1097477 WSO International Journal of Stroke X(X)Brady

Systematic Review

International Journal of Stroke

Precision rehabilitation for aphasia

2022, Vol. 17(10) 1067­–1077
https://doi.org/10.1177/17474930221097477
© 2022 World Stroke Organization

by patient age, sex, aphasia severity, Article reuse guidelines:

and time since stroke? A prespecified, sagepub.com/journals-permissions

DOI: 10.1177/17474930221097477
journals.sagepub.com/home/wso
systematic review-based, individual
participant data, network, subgroup
meta-analysis

The RELEASE Collaborators

Abstract
Background: Stroke rehabilitation interventions are routinely personalized to address individuals’ needs, goals, and
challenges based on evidence from aggregated randomized controlled trials (RCT) data and meta-syntheses. Individual
participant data (IPD) meta-analyses may better inform the development of precision rehabilitation approaches, quantify-
ing treatment responses while adjusting for confounders and reducing ecological bias.
Aim: We explored associations between speech and language therapy (SLT) interventions frequency (days/week), inten-
sity (h/week), and dosage (total SLT-hours) and language outcomes for different age, sex, aphasia severity, and chronicity
subgroups by undertaking prespecified subgroup network meta-analyses of the RELEASE database.
Methods: MEDLINE, EMBASE, and trial registrations were systematically searched (inception-Sept2015) for RCTs,
including ⩾ 10 IPD on stroke-related aphasia. We extracted demographic, stroke, aphasia, SLT, and risk of bias data.
Overall-language ability, auditory comprehension, and functional communication outcomes were standardized. A one-
stage, random effects, network meta-analysis approach filtered IPD into a single optimal model, examining SLT regimen
and language recovery from baseline to first post-intervention follow-up, adjusting for covariates identified a-priori.
Data were dichotomized by age (⩽/> 65 years), aphasia severity (mild–moderate/ moderate–severe based on language
outcomes’ median value), chronicity (⩽/> 3 months), and sex subgroups. We reported estimates of means and 95%
confidence intervals. Where relative variance was high (> 50%), results were reported for completeness.
Results: 959 IPD (25 RCTs) were analyzed. For working-age participants, greatest language gains from baseline occurred
alongside moderate to high-intensity SLT (functional communication 3-to-4 h/week; overall-language and comprehen-
sion > 9 h/week); older participants’ greatest gains occurred alongside low-intensity SLT (⩽ 2 h/week) except for audi-
tory comprehension (> 9 h/week). For both age-groups, SLT-frequency and dosage associated with best language gains
were similar. Participants ⩽ 3 months post-onset demonstrated greatest overall-language gains for SLT at low intensity/
moderate dosage (⩽ 2 SLT-h/week; 20-to-50 h); for those > 3 months, post-stroke greatest gains were associated with
moderate-intensity/high-dosage SLT (3–4 SLT-h/week; ⩾ 50 hours). For moderate–severe participants, 4 SLT-days/week
conferred the greatest language gains across outcomes, with auditory comprehension gains only observed for ⩾ 4 SLT-
days/week; mild–moderate participants’ greatest functional communication gains were associated with similar frequency
(⩾ 4 SLT-days/week) and greatest overall-language gains with higher frequency SLT (⩾ 6 days/weekly). Males’ greatest
gains were associated with SLT of moderate (functional communication; 3-to-4 h/weekly) or high intensity (overall-
language and auditory comprehension; (> 9 h/weekly) compared to females for whom the greatest gains were associated
with lower-intensity SLT (< 2 SLT-h/weekly). Consistencies across subgroups were also evident; greatest overall-­
language gains were associated with 20-to-50 SLT-h in total; auditory comprehension gains were generally observed
when SLT > 9 h over ⩾ 4 days/week.

Corresponding author:
Marian C Brady, NMAHP Research Unit, Glasgow Caledonian University, Cowcaddens Road, Glasgow G4 0BA, UK.
Email: [email protected]

International Journal of Stroke, 17(10)

1068 International Journal of Stroke 17(10)

Conclusions: We observed a treatment response in most subgroups’ overall-language, auditory comprehension, and
functional communication language gains. For some, the maximum treatment response varied in association with dif-
ferent SLT-frequency, intensity, and dosage. Where differences were observed, working-aged, chronic, mild–moderate,
and male subgroups experienced their greatest language gains alongside high-frequency/intensity SLT. In contrast, older,
moderate–severely impaired, and female subgroups within 3 months of aphasia onset made their greatest gains for
lower-intensity SLT. The acceptability, clinical, and cost effectiveness of precision aphasia rehabilitation approaches
based on age, sex, aphasia severity, and chronicity should be evaluated in future clinical RCTs.

Keywords
Stroke, aphasia, rehabilitation, speech and language therapy, individual participant data, network meta-analysis

