rehabilitation 33(6)

Figure 2. Hedges’ g (95% CI) and summary of effect size (95% CI) on somatosensory outcomes.
SES: summary effect size; CI: confidence interval; RPM: repeated passive movement; RAM: repeated active movement; PAR: pas-
sive angle repositioning; AAR: active angle repositioning.
The squares on the forest plot are of the same size, instead of proportional to study weight, as the forest plot was generated on
Microsoft Excel.

Intervention effects sensitivity analysis was conducted with the exclu-

sion of this data set, and a consistently non-signifi-
Study results, calculated effect sizes, and CIs cant SES was found (SES: 1.09; 95% CI: –0.06 to
are summarised in Supplemental Table S3. 2.24). Gait velocity SES was not significant (SES:
Somatosensory outcomes, which included JPS, 0.42; 95% CI: –0.58 to 1.41; Figure 4).
light touch, and two-point discrimination, showed a High clinical heterogeneity and insufficient data
significant heterogeneous positive SES (SES: 0.52; prevented meaningful pooling of postural sway
95% CI: 0.04 to 1.01; Figure 2). However, sub- area, motor impairment, and leg function outcomes
group analyses of JPS (SES: 0.36; 95% CI: –0.25 to (Supplemental Table S3). Effect sizes of postural
0.96) and light touch (SES: 0.28; 95% CI: –0.86 to sway area, both eyes open and closed, were signifi-
1.41) were not significant. A significant heterogene- cant in one study,40 but not significant in the other.44
ous positive SES was found for BBS scores (SES: For results pertaining to motor impairment and leg
0.62; 95% CI: 0.10 to 1.14; Figure 3). SES of function, only the Barthel Index effect size was sig-
weight-bearing on the affected side was not signifi- nificant (SMD: 1.07; 95% CI: 0.24 to 1.89;
cant (SES: 1.52; 95% CI: –1.71 to 4.74). There was P = 0.01).41 Non-significant findings included the
an outlying data set in weight-bearing distribution FMA for lower extremity (P-values ranging from
on the affected side (Supplemental Table S3). A 0.13 to 0.61),38,46,51 Iowa Level of Assistance Scale
Chia et al. 973

Figure 3. Hedges’ g (95% CI) and summary effect size (95% CI) on Berg Balance Scale scores.
SES: summary effect size; CI: confidence interval.
The squares on the forest plot are of the same size, instead of proportional to study weight, as the forest plot was generated on
Microsoft Excel.

Figure 4. Hedges’ g (95% CI) and summary effect size (95% CI) on gait velocity.
SES: summary effect size; CI: confidence interval.
The squares on the forest plot are of the same size, instead of proportional to study weight, as the forest plot was generated on
Microsoft Excel.

(SMD: 0.00; 95% CI: –0.86 to 0.86; P = 1.00),22 discrimination,41,44 vibration,48 skin sensitivity,45 and
and TUG (SMD: 0.19; 95% CI: –0.46 to 0.85; DPT.22,23,41 Significant improvements were reported
P = 0.56).42 Details of data synthesis are available for hardness discrimination,44 configuration of soma-
from the corresponding author. tosensory evoked potentials,45 and in two out of three
Data from five studies were not included in data subjects for light touch.23 The TENS cross-over trial
synthesis due to having a cross-over design,49 data reported significant improvement post intervention
being unavailable,48 or having no control group.23,39,45 in ankle plantarflexion JPS and plantarflexor
Non-significant findings were reported for two-point strength, but not ankle dorsiflexion JPS and
974 Clinical Rehabilitation 33(6)

dorsiflexor strength.49 The assisted movement with significant improvements post intervention.
enhanced sensation robotic therapy trial39 reported Although unlikely, it is possible that this one study52
that 100% of subjects had 10% or more improve- may have skewed the results due to their mixed
ment in ankle JPS, 73% of subjects had 10% or more findings: a significant decrease in JPS error in the
improvement in ankle dorsiflexor strength, and 91% repeated passive movement group and an increase
of subjects had 10% or more improvement in ankle in JPS error in the repeated active movement group.
plantarflexor strength post intervention. Significant Studies that measured proprioception using the
improvements were reported in weight-bearing on DPT, unable to be included in the subgroup analy-
the affected side39 and Forward Reach Test49 post sis, reported no statistically significant improve-
intervention, and a downward trend over time in pos- ment,22,23 although one of them reported clinical
tural sway area.23 A significant improvement in gait improvement.23 This apparent lack of improvement
velocity was reported by two studies,39,49 but not sig- was attributed to the lack of sensitivity of DPT.22,23
nificant in one study.45 Light touch training effects appear ambiguous as
there were inconsistent findings among the three
included studies. One study reported between-
Adverse effects group difference in only one (first metatarsal) of
Four studies addressed adverse effects.39,46,47,49 Two seven points of the foot,22 one reported significant
studies reported no adverse effects.46,47 Reported improvement in two out of three subjects,23 and the
adverse effects were skin abrasion from self-over- third41 showed a non-significant effect size. Two-
treatment (one subject)39 and one subject reported a point discrimination similarly demonstrated no
day of pain post treatment.49 improvement associated with retraining of soma-
tosensory impairment,44 although this may also be
due to the lack of sensitivity of the instrument.41
Discussion However, the overall positive findings support the
This review aimed to examine the effects of inter- incorporation of interventions for addressing leg
ventions for leg somatosensory impairment after somatosensory impairment in stroke rehabilitation.
stroke primarily on somatosensory impairment and Specifically, the JPS modality may be a suitable
secondarily on balance, gait, motor impairment, starting point of retraining.
and leg function. Results of meta-analyses suggest Results of this review also suggest that interven-
that there is evidence that these interventions tions for post-stroke leg somatosensory impairment
improve somatosensory function and balance, but improve balance. Although pooling was not possi-
not gait, outcomes. However, it may be premature ble for postural sway area and a non-significant
to make firm conclusions about gait outcomes as finding was found for weight-bearing on the
pertinent variables other than gait velocity, such as affected side, three of seven studies that reported
gait symmetry, have not been assessed. The effects these outcomes showed significant positive effect
of these interventions on motor impairment and leg sizes.38,40,46 The remaining studies reported either
function remain unclear because pooling of data significant improvements post intervention39,42,44 or
was not possible due to a high degree of clinical a downward trend over time23 (see Supplemental
heterogeneity and insufficient data. Table S3). A potential reason for improvement in
The findings in this review suggest that interven- balance, as a result of addressing leg somatosensory
tions for post-stroke leg somatosensory impairment impairment, may be the improved perceptive ability
improve somatosensory function. Meta-analysis of through perceptual and motor learning, which is
somatosensory function was limited to the proprio- transferred to the motor performance of improved
ception (JPS error), light touch, and two-point dis- postural control.44 Improvement in balance may in
crimination modalities of the leg. Although JPS turn reduce falls risk in stroke survivors. This is
error subgroup analysis was not significant, all but especially important given the association between
one study52 included in the analysis reported somatosensory impairment and a higher falls inci-
Chia et al. 975

dence in stroke survivors compared to those with- be assessed when evaluating effectiveness of leg
out somatosensory impairment.3 somatosensory interventions in future trials.