Received: 29 October 2021; accepted: 1 March 2022

Introduction language assessments into a standardized mean differ-

ence11,12, making clinical interpretation and application of
Personalized health care is a central tenet of stroke rehabili- emerging evidence a challenge.9,13 In addition, meta-synthe-
tation; therapists routinely tailor interventions to patients’ ses of aggregate data risk masking the responsiveness of dif-
individual goals, preferences, the optimal difficulty level, ferent subgroups to specific interventions, known as
the local environment, risks, their health, and cognitive sta- ecological bias.14 Aphasia meta-syntheses have presented
tus and limitations.1 Precision medicine refers to data- findings of interest (and concern); for example, one meta-
driven decision making based on patient subpopulations analysis highlighted the benefits of intensive SLT early
and responsiveness to specific interventions.2 Data-driven post-onset though these findings were confounded by higher
“precision” rehabilitation intervention decisions, with gains participant dropouts (n = 35/114) compared to lower-inten-
in acceptability, clinical, and cost effectiveness, require sity SLT participants (n = 17/102; p = 0.01).5 In contrast,
exploration of subgroups’ responsiveness to interventions among the RCTs which recruited participants years post-
based on, for example, clinical demographics, stroke sever- onset, neither dropouts nor significant language gains were
ity, and time post-onset.3 evident.5 Further examination of the potential benefits,
risks, and evident variability in tolerance and responsive-
Effectiveness of rehabilitation interventions ness to the different SLT regimens may be worthwhile but
could not be advanced until individual participants’ data
Examining the effectiveness of complex rehabilitation inter- (IPD) on demographics were available.
ventions is challenging.4 Aphasia interventions, for exam-
ple, vary by regime (frequency, intensity, duration, and
dosage), provider, delivery model, and therapeutic approach. Individual participant data meta-analyses
Therapists’ clinical decision making is typically informed Where strong theoretical reasons and clinical plausibility
by findings from high-quality, group-level comparisons suggest differential responses to an intervention (e.g., apha-
within randomized controlled trials (RCTs) and estimates of sia rehabilitation), large IPD meta-analysis is the gold stand-
the average effect based on meta-synthesis of aggregate ard.14 Our earlier IPD meta-analysis found that trial
group data, which in turn inform clinical guidelines.5,6 High- participants’ greatest overall and auditory comprehension
level aphasia rehabilitation evidence has, to date, been pri- language gains were associated with > 20–50 h SLT dosage,
marily based on aggregate data from high-quality RCTs and delivered 2-to-4 h each week, and between 3 and 5 + days.
meta-synthesis,5,7 thus limiting the degree to which it might Greatest auditory comprehension gains were associated
inform precision rehabilitation decisions.8 with 9 + SLT-h weekly over 4-to-5 days.15 Meta-analyses
using IPD also support the exploration of differential treat-
ment response across participants,8,16 reducing the risk of
Aggregate data limitations ecological bias, facilitating the inclusion of previously unre-
Aggregate data based on relatively large samples and ade- ported dropouts, outcome measurement, and follow-up data
quate statistical power provide crucial evidence of the effec- while increasing statistical power.9,10,17 In a highly heteroge-
tiveness of rehabilitation interventions across recruited neous population such as people with aphasia, investiga-
populations. The ability of an isolated trial to determine sub- tions of subgroups’ (based on age, aphasia severity,
groups’ differential treatment response, however, is lim- chronicity, or sex) treatment responsiveness might be
ited.8,9 Limitations to aggregate data meta-syntheses include explored with greater statistical power and adjustments for
the risk of outcome reporting bias,10 restricted inclusion of confounders than at trial level.9,14,18 Information on differen-
language data, and the need to synthesize various clinical tial responsiveness of clinically relevant subgroups may

International Journal of Stroke, 17(10)

Brady 1069

inform the development of precision aphasia rehabilitation ability, auditory comprehension, and functional communi-
approaches and future RCT-based treatment evaluations cation). Language recovery was defined as the change in
absolute language score from baseline to first post-inter-
vention follow-up. We checked and collated language
Aims domain measurements, as agreed a priori by the RELEASE
Following a large IPD network meta-analysis, our pre- collaborators. Whenever possible, baseline and subsequent
15

specified subgroup analysis aimed to explore language timepoint data extraction was confirmed and additional
21
change from baseline to first post-intervention follow-up unreported data sought from the primary researchers.
(measuring overall-language, auditory comprehension, and For each outcome, the most frequently used measure by
functional communication) for various levels of rehabilita- dataset was identified as the anchor measure. All remaining
tion, intensity, dosage in aphasia subgroups, varying by measures of that outcome (minority measures) were trans-
participants’ age, sex, aphasia chronicity and severity at formed to match the anchor measure’s range and format,
baseline. thus retaining the clinical relevance of the anchor meas-
ures’ change score.19 Anchor measures comprised the
Western aphasia battery-aphasia quotient (WAB-AQ)22 for
Methods overall-language ability; the Aachen Aphasia test-token test
23
Approval was granted for this database study (UK IRAS (AAT-TT) scored positively for auditory comprehension;
registration ID 179505; Glasgow Caledonian University and the Aachen Aphasia test-spontaneous speech commu-
Health and Life Sciences Ethical Committee HLS/ nication (AAT-SSC) 23
rating domain score for functional
NCH/15/09). communication.
SLT interventions targeting language recovery were cat-
egorized by regimen (frequency, intensity, and dosage;
Search strategy and selection criteria Supplemental Material A). All data were checked by an
We created the international RELEASE IPD database of independent researcher and, where possible, with available
aphasia research to support several planned analyses,15,19,20 documentation and primary researchers. We recorded una-
systematically identifying and reviewing published and vailable data as “unreported.” Aphasia with a non-stroke
unpublished datasets with ⩾ 10 IPD on aphasia, language etiology, unreported time post-onset, and duplicate IPD
outcome, and time since stroke (reported in full else- were excluded. In the absence of SLT intervention records at
where).15,19,21 Briefly, several electronic databases, including IPD level, we applied group-level SLT descriptions to the
MEDLINE and EMBASE, were searched (inception-­ IPD accordingly. Final data formatting decisions were made
September 2015 for eligible datasets). Anticipating lengthy through collaborator discussion. Categorical data formats
data cleaning and analysis procedures, we also searched trial (e.g., 5-to-6 weeks) were recorded as means (5.5 weeks).
registrations for eligible trials completing beyond the elec- Language co-interventions (e.g., pharmacological) were
tronic search date.15,21 Non-English datasets were translated. documented and IPD were included up to the point of any
Language data derived from stroke or screening measures crossover.
were excluded. Non-randomized trials, case series, and clini-
cal registries were included in the database and supported
IPD network meta-analysis
previous analyses20 but were excluded from this subgroup
analysis. Drawing only RCT IPD from the wider RELEASE database
Full text reports were reviewed independently by two (21), we conducted a network meta-analysis of SLT interven-
reviewers; a third resolved disagreements. Potentially eligi- tions delivered by language outcomes (15). Where tradi-
ble datasets were invited to contribute IPD. One reminder tional meta-analyses consider pair-wise trial comparisons
was sent to non-respondents, followed by attempts to con- (e.g., treatment 1 vs 2), network meta-analysis considers
tact co-authors. We confirmed dataset eligibility with three or more interventions simultaneously by making direct
respondents prior to contribution. A protocol guided data (treatment 1 vs 2; treatment 2 vs 3) and indirect comparisons
searching, identification, extraction, and analyses (treatment 1 vs 3), thus yielding more precise estimates than
(PROSPERO CRD42018110947).21 Included datasets paired direct/indirect estimates and making it possible to
reported relevant ethical and gatekeeper approvals. compare the effectiveness of interventions. We used datasets
as random effects and demographics and interventions as
fixed effects. Data analysis for this article used SAS™ soft-
Data extraction and preparation ware (9.4 using PROCMIXED). Using a statistical inferenc-
For this prespecified subgroup analysis, we extracted data ing approach, we sought to highlight important research
on demography (including sex, age, and language used), questions, considerations for future trial design, and clinical
stroke (time post-onset, hemisphere, and aphasia severity), implementation. Language recovery was defined as the mean
SLT intervention, and language outcome (overall-language of the absolute change from baseline to the first follow-up on