There are a few possible reasons for the finding Two previous systematic reviews12,13 that inves-
that interventions for post-stroke leg somatosen- tigated the effects of interventions for somatosen-
sory impairment had no effect on gait outcomes. sory impairment in the stroke population, although
First, gait post stroke can be influenced not only by not specific to the leg, reported that there had been
somatosensory information,11,53 but also by other insufficient evidence to determine their effects and
factors including muscle strength,11,53–55 spastic- highlighted the need for high-quality controlled tri-
ity,55 cognition,56 visuospatial perception,57 motor als. Results from this review suggest that although
function,53,56 and balance.53,54 Changes to soma- several additional controlled trials examining the
tosensation alone may not be enough to influence effects of interventions for somatosensory impair-
gait. It may be necessary to retrain somatosensory ment, particularly in the leg, have been undertaken
function in conjunction with interventions that in recent years, the quality of these recent trials
address these other factors. Second, the interven- either remains poor or is difficult to assess due to
tions in most of the studies included in this review incomplete reporting. In view of this, future studies
may not have been applied in tasks specific to gait. should adhere to reporting guidelines for transpar-
There is strong evidence for effectiveness of task- ent reporting, such as CONSORT for randomised
specific training for recovery after stroke,58 as a trials65 and TREND for non-randomised trials.66
result of neuroplasticity.59 The use of intensive Contrary to these previous reviews, the medium
gait-specific training has been recommended for summary effects of this review provide preliminary
improving gait ability after stroke.60,61 An example evidence to support retraining of somatosensory
of gait-specific retraining of somatosensory func- function in the leg after stroke, for improving
tion may be ankle proprioceptive discrimination somatosensory function and balance.
throughout a gait cycle. Third, results may have Quantifiable and precise somatosensory assess-
been influenced by the small sample sizes and var- ment measures are vital in order to diagnose
ying methodology of the included studies. impairment, evaluate the extent of impairment and
Only gait velocity was measured in the three treatment effectiveness, and facilitate clinical
studies included in the meta-analysis. Another vari- decision-making about outcomes being achieved.67
able pertaining to gait that could be considered in The psychometric properties of many of the soma-
assessing treatment effectiveness is symmetry.7 tosensory outcome measures used in studies
Gait symmetry can be measured using step length, included in this review have not been established
or temporal measures such as stance or swing or were not reported in the studies, which is con-
times.62 It has been suggested that leg propriocep- sistent with an observation about the dearth of lit-
tive and tactile information provides critical feed- erature examining frequently-used somatosensory
back that is able to modify gait patterns,7 potentially assessment tools.68 This had implications for qual-
improving gait symmetry in stroke survivors. ity ratings of the included studies. Several studies
Improved somatosensory feedback can contribute used a range of measures; some such as the timed
to more accurate timing and amplitude of muscle 10-metre walk test69 and the BBS70 have demon-
contractions in response to the external environ- strated good psychometric properties,71,72 while
ment,63,64 thereby improving gait symmetry. It is not others such as the DPT73 have not had psychomet-
possible to make the conclusion that retraining of ric properties established. In order to provide a fair
leg somatosensory function would not affect gait at rating, five studies were given two different ratings:
all, as none of the studies statistically analysed gait a strong rating for balance or gait measures and a
symmetry. One study assessed stride length and weak rating for somatosensory measures.22,23,39,41,45
reported a significant within-group improvement,39 Furthermore, there are concerns raised about the
but data were not included in the meta-analysis due lack of standardisation, responsiveness, and general-
to a lack of a control group. Gait symmetry should isability of somatosensory measures used in stroke
976 Clinical Rehabilitation 33(6)

rehabilitation.68 For example, a study examining the been demonstrated to be associated with perception
psychometric properties of the sensory subscale of of verticality,82 which in turn is related to balance.83
the Fugl-Meyer Assessment (FMA) found that the It is possible that enhanced proprioceptive and tac-
high ceiling effect, and poor to moderate reliability, tile feedback contributes to a more accurate percep-
validity, and responsiveness, did not support its tion of verticality, thereby positively influencing
clinical use in stroke rehabilitation.74 The FMA was balance. Furthermore, increased weight-bearing on
used in three of the included studies in this the affected side has been found to be associated
review.38,46,51 On a positive note, several leg pro- with a reduction in postural sway.84 It was suggested
prioception assessment tools included in this review that a reduction in postural sway could be due to
have been tested for their psychometrics, including enhancement of somatosensory information enabled
electrogoniometers75 and digital inclinometers.76,77 by increased weight-bearing on that leg.84 Based on
However, these tools have not been tested in the this review’s meta-analysis, interventions aimed to
stroke population, and their usage in clinical set- improve somatosensory function did not appear to
tings remains limited. Future research should focus increase weight-bearing on the affected side.
on establishing the psychometric properties of these However, in two studies38,46 where interventions
tools in stroke rehabilitation. Development of leg specifically targeted weight-bearing on the affected
somatosensory measures that are quantifiable, sen- side, there were larger improvements of weight-
sitive to change, and available for clinical use may bearing on the affected side in the experimental
also be required for assessing treatment effective- group compared to the control group. Further
ness and enabling better quality trials in somatosen- research into interventions aimed to increase weight-
sory rehabilitation. bearing on the affected side may be useful in clarify-
There was a diverse range of interventions used ing its role in reduction of postural sway.
in the studies included. This highlights a need for There are obvious strengths and limitations in
developing standardised retraining methods of leg this review. The main strength is the use of the
somatosensory function that can be reliably repli- PRISMA guidelines,17 which enable transparent
cated across trials and in clinical settings, to and complete reporting. One of the limitations is
increase consistency of interventions. An approach that the inclusion of non-randomised as well as
for retraining of somatosensory function has been randomised trials of varying quality permitted
developed,78 which has been demonstrated to be inclusion of a high risk of bias across studies.
effective albeit only in arm studies.79–81 The However, the reviewers wanted to report all avail-
approach is derived from theories of perceptual able studies on post-stroke interventions that aimed
learning and somatosensory processing neurophys- to improve leg somatosensory function. All the
iology, and are consistent with the learning- included studies had small sample sizes with the
dependent principles of neuroplasticity.78 The key highest being 62 participants.40 These small sample
elements of the approach are task-based (goal- sizes, plus the high risk of selection bias within
directed), guided attentive exploration with vision studies as noted in quality appraisal, mean that out-
occluded, immediate and precise feedback, calibra- comes from these studies may not be representative
tion (within and across different sensory modali- of the wider stroke population. In addition, inter-
ties, e.g. other limb and vision), anticipation trials, pretation of the results may have been influenced
repetition, graded progression, and transfer of by the heterogeneity of interventions and outcomes
training effects to novel stimuli. Further explora- included. A risk of publication bias also exists due
tion regarding the application and effectiveness of to the exclusion of non-English publications and
these principles, particularly in the leg, may be unpublished studies.24 Nonetheless, the inclusion
beneficial in establishing a standardised approach of non-English studies of randomised trials in sys-
to addressing somatosensory impairment. tematic reviews found that language restrictions
Somatosensory information, both from the joint did not appear to bias results of conventional
(proprioception) and from the skin (tactile), has interventions.85
Chia et al. 977

individual patient data. Top Stroke Rehabil 2013; 20(5):

441–449.
Clinical messages 4. Carey LM, Matyas TA and Baum C. Effects of somatosen-
•• Interventions aimed at retraining leg sory impairment on participation after stroke. Am J Occup
Ther 2018; 72(3): 7203205100p1–7203205100p10.
somatosensory function post stroke are 5. Rand D, Gottlieb D and Weiss P. Recovery of patients
shown to be effective for improving with a combined motor and proprioception deficit dur-
somatosensory impairment and balance, ing the first six weeks of post stroke rehabilitation. Phys
but not gait. Occup Ther Geriatr 2001; 18: 69–87.
•• Many of the somatosensory assessment 6. Parsons SL, Mansfield A, Inness EL, et al. The relation-
ship of plantar cutaneous sensation and standing balance
tools used in the leg have not been tested post-stroke. Top Stroke Rehabil 2016; 23(5): 326–332.
for their validity and reliability in stroke 7. Wutzke CJ, Mercer VS and Lewek MD. Influence of
rehabilitation. lower extremity sensory function on locomotor adapta-
•• There is a varied range of intervention tion following stroke: a review. Top Stroke Rehabil 2013;
methods intended for retraining leg 20(3): 233–240.
8. Duysens J and Massaad F. Stroke gait rehabilitation: is
somatosensory function after stroke. load perception a first step towards load control. Clin
Neurophysiol 2015; 126(2): 225–226.
9. Lin S-I. Motor function and joint position sense in relation
Acknowledgements to gait performance in chronic stroke patients. Arch Phys
Med Rehabil 2005; 86(2): 197–203.