International Journal of Stroke, 17(10)

1070 International Journal of Stroke 17(10)

the transformed standardized measure. Effect sizes were esti- SLT-days/week (10 IPD; followed by 4 SLT-days/week; 28
mated and reported (95% CI). Our minimum sample size for IPD). Auditory comprehension gains were absent when
each analysis was 20 IPD (2 RCTs). SLT < 4 days/week. The greatest gains for both age-groups
We included prespecified potential confounders in the were observed at 4 SLT-days/week, the only significant gain
base model (age, sex, aphasia severity, and time post-onset) for older participants; working-aged participants made simi-
and simultaneously examined the impact of IPD and lan- lar gains at 5 days/week. Working-age participants’ greatest
guage variables on the intervention effect. Our one-stage functional communication gains occurred for ⩾ 5 SLT-days/
network meta-analysis examined IPD and SLT intervention week and were observed for older participants for 4 SLT-
regimen variations by age, time since onset, aphasia sever- days/week (with similar gains at 5 SLT-days/week; Table 1;
ity at baseline and sex subgroups, and associated estimates Supplemental Material H(a) to (c)).
of mean language gains from baseline (Supplemental
Material A). Continuous regimen variables were grouped Intensity.  Working-age participants made their greatest
for comparison (e.g., 3 vs 4 SLT-days/week). We dichoto- overall-language and auditory comprehension gains along-
mized IPD based on key demographic and clinical data; side > 9 SLT-h/week. Older participants’ greatest overall-
males and females, median age (working age ⩽ 65 years language gains occurred for < 2 SLT-h/week (similar for
and older > 65 years; rehabilitation timing after aphasia 3–4 SLT-h/week) while their only comprehension gain
onset (early ⩽ 3 months and late > 3 months); moderate– occurred when SLT > 9 h/week. Functional communica-
severe and mild–moderate groups based on the overall tion gains were greatest for working-age participants’ when
median language modality score. SLT 3–4 h/week and ⩽ 2 h/week for the older (Table 1;
We considered clustering by dataset distinguishing IPD Supplemental Material H(d) to (f)).
from dataset-based interactions.15,21 Where > 20% of a
dataset variable was missing, we excluded it from that net- Dosage.  Both age-groups’ greatest overall-language gains
work analysis. Patterns of loss were checked; we compared occurred alongside 20-to-50 SLT-h (few IPD; see Table 1)
missing data to demographic and other variables using the while other gains, based on greater IPD, were similar across
t-test or Mann–Whitney U test. In the absence of evidence dosages. Working-age participants’ auditory comprehension
of influence, we studied the data missing at random. We gains were greatest for 20–50 SLT-h (with similar gains for
excluded any data not missing at random. For risk of bias 14–20 h); older participants’ significant gains were only
and heterogeneity checking, see Supplemental Material B. observed for > 20 SLT-h. Older participants made greatest
functional communication gains alongside 20–50 SLT-h; for
working-age participants gains observed for 14-to-20 SLT-h
Results were based on 6 IPD, followed by > 50 SLT-h (87 IPD)
Filtering by available demographic, language, intervention (Table 1; Supplemental Material H(g) to (i)).
data, and time points, 959 IPD (25 RCTs; Supplemental
Material C to F) informed a prespecified IPD subgroup net- Early versus late rehabilitation
work meta-analysis of therapy regimen and language out-
comes: overall-language ability (WAB-AQ 482 IPD; 11 Frequency.  The greatest overall-language gains were
RCTs); functional communication (AAT-SSC 533 IPD; 14 observed for three participants that received early SLT
RCTs); and auditory comprehension (AAT-TT 550 IPD; 16 3 days/week, but the greatest gains for most early-rehabilita-
RCTs). Participants experienced predominantly left hemi- tion participants (150 IPD) were noted for 5 SLT-days/
sphere (683 IPD; 97.7%); ischemic first strokes (685 IPD; week. Significant auditory comprehension gains were not
88.9%) with English predominant across languages repre- observed, but functional communication gains were greatest
sented (255 IPD; 26.6%; Supplemental Material G). We for early 4 SLT-days/week. Participants that received
examined within study clustering; the findings were non- SLT > 3 months post-aphasia onset in a trial context made
significant or caused a model failure (the G-matrix was not greatest overall-language and auditory comprehension gains
positive definite).20,21 Network geometries were stable.15 for 5 SLT-days/week. No significant functional communi-
Overall, the greatest language gains from baseline to the first cation gains from baseline were observed in this late-reha-
follow-up occurred among working-age, female, moderate– bilitation group (Table 1; Supplemental Material I(a) to (c)).
severe aphasia severity subgroups, and those within 3 months
of stroke onset (Table 1; Supplemental Material H to K). Intensity.  The early-SLT group’s greatest overall-language
gains occurred for up to 2 SLT-h/week with similar gains
observed for 3-to-4 and > 9 SLT-h/week. Overall-language
Age and language rehabilitation gains for late-rehabilitation participants were less pro-
Frequency.  Working-age participants’ greatest overall-language nounced, with the greatest of these associated with SLT
gains occurred for 5 SLT-days/week (similar gains observed 3-to-4 h/week, with similar gains when SLT > 9 h/week.
for 3-to-6  + 
days/week) and for older participants’  ⩾ 6 Auditory comprehension gains were only observed when

International Journal of Stroke, 17(10)

 Table 1.  Subgroup analysis; greatest significant gains [95% CI] from baseline by SLT regimen, language outcome, and IPD (RCTs).
Brady