We thank the authors of the included studies who 10. Lee M-J, Kilbreath SL and Refshauge KM. Movement
responded to our requests for information. detection at the ankle following stroke is poor. Aust J
Physiother 2005; 51(1): 19–24.
Declaration of conflicting interests 11. Hsu A-L, Tang P-F and Jan MH. Analysis of impairments
influencing gait velocity and asymmetry of hemiplegic
The author(s) declared no potential conflicts of interest patients after mild to moderate stroke. Arch Phys Med
with respect to the research, authorship, and/or publication Rehabil 2003; 84(8): 1185–1193.
of this article. 12. Schabrun SM and Hillier S. Evidence for the retraining of
sensation after stroke: a systematic review. Clin Rehabil
Funding 2009; 23: 27–39.
13. Doyle S, Bennett S, Fasoli SE, et al. Interventions for sen-
The author(s) received no financial support for the sory impairment in the upper limb after stroke. Cochrane
research, authorship, and/or publication of this article. Database Syst Rev 2010; 6: CD006331.
14. Aman JE, Elangovan N, Yeh IL, et al. The effectiveness
ORCID iD of proprioceptive training for improving motor function: a
systematic review. Front Hum Neurosci 2014; 8: 1075.
Fenny SF Chia https://orcid.org/0000-0002-9243-
15. McCluskey A, Logan P, Carey L, et al. Critically
8776
appraised papers. Evidence for the retraining of sensation
after stroke remains limited. Aust Occup Ther J 2010; 57:
Supplemental material 200–202.
Supplemental material for this article is available online. 16. National Institute for Health Research. PROSPERO:
international Prospective Register of Systematic Reviews,
https://www.crd.york.ac.uk/PROSPERO/ (2017, accessed
References 10 February 2017).
1. Connell LA, Lincoln NB and Radford KA. 17. Moher D, Liberati A, Tetzlaff J, et al. Preferred report-
Somatosensory impairment after stroke: frequency of ing items for systematic reviews and meta-analyses: the
different deficits and their recovery. Clin Rehabil 2008; PRISMA statement. PLoS Med 2009; 6: e1000097.
22(8): 758–767. 18. Carey LM. Touch and body sensations. In: Carey LM
2. Tyson SF, Hanley M, Chillala J, et al. Sensory loss in hos- (ed.) Stroke rehabilitation: insights from neuroscience
pital-admitted people with stroke: characteristics, associ- and imaging. New York: Oxford University Press, 2012,
ated factors, and relationship with function. Neurorehabil pp.157–172.
Neural Repair 2008; 22(2): 166–172. 19. Borstad AL and Nichols-Larsen DS. Assessing and treat-
3. Tyson SF, Crow JL, Connell L, et al. Sensory impair- ing higher level somatosensory impairments post stroke.
ments of the lower limb after stroke: a pooled analysis of Top Stroke Rehabil 2014; 21(4): 290–295.
978 Clinical Rehabilitation 33(6)

20. Carey LM. Somatosensory loss after stroke. Crit Rev Phys step-by-step guide focusing on descriptive data analysis.
Rehabil Med 1995; 7: 51–91. BMC Res Notes 2012; 5: 52.
21. Kim JS and Choi-Kwon S. Discriminative sensory dysfunc- 37. Cross Validated. Combining outcome measures for meta-
tion after unilateral stroke. Stroke 1996; 27(4): 677–682. analysis, https://stats.stackexchange.com/questions/92134
22. Lynch EA, Hillier SL, Stiller K, et al. Sensory retraining of /combining-outcome-measures-for-meta-analysis (2014,
the lower limb after acute stroke: a randomized controlled accessed 10 January 2019).
pilot trial. Arch Phys Med Rehabil 2007; 88(9): 1101–1107. 38. Mohapatra S, Eviota AC, Ringquist KL, et al. Compelled
23. Hillier S and Dunsford A. A pilot study of sensory retrain- body weight shift technique to facilitate rehabilitation of
ing for the hemiparetic foot post-stroke. Int J Rehabil Res individuals with acute stroke. ISRN Rehabil 2012; 2012:
2006; 29(3): 237–242. 328018.
24. Centre for Reviews and Dissemination. Systematic reviews: 39. Cordo P, Lutsep H, Cordo L, et al. Assisted movement
CRD’s guidance for undertaking reviews in health care. with enhanced sensation (AMES): coupling motor and
York: CRD University of York, 2009. sensory to remediate motor deficits in chronic stroke
25. Feng Y, Schlosser FJ and Sumpio BE. The Semmes patients. Neurorehabil Neural Repair 2009; 23(1): 67–77.
Weinstein monofilament examination as a screening tool 40. Han SK, Kim MC and An CS. Comparison of effects of a
for diabetic peripheral neuropathy. J Vasc Surg 2009; proprioceptive exercise program in water and on land the
50(3): 675–682.e1. balance of chronic stroke patients. J Phys Ther Sci 2013;
26. University of Nottingham. Nottingham sensory assess- 25(10): 1219–1222.
ment. https://www.nottingham.ac.uk/medicine/documents 41. Dogru Huzmeli E, Yildirim SA and Kilinc M. Effect of
/publishedassessments/nsainstructionsrevised.pdf (2007, sensory training of the posterior thigh on trunk control and
accessed 15 June 2017). upper extremity functions in stroke patients. Neurol Sci
27. Fugl-Meyer AR, Jaasko L, Leyman I, et al. The post-stroke 2017; 38(4): 651–657.
hemiplegic patient. A method for evaluation of physical 42. Lee H, Kim H, Ahn M, et al. Effects of proprioception
performance. Scand J Rehabil Med 1975; 7: 13–31. training with exercise imagery on balance ability of stroke
28. Effective public health practice project. Quality assess- patients. J Phys Ther Sci 2015; 27(1): 1–4.
ment tool for quantitative studies, https://merst.ca/wp- 43. Moon S-J and Kim YW. Effect of blocked vision tread-
content/uploads/2018/02/quality-assessment-tool_2010. mill training on knee joint proprioception of patients with
pdf (2009, accessed 9 March 2017). chronic stroke. J Phys Ther Sci 2015; 27(3): 897–900.
29. Thomas BH, Ciliska D, Dobbins M, et al. A process for 44. Morioka S and Yagi F. Effects of perceptual learning exer-
systematically reviewing the literature: providing the cises on standing balance using a hardness discrimination
research evidence for public health nursing interventions. task in hemiplegic patients following stroke: a randomized
Worldviews Evid Based Nurs 2004; 1(3): 176–184. controlled pilot trial. Clin Rehabil 2003; 17(6): 600–607.
30. Effective public health practice project. Quality assess- 45. Peurala SH, Pitkanen K, Sivenius J, et al. Cutaneous elec-
ment tool for quantitative studies dictionary, https://merst trical stimulation may enhance sensorimotor recovery in
.ca/wp-content/uploads/2018/02/qualilty-assessment-dic chronic stroke. Clin Rehabil 2002; 16(7): 709–716.
tionary_2017.pdf (1998, accessed 9 March 2017). 46. Aruin AS, Rao N, Sharma A, et al. Compelled body
31. Centre for Evaluation Monitoring. Effect size calcula- weight shift approach in rehabilitation of individuals with
tor, http://www.cem.org/effect-size-calculator (2017, chronic stroke. Top Stroke Rehabil 2012; 19(6): 556–563.
accessed 18 July 2017). 47. Mazuchi FDAES, Bigongiari A, Francica JV, et al.
32. Borenstein M, Hedges LV, Higgins JPT, et al. Introduction Aerobic training in aquatic environment improves the
to meta-analysis. Hoboken, NJ: John Wiley & Sons, 2009, position sense of stroke patients: a randomized clinical
pp.21–32. trial. Motriz: J Phys Ed 2018; 24: 1–7.