Frequency* Points [95% CI] Intensity* Points [95% CI] Dosage* Points [95% CI]
Subgroup days weekly IPD (RCTs) hours weekly IPD (RCTs) total SLT-hours IPD (RCTs)

Overall-language ability: WAB-AQ 0–100 points

⩽ 65 years 5 (⩾ 3-to-5) 15.1 [8.2, 22.1] > 9 (< 2) 17.0 [10.0, 24.0] 20-to-50 (> 50) 23.4 [13.5, 33.3]
IPD 107 (6) IPD 78 (3) IPD 15(3)

> 65 years ⩾ 5 (4) 17.2 [3.9, 30.5] up to 2 (3-to-4) 16.9 [3.8, 30.0] 20-to-50 (14-to-20) 15.95 [5.1, 26.8]
IPD 10 (2) IPD 37 (3) IPD 16(4)

SLT ⩽ 3 months 5 (⩾ 4) 27.7 [3.6, 51.9] up to 2 (3-to-4 and > 9) 24.3 [13.4, 35.2] 20-to-50 27.5 [18.3, 36.7]
IPD 3 (1) IPD 62 (2) IPD 27(3)

SLT > 3 months 5 (⩾ 4) 6.32 [1.6, 11.1] 3-to-4 (> 9) 6.3 [2.2, 10.3] ⩾ 50 10.1 [4.2, 16.0]
IPD 44 (1) IPD 25 (2) IPD 15 (1)

Severe–moderate aphasia 4 (2-to > 5) 18.3 [8.0, 28.5] 3-to-4 (up to 2 and > 9) 20.0 [10.4, 28.8] 20-to-50 (5-to-50) 23.5 [13.5, 33.5]
IPD 38 (5) IPD 48 (3) IPD 23 (3)

Mild–moderate aphasia > 5 (2-to > 5) 9.9 [4.4, 15.3] > 9 (up to 2 and > 3) 8.0 [3.4, 12.6] 20-to-50 (5-to-50) 8.7 [2.0, 15.5]
IPD 16 (2) IPD 48 (3) IPD 8 (2)

Female 5 (2-to > 5) 24.2 [7.7, 40.7] up to 2 (2 to > 9) 18.5 [7.5, 29.6] 20-to-50 24.4 [12.2, 36.5]
IPD 5 (2) IPD 37 (3) IPD 13 (2)

Male ⩾ 5 (⩾ 4) 13.6 [4.3, 22.9] > 9 (up to 2 and 3-to-4) 15.1 [7.9, 22.3] 20-to-50 (2-to-20) 15.6 [6.2, 25.0]
IPD 21 (2) IPD 61 (3) IPD 18 (4)

Auditory comprehension: AAT-TT 0–50 points

⩽ 65 years 4 (5) 6.8 [2.3, 11.2] > 9 9.0 [5.4, 12.6] 20-to-50 6.1 [1.8, 10.4]
IPD 64 (5) IPD 108 (4) IPD 59(6)

> 65 years 4 8.5 [2.0, 14.9] > 9 5.3 [0.7, 10.0] 20-to-50 5.8 [1.3, 10.2]
IPD 50 (3) IPD 33(6) IPD 34(7)

SLT ⩽ 3 months NS NS > 9 9.3 [2.1, 16.5] NS NS

IPD 20 (2)

SLT > 3 months 5 (4) 3.7 [1.4, 6.0] > 9 4.6 [2.4, 6.8] ⩾ 50 (> 14-to-50) 4.2 [1.7, 6.7]
IPD 89 (5) IPD 121 (4) IPD 76 (3)

Severe–moderate aphasia 4 8.5 [3.7, 13.3] > 9 (up to 2) 9.1 [2.6, 15.6] ⩾ 50 (> 14-to-20) 8.9 [4.5, 13.3]
IPD 80 (5) IPD 9 (2) IPD 142 (6)

Mild–moderate Aphasia NS NS NS NS NS NS
(Continued)

International Journal of Stroke, 17(10)

1071
 Table 1. (Continued)
1072

Frequency* Points [95% CI] Intensity* Points [95% CI] Dosage* Points [95% CI]
Subgroup days weekly IPD (RCTs) hours weekly IPD (RCTs) total SLT-hours IPD (RCTs)

Female 4 (5) 8.1 [2.7, 13.6] up to 2 (3-to-4 and > 9) 7.9 [1.8, 14.0] ⩾ 50 (20-to-50) 7.0 [2.9, 11.2]
IPD 52 (5) IPD 10 (2) IPD 77 (6)

Male 5 (4) 5.1 [1.4, 8.7] > 9 (3-to-4) 7.6 [3.8, 11.5] 20-to-50 6.5 [1.8, 11.2]
IPD 99 (8) IPD 91 (6) IPD 56 (7)

Functional communication: AAT-SSC, score 0–5 points

⩽ 65 years ⩾ 5 1.2 [0.3, 2.1] 3-to-4 0.9 [0.5, 1.2] 14-to-20 (⩾ 50) 1.1 [0.5, 1.8]
IPD 3 (2) IPD 74 (5) IPD 6 (3)

International Journal of Stroke, 17(10)

> 65 years 4 (5) 0.8 [0.2, 1.4] up to 2 1.0 [0.4, 1.6] 20-to-50 (> 5-to-14) 0.86 [0.4, 1.4]
IPD 54 (3) IPD 41 (4) IPD 31(9)

SLT ⩽ 3 months 4 1.6 [ 0.6, 2.5] up to 3 1.3 [0.6, 1.9] 20-to-50 1.2 [0.7, 1.8]
IPD 80 (1) IPD 57 (2) IPD 31 (3)

SLT > 3 months NS NS NS NS NS NS

Severe–moderate aphasia 4 1.0 [0.5, 1.5] < 2-to-3 1.0 [0.3, 1.6] ⩾ 50 (14-to-20) 1.1 [0.4, 1.8]
IPD 55 (2) IPD 42 (3) IPD 9 (3)

Mild–moderate aphasia > 5 (⩾ 4) 0.7 [0.4, 1.0] > 9 (< 2 and > 4) 0.63 [0.26, 1.0] > 14-to-20 (⩾ 50) 0.7 [0.2, 1.2]
IPD 55 (8) IPD 27 (4) IPD 8 (3)

Female 4 (3) 1.0 [0.2, 1.8] up to 2 1.2 [0.5, 1.8] > 14-to-20 (5-to- 1.6 [0.6, 2.6]
IPD 49(3) IPD 39(4) ⩾ 50) IPD 4 (2)

Male > 5 0.8 [0.1, 1.5] 3-to-4 (> 9) 0.7 [0.4, 0.9] ⩾ 50 (20-to-50) 0.7 [0.5, 0.9]
IPD 6(3) IPD 97(5) IPD 93 (6)

Key: MD mean difference. Underline text > 50% relative covariance reported for completeness. Bold refers to greatest gain associations. Italics refer to clinically similar gains.
International Journal of Stroke 17(10)
Brady 1073

SLT > 9 h/week regardless of the timing of the interven- Sex and language rehabilitation
tion. No functional communication gains were observed for Frequency.  Females’ greatest overall-language gains occurred
the late-SLT group; the early-SLT group’s gains were for 3 SLT-days/week (5 IPD) followed by 5 SLT-days/week;
greatest for 2–3 SLT-h/week (Table 1; Supplemental Mate- similar gains observed across frequencies. Males’ greatest
rial I(d) to (f)). overall-language gains were for ⩾ 6 SLT-days/week (similar
gains for 4-to-5 SLT-days/week) while greatest functional
Dosage.  The early-SLT group achieved their greatest over- communication gains occurred for  ⩾ 5 SLT-day/week.
all-language and functional communication gains for 20– Females’ functional communication gains were greatest for 4
50 SLT-h; auditory comprehension gains were not observed SLT-days/week with similar gains observed for 3 SLT-days/
at any dosage. The late-SLT group’s only significant over- week. Both female and male groups’ greatest and only com-
all-language gain from baseline occurred for > 50 SLT-h; prehension gains occurred for 4-to-5 SLT-days/week. No
gains in auditory comprehension were significant for > 14 auditory comprehension gains were observed when SLT < 4
SLT-h with the greatest of these for > 50 SLT-h. The late- or 6 days/week (Table 1; Supplemental Material K(a) to (c)).
SLT group made no significant functional communication
gains from baseline at any dosage (Table 1; Supplemental Intensity.  Females’ greatest overall-language and func-
Material I(g) to (i)). tional communication gains occurred for < 2 SLT-h/week
as did their greatest auditory comprehension gains (the lat-
Aphasia severity and language rehabilitation ter based on few IPD). The next greatest comprehension
gains were observed for ⩾ 9 SLT-h/weekly. Similar over-
Frequency.  When SLT was 4 days/week, participants with all-language gains also occurred at 2 to > 9 SLT-h/week.
moderate–severe aphasia experienced the greatest overall- Males’ greatest overall-language and comprehension
language, auditory comprehension, and functional commu- gains occurred for > 9 SLT-h/week, but greatest functional
nication gains. Comprehension gains were only observed communication gains occurred for 3–4 SLT-h/week. Similar
for ⩾ 4 SLT-days/week. In contrast, the mild–moderate gains were observed for > 9 SLT-h/week for overall lan-
group’s greatest gains occurred for ⩾ 6 SLT-days/week guage, auditory comprehension. Comprehension gains
(overall language) and functional communication from ⩾ 4 among male participants were absent when SLT ⩽ 3 h/week
SLT-days/week. Relative variance for auditory compre- (Table 1; Supplemental Material K(d) to (f)).
hension analysis was high (> 50%) (Table 1; Supplemental
Material J(a) to (c)). Dosage.  Females and males’ greatest overall-language gains
occurred alongside 20-to-50 SLT-h (males made similar
Intensity.  The moderate–severe group made significant gains for > 50 SLT-h). For males’ greatest comprehension,
gains on overall language at all intensities, with greatest gains were also observed for 20-to-50 SLT-h, whereas
gains for 3-to-4 SLT-h/week. Their greatest auditory com- females’ comprehension gains were greatest for > 50 SLT-h.
prehension gains were observed for < 2 SLT-h/week but Comprehension gains were absent for both groups for < 20
based on few participants, while similar gains were SLT-h. Females’ greatest functional communication gains
observed for > 9 SLT-h/week but informed by more IPD. occurred for > 14–20 SLT-h (based on few IPD; gains
When SLT was > 9 h/week, the mild–moderate group observed at all dosages > 5 SLT-h). Males’ functional com-
made their greatest overall-language gains (similar gains < 2 munication was greatest alongside ⩾ 50 SLT-h with clini-
and > 3 SLT-h/week). They made significant functional cally similar gains observed for > 20–50 SLT-h (Table 1;
communication gains across intensities. Relative variance Supplemental Material K(g) to (i)).
of the auditory comprehension analysis for this group
was > 50% (Table 1; Supplemental Material J(d) to (f)).
Discussion
Dosage.  The moderate–severe group made greatest over-
all-language gains for 20-to-50 SLT-h with similar gains Our IPD network meta-analysis of 959 individual datasets
for 5–50 + SLT-h. Greatest gains were observed for the (25 RCTs) explored patterns of interaction between SLT-
mild–moderate group for > 20 h. The relative variance frequency, intensity, and dosage and aphasia language
was > 50% for the mild–moderate group’s auditory com- ­outcomes by age, sex, aphasia chronicity and severity sub-
prehension analysis; no gains were observed. The moder- groups. Some subgroup consistencies were evident; great-
ate–severe group’s greatest comprehension and functional est overall-language gains were associated with 20–50
communication gains occurred for ⩾ 50 SLT-h. The mild– SLT-h; comprehension gains were only evident > 9 SLT-h
moderate group made significant functional communica- over ⩾ 4 SLT-days/week.15 Most subgroups demonstrated
tion gains at all dosages (the greatest observed for > 14–20 gains from baseline to first post-intervention follow-up
SLT-h but based on 8 IPD). (Table 1; Supplemental across outcomes, consistent with previous pairwise, aggre-
­Material J(g) to (i)). gate data, and meta-analyses.5,7 Our findings also suggest

International Journal of Stroke, 17(10)