33. Higgins JPT and Deeks JJ. Selecting studies and collect- 48. Zankel HT. Pallesthesia studies in stroke patients. South
ing data. In: Higgins J and Green S (eds) Cochrane hand- Med J 1969; 62: 8–11.
book for systematic reviews of interventions. Hoboken, 49. Tyson SF, Sadeghi-Demneh E and Nester CJ. The effects
NJ: John Wiley & Sons, 2008, pp.151–185. of transcutaneous electrical nerve stimulation on strength,
34. Hedges LV. Distribution theory for Glass’s estimator of proprioception, balance and mobility in people with
effect size and related estimators. J Educ Behav Stat 1981; stroke: a randomized controlled cross-over trial. Clin
6: 107–128. Rehabil 2013; 27(9): 785–791.
35. Deeks JJ, Higgins JPT and Altman DG. Analysing data 50. Biodex. Dynamometers, http://www.biodex.com/physi
and undertaking meta-analyses. In: Higgins J and Green cal-medicine/products/dynamometers (2019, accessed 9
S (eds) Cochrane handbook for systematic reviews of January 2019).
interventions. Hoboken, NJ: John Wiley & Sons, 2008, 51. Da Silva Ribeiro NM, Ferraz DD, Pedreira É, et al.
pp.243–296. Virtual rehabilitation via Nintendo Wii® and conventional
36. Neyeloff JL, Fuchs SC and Moreira LB. Meta-analyses physical therapy effectively treat post-stroke hemiparetic
and forest plots using a Microsoft Excel spreadsheet: patients. Top Stroke Rehabil 2015; 22: 299–305.
Chia et al. 979

52. Kwon OS and Lee SW. Effect of continuing repeated 70. Berg KO, Wood-Dauphinee SL, Williams JI, et al.
passive and active exercises on knee’s position senses Measuring balance in the elderly: validation of an instru-
in patients with hemiplegia. Neurorehabilitation 2013; ment. Can J Public Health 1992; 83: S7–S11.
33(3): 391–397. 71. Tyson SF and Connell L. The psychometric properties
53. Nadeau S, Arsenault AB, Gravel D, et al. Analysis of and clinical utility of measures of walking and mobil-
the clinical factors determining natural and maximal gait ity in neurological conditions: a systematic review. Clin
speeds in adults with a stroke. Am J Phys Med Rehabil Rehabil 2009; 23(11): 1018–1033.
1999; 78(2): 123–130. 72. Berg K, Wood-Dauphinee S and Williams JI. The Balance
54. Patterson SL, Forrester LW, Rodgers MM, et al. Determinants Scale: reliability assessment with elderly residents and
of walking function after stroke: differences by deficit sever- patients with an acute stroke. Scand J Rehabil Med 1995;
ity. Arch Phys Med Rehabil 2007; 88(1): 115–119. 27(1): 27–36.
55. Lin P-Y, Yang Y-R, Cheng S-J, et al. The relation between 73. Bohannon RW. Evaluation and treatment of sensory and
ankle impairments and gait velocity and symmetry in perceptual impairments following stroke. Top Geriatr
people with stroke. Arch Phys Med Rehabil 2006; 87(4): Rehabil 2003; 19: 87–97.
562–568. 74. Lin J-H, Hsueh I-P, Sheu C-F, et al. Psychometric proper-
56. Cho KH, Lee JY, Lee KJ, et al. Factors related to gait ties of the sensory scale of the Fugl-Meyer Assessment in
function in post-stroke patients. J Phys Ther Sci 2014; stroke patients. Clin Rehabil 2004; 18(4): 391–397.
26(12): 1941–1944. 75. Smith TO, Davies L and Hing CB. A systematic review
57. Roh H. Effect of visual perceptual disturbance on gait and to determine the reliability of knee joint position sense
balance. J Phys Ther Sci 2015; 27(10): 3109–3111. assessment measures. Knee 2013; 20(3): 162–169.