1074 International Journal of Stroke 17(10)

that differential aphasia rehabilitation responses may exist and language change data were unavailable. Concurrent
for some subgroups. Older participants’ overall-language impairment or comorbidities were rarely reported. Other
gains were greatest when associated with lower-intensity demographic data were inconsistently available.19 Our
SLT than working-age participants’ gains (⩽ 2 vs > 9 analysis controlled for time post-stroke and aphasia sever-
SLT-h/week); optimal frequency and dosage were similar ity, though other threats to the validity of the effect estima-
across groups. Generally, early-intervention participants’ tion may exist, including participants’ tolerance to highly
greatest overall-language gains occurred for up to 2 SLT-h/ intensive SLT early post-stroke. Our statistical inferencing,
week for 20-to-50 h; when SLT > 3 months post-onset hypothesis generating approach carries a risk of false nega-
greatest gains occurred for 3-to-4 SLT-h/week for ⩾ 50 h. tive and positive findings. Replication of these findings
Moderate–severe participants made significant overall- through confirmatory clinical trials is required.
language gains across intensities (greatest for 3–4 SLT-h/
weekly); functional communication gains were greatest
Clinical implications
for < 3 SLT-h/week. In contrast, mild–moderate partici-
pants’ gains were associated with higher-intensity SLT Greatest overall-language and auditory comprehension
(> 9 h/week). Males’ greatest language gains were associ- gains across subgroups were associated with higher dosage
ated with high-frequency and -intensity SLT (⩾ 5 SLT- levels > 20–50 SLT-h or above) than current clinical provi-
days/week; > 9 h/week), while females’ greatest language sion reported for 4–16 h.27–30 Higher dosage rehabilitation
gains were associated with SLT-frequency and intensity of within existing resources may be challenging, requiring
4-to-5 SLT-days/week; < 2 h/weekly. alternative delivery models to achieve the requisite dose,
Previous aggregate meta-analyses suggested optimal such as telerehabilitation, self-management, trained family
SLT intensities were ⩾ 2 SLT-h/week24 or in the region of 9 members, prescribed home-practice tasks, and group ther-
SLT-h/week7, while dosage of > 90 SLT-h was likely to apy. Our findings also highlighted frequency–intensity–
confer language benefit but < 44 SLT-h may not.24 Our dosage ranges below which language gains from baseline
analysis refines these estimates, highlighting variations by were not observed. Some plausible, clinically relevant sub-
language outcome and subgroups. groups may benefit from precision rehabilitation approaches
Subgroup meta-analyses carry intrinsic strengths and limi- based on age, aphasia severity, chronicity, and sex.
tations.25,26 The high number of diverse IPD (including public
domain datasets, languages, clinical, and regional contexts) Research implications
sought to maximize data inclusion, ensure sufficient data
overlap, and support generalization of findings. Our inclusion High-quality targeted SLT RCTs should be conducted to
of trials completed beyond our search–end date ensured the evaluate the acceptability, clinical, and cost effectiveness of
dataset’s currency. Trial registration is a requirement met by precision aphasia rehabilitation approaches. Participants
high-quality trials, but an unregistered trial may have been should be selected by age-group, time since stroke, and
missed from this search strategy. Strong rationale, early severity or stratifying intervention by subgroup. Despite no
empirical evidence, and plausible clinical perceptions of dif- language restrictions, our predominantly English-speaking
ferential responsiveness to SLT supported our multiple, participant data highlighted an underrepresentation of non-
planned, and subgroup network meta-analyses. Included English aphasia research. Minimal SLT-frequency–
RCTs had a low risk of bias (Supplemental Material N), and intensity–dosage levels should be applied to intervention
of the 25 included RCTs, only ten participants did not have development and evaluation, in addition to stratification by
the data points necessary to contribute to our planned analy- age, time post-stroke, and severity. Continued collaborative
sis. Our meta-synthesis preserved clinically relevant meas- approaches including research data sharing will support the
urements, supporting clinical interpretation. reduction of research waste and further insights into preci-
We acknowledge that spontaneous recovery may also sion stroke rehabilitation, including SLT for aphasia.
impact on treatment gains observed, with median overall-
language gains higher for the early-rehabilitation subgroup
Conclusion
than the late-rehabilitation group. In addition, patients in
the acute stroke stage may have reduced capacity to engage Exploratory IPD meta-analysis based on aphasia RCT IPD
in SLT or RCT activities.6 The extent of such impacts demonstrated that most subgroups with aphasia made sig-
remains to be determined. Our exploratory IPD meta-anal- nificant overall-language ability, auditory comprehension,
ysis reflects highly selected participants, interventions, lan- and functional communication gains from baseline and sug-
guage outcomes, and the availability of sufficiently detailed gested that some subgroups may achieve their greatest lan-
records within included RCTs. Where limited data were guage gains in the context of specific SLT-frequency,
available, there remain uncertainties in our findings. intensity, and dosage regimens. Where differences arose,
Language gains observed reflect the change from RCT older, moderate–severely impaired, and female subgroups’
baseline to the first follow-up only. Pre-randomization SLT greatest gains were associated with lower-intensity SLT.

International Journal of Stroke, 17(10)

Brady 1075

Working-aged, mild–moderate aphasia, and male subgroups’ recommendations for conduct and planning. Stat Med 2020;
greatest language gains were associated with high-frequency/ 39: 2115–2137.
intensity SLT. 9. Riley RD, Lambert PC and Abo-Zaid G. Meta-analysis of
individual participant data: rationale, conduct, and reporting.
Acknowledgements BMJ 2010; 340: c221.
10. Kirkham JJ, Dwan KM, Altman DG, Gamble C, Dodd S,
Detailed in Supplemental material. Smyth R and Williamson PR. The impact of outcome report-
ing bias in randomised controlled trials on a cohort of system-
Authorship atic reviews. BMJ 2010; 340: c365.
The RELEASE Collaborators (listed in the Appendix 1). 11. Tierney JF, Vale C, Riley R, Smith CT, Stewart L, Clarke M
and Rovers M. Individual participant data (IPD) meta-anal-
yses of randomised controlled trials: guidance on their use.
Declaration of conflicting interests
PLoS Med 2015; 12: e1001855.
Detailed in Supplemental material. 12. Cochrane Training. Cochrane handbook for systematic reviews
of interventions, https://training.cochrane.org/handbook (2020,
Funding accessed 12 April 2022).
The author disclosed receipt of the following financial support for 13. Efthimiou O, Debray TPA, van Valkenhoef G, et al. GetReal
the research, authorship, and/or publication of this article: This in network meta-analysis: a review of the methodology. Res
research was funded by the National Institute for Health Research Synth Methods 2016; 7: 236–263.
Health Services and Delivery Research (14/04/22; Grantholders 14. Smith GD, Egger M and Phillips AN. Meta-analysis: beyond
MB, MA, AB, EG, KH, JH, SH, DH, TK, ACL, BMacW, RP, CP, the grand mean? BMJ 1997; 315: 1610–1614.
SAT, EVB, LW) and the Tavistock Trust for Aphasia (Grantholder 15. REhabilitation and recovery of peopLE with Aphasia after
MB) and was undertaken within the Tavistock Trust for Aphasia StrokE (RELEASE) Collaborators. Dosage, intensity, and fre-
funded Collaboration of Aphasia Trialists. quency of language therapy for aphasia: a systematic review–
based, individual participant data network meta-analysis.
Protocol registration Stroke 2022; 53: 956–967.
16. Belias M, Rovers MM, Reitsma JB, Debray TPA and IntHout
PROSPERO CRD42018110947. J. Statistical approaches to identify subgroups in meta-anal-
ysis of individual participant data: a simulation study. BMC
Supplemental material Med Res Methodol 2019; 19: 183.
17. Debray TPA, Schuit E, Efthimiou O, et al. An overview

Supplemental material for this article is available online.
of methods for network meta-analysis using individual
­participant data: when do benefits arise? Stat Methods Med
References Res 2018; 27: 1351–1364.
1. Mawson S, Nasr N, Parker J, Zheng H, Davies R and Mountain 18. Tierney JF, Pignon JP, Gueffyier F, Clarke M, Askie L, Vale CL
G. Developing a personalised self-management system for and Burdett S; Cochrane IPD Meta-analysis Methods Group.
post stroke rehabilitation; utilising a user-centred design meth- How individual participant data meta-analyses have i­ nfluenced
odology. Disabil Rehabil Assist Technol 2014; 9: 521–528. trial design, conduct, and analysis. J Clin Epidemiol 2015; 68:
2. Kosorok MR and Laber EB. Annual review of statistics and its 1325–1335.
application. Precis Med 2019; 6: 263–286. 19. Williams LR, Ali M, VandenBerg K, et al. Utilising a sys-
3. Bachmann S, Finger C, Huss A, Egger M, Stuck AE and tematic review-based approach to create a database of indi-
Clough-Gorr KM. Inpatient rehabilitation specifically designed vidual participant data for meta- and network meta-analyses:
for geriatric patients: systematic review and meta-analysis of the RELEASE database of aphasia after stroke. Aphasiology
randomised controlled trials. BMJ 2010; 340: c1718. 2022; 36: 513–533.
4. Bernhardt J, Hayward KS, Dancause N, et al. A stroke recovery 20. REhabilitation and recovery of peopLE with Aphasia after
trial development framework: consensus-based core recom- StrokE (RELEASE) Collaborators. Predictors of poststroke
mendations from the second stroke recovery and rehabilitation aphasia recovery. Stroke 2021; 52: 1778–1787.
roundtable. Neurorehabil Neural Repair 2019; 33: 959–969. 21. Brady MC, Ali M, VandenBerg K, et al. RELEASE: a pro-
5. Brady MC, Kelly H, Godwin J, Enderby P and Campbell P. tocol for a systematic review based, individual participant
Speech and language therapy for aphasia following stroke. data, meta- and network meta-analysis, of complex speech-
Cochrane Database Syst Rev 2016; 2016: CD000425. language therapy interventions for stroke-related aphasia.
6. Royal College of Physicians. National clinical guidelines for Aphasiology 2020; 34: 137–157.
stroke, https://www.strokeaudit.org/SupportFiles/Documents/ 22. Kertesz A. The Western aphasia battery. 1st ed. San Antonio,
Guidelines/2016-National-Clinical-Guideline-for-Stroke-5t-(1). TX: The Psychological Corporation, 1982.
aspx (2016, accessed 12 April 2022). 23. Huber W, Poeck K, Weniger D and Willmess K. Aachener
7. Bhogal SK, Teasell R and Speechley M. Intensity of aphasia apahsie test: verlag fur psychologie. 2nd ed. Florence:
therapy, impact on recovery. Stroke 2003; 34: 987–993. Organizzazioni Speciali Firenze, 1996.
8. Riley RD, Debray TPA, Fisher D, et al. Individual partici- 24. Robey RR. A meta-analysis of clinical outcomes in the
pant data meta-analysis to examine interactions between treatment of aphasia. J Speech Lang Hear Res 1998; 41:
treatment effect and participant-level covariates: statistical 172–187.

International Journal of Stroke, 17(10)