58. Veerbeek JM, van Wegen E, van Peppen R, et al. What is 76. Romero-Franco N, Montano-Munuera JA and Jimenez-
the evidence for physical therapy poststroke? A systematic Reyes P. Validity and reliability of a digital inclinometer
review and meta-analysis. PLoS ONE 2014; 9(2): e87987. to assess knee joint position sense in a closed kinetic
59. Richards LG, Stewart KC, Woodbury ML, et al. chain. J Sport Rehabil 2017; 26(1): 1–5.
Movement-dependent stroke recovery: a systematic 77. Suner-Keklik S, Cobanoglu-Seven G, Kafa N, et al. The
review and meta-analysis of TMS and fMRI evidence. validity and reliability of knee proprioception measure-
Neuropsychologia 2008; 46(1): 3–11. ment performed with inclinometer in different positions.
60. Eng JJ and Tang PF. Gait training strategies to optimize J Sport Rehabil 2017; 26(6): 1–15.
walking ability in people with stroke: a synthesis of the 78. Carey LM. Loss of somatic sensation. In: Selzer M, Clarke
evidence. Expert Rev Neurother 2007; 7(10): 1417–1436. S, Cohen L and et al (eds) Textbook of neural repair and
61. Pascal M, Barnable E, Bisset R, et al. Task-specific gait rehabilitation. Cambridge: Cambridge University Press,
training for individuals with chronic CVA: a systematic 2006, pp.231–247.
review. Arch Phys Med Rehabil 2017; 98: e142. 79. Carey L, Macdonell R and Matyas TA. SENSe: Study of
62. Patterson KK. Gait asymmetry post-stroke. PhD Thesis, the Effectiveness of Neurorehabilitation on Sensation: a
University of Toronto, Toronto, ON, Canada, 2010. randomized controlled trial. Neurorehabil Neural Repair
63. Dietz V, Muller R and Colombo G. Locomotor activity in 2011; 25(4): 304–313.
spinal man: significance of afferent input from joint and 80. Carey LM and Matyas TA. Training of somatosensory
load receptors. Brain 2002; 125(Pt 12): 2626–2634. discrimination after stroke: facilitation of stimulus gener-
64. Rossignol S, Dubuc R and Gossard JP. Dynamic senso- alization. Am J Phys Med Rehabil 2005; 84(6): 428–442.
rimotor interactions in locomotion. Physiol Rev 2006; 81. Turville M, Carey LM, Matyas TA, et al. Change in func-
86(1): 89–154. tional arm use is associated with somatosensory skills
65. Schulz KF, Altman DG, Moher D, et al. CONSORT 2010 after sensory retraining poststroke. Am J Occup Ther
Statement: updated guidelines for reporting parallel group 2017; 71(3): 7103190070p1–7103190070p9.
randomised trials. Br Med J 2010; 340: 697–702. 82. Saeys W, Vereeck L, Truijen S, et al. Influence of sen-
66. Des Jarlais DC, Lyles C and Crepaz N. Improving the sory loss on the perception of verticality in stroke patients.
reporting quality of nonrandomized evaluations of behav- Disabil Rehabil 2012; 34(23): 1965–1970.
ioral and public health interventions: the TREND state- 83. Bonan IV, Guettard E, Leman MC, et al. Subjective visual
ment. Am J Public Health 2004; 94(3): 361–366. vertical perception relates to balance in acute stroke. Arch
67. Sullivan EJ, Andrews WA, Lanzino AD, et al. Outcome Phys Med Rehabil 2006; 87(5): 642–646.
measures in neurological physical therapy practice: part 84. Ju SK and Yoo WG. The effect of somatosensory and
II. J Neurol Phys Ther 2011; 35(2): 65–74. cognitive-motor tasks on the paretic leg of chronic stroke
68. Sullivan JE and Hedman LD. Sensory dysfunction fol- patients in the standing posture. J Phys Ther Sci 2014;
lowing stroke: incidence, significance, examination, and 26(12): 1869–1870.
intervention. Top Stroke Rehabil 2008; 15(3): 200–217. 85. Moher D, Pham B, Lawson ML, et al. The inclusion of
69. Wade DT, Wood VA, Heller A, et al. Walking after reports of randomised trials published in languages other
stroke. Measurement and recovery over the first 3 months. than English in systematic reviews. Health Technol Assess
Scand J Rehabil Med 1987; 19: 25–30. 2003; 7(41): 1–90.