1076 International Journal of Stroke 17(10)

25. Spears MR, James ND and Sydes MR. ‘Thursday’s child has
Anthony Pak-Hin Kong https://orcid.org/0000-0002-6211-0358
far to go’—interpreting subgroups and the STAMPEDE trial.
Ann Oncol 2017; 28: 2327–2330. Tarja Kukkonen https://orcid.org/0000-0002-8189-0337
26. Wang R, Lagakos SW, Ware JH, Hunter DJ and Drazen JM. Marina Laganaro https://orcid.org/0000-0002-4054-0939
Statistics in medicine—reporting of subgroup analyses in Matthew A Lambon Ralph https://orcid.org/0000-0001-5907-2488
clinical trials. N Engl J Med 2007; 357: 2189–2194.
27. Godecke E, Armstrong E, Rai T, et al. The Very Early
Ann Charlotte Laska https://orcid.org/0000-0002-7330-940X
Rehabilitation in Speech (VERSE) after stroke trial: an inter- Béatrice Leemann https://orcid.org/0000-0003-2226-6777
national 3-arm clinical trial to determine the effectiveness Alexander P Leff https://orcid.org/0000-0002-0831-3541
of early, intensive, prescribed, direct aphasia therapy. Front
Roxele R Lima https://orcid.org/0000-0002-9914-4789
Hum Neurosci. Epub ahead of print 22 January 2019. DOI:
10.3389/conf.fnhum.2018.228.00034. Antje Lorenz https://orcid.org/0000-0002-0200-1977
28. Code C and Petheram B. Delivering for aphasia. Int J Speech Brian MacWhinney https://orcid.org/0000-0002-4988-1342
Lang Pathol 2011; 13: 3–10. Rebecca Shisler Marshall https://orcid.org/0000-0001-9313-5454
29. Verna A, Davidson B and Rose TA. Speech-language pathol-
Flavia Mattioli https://orcid.org/0000-0002-4912-5520
ogy services for people with aphasia: a survey of current prac-
tice in Australia. Int J Speech Lang Pathol 2009; 11: 191–205. İlknur Maviş https://orcid.org/0000-0003-3924-1138
30. Palmer R, Witts H and Chater T. What speech and language Marcus Meinzer https://orcid.org/0000-0003-1370-3947
therapy do community dwelling stroke survivors with aphasia Reza Nilipour https://orcid.org/0000-0003-4180-7989
receive in the UK? PLoS ONE 2018; 13: e0200096.
Enrique Noé https://orcid.org/0000-0002-2547-8727
Nam-Jong Paik https://orcid.org/0000-0002-5193-8678
ORCID iDs Rebecca Palmer https://orcid.org/0000-0002-2335-7104
Ilias Papathanasiou https://orcid.org/0000-0003-0999-696X
Marian C Brady https://orcid.org/0000-0002-4589-7021
Brígida F Patrício https://orcid.org/0000-0002-2619-470X
Myzoon Ali https://orcid.org/0000-0001-5899-2485
Isabel Pavão Martins https://orcid.org/0000-0002-9611-7400
Kathryn VandenBerg https://orcid.org/0000-0001-5035-9650
Cathy Price https://orcid.org/0000-0003-0111-9364
Louise R Williams https://orcid.org/0000-0003-2430-1142
Tatjana Prizl Jakovac https://orcid.org/0000-0002-5018-9556
Linda J Williams https://orcid.org/0000-0002-6317-1718
Elizabeth Rochon https://orcid.org/0000-0001-5521-0513
Masahiro Abo https://orcid.org/0000-0001-6701-4974
Miranda L Rose https://orcid.org/0000-0002-8892-0965
Frank Becker https://orcid.org/0000-0002-0857-0628
Charlotte Rosso https://orcid.org/0000-0001-7236-1508
Audrey Bowen https://orcid.org/0000-0003-4075-1215
Ilona Rubi-Fessen https://orcid.org/0000-0002-9775-3812
Caitlin Brandenburg https://orcid.org/0000-0002-6992-7790
Marina B Ruiter https://orcid.org/0000-0001-6147-5235
Caterina Breitenstein https://orcid.org/0000-0002-6408-873X
Claerwen Snell https://orcid.org/0000-0001-8606-7801
Stefanie Bruehl https://orcid.org/0000-0003-4826-1990
Benjamin Stahl https://orcid.org/0000-0003-3957-1495
David A Copland https://orcid.org/0000-0002-2257-4270
Jerzy P Szaflarski https://orcid.org/0000-0002-5936-6627
Tamara B Cranfill https://orcid.org/0000-0001-7608-6443
Shirley A Thomas https://orcid.org/0000-0003-0704-9387
Marie di Pietro-Bachmann https://orcid.org/0000-0001-8027-2337
Mieke van de Sandt-Koenderman https://orcid.org/0000-0002-
Pamela Enderby https://orcid.org/0000-0002-4371-9053
8104-6937
Joanne Fillingham https://orcid.org/0000-0002-0363-8021
Ineke van der Meulen https://orcid.org/0000-0002-6156-3873
Federica Lucia Galli https://orcid.org/0000-0001-9244-9179
Evy Visch-Brink https://orcid.org/0000-0001-7833-0112
Marialuisa Gandolfi https://orcid.org/0000-0002-0877-4807
Linda Worrall https://orcid.org/0000-0002-3283-7038
Bertrand Glize https://orcid.org/0000-0001-9618-2088
Heather Harris Wright https://orcid.org/0000-0001-6922-6364
Erin Godecke https://orcid.org/0000-0002-7210-1295
David Howard https://orcid.org/0000-0001-9141-5751
Katerina Hilari https://orcid.org/0000-0003-2091-4849
Neil Hawkins https://orcid.org/0000-0003-3199-221X
Jacqueline Hinckley https://orcid.org/0000-0002-4052-1420
Simon Horton https://orcid.org/0000-0002-2133-1410
Appendix 1
Petra Jaecks https://orcid.org/0000-0002-5878-1327
Elizabeth Jefferies https://orcid.org/0000-0002-3826-4330 The RELEASE collaboration Marian C Brady1 , Myzoon
Luis MT Jesus https://orcid.org/0000-0002-8534-3218 Ali1 , Kathryn VandenBerg1 , Linda J Williams2,
Maria Kambanaros https://orcid.org/0000-0002-5857-9460 Louise R Williams1 , Masahiro Abo3 , Frank Becker4 ,
Eun Kyoung Kang https://orcid.org/0000-0001-5315-1361 Audrey Bowen5 , Caitlin Brandenburg6 , Caterina
Eman M Khedr https://orcid.org/0000-0001-5679-9833 Breitenstein7 , Stefanie Bruehl5,8 , David A Copland6 ,

International Journal of Stroke, 17(10)

Brady 1077

18. City, University of London, UK

Tamara B Cranfill9 , Marie di Pietro-Bachmann10 ,
19. Nova Southeastern University, USA
Pamela Enderby11 , Joanne Fillingham12 , Federica 20. University of East Anglia, UK
Lucia Galli13 , Marialuisa Gandolfi14 , Bertrand 21. Newcastle University, UK
Glize15 , Erin Godecke16 , Neil Hawkins17 , Katerina 22. Bielefeld University, Germany
Hilari18 , Jacqueline Hinckley19 , Simon Horton20 , 23. University of York, UK
David Howard21 , Petra Jaecks22 , Elizabeth Jefferies23 , 24. University of Aveiro, Portugal
25. Cyprus University of Technology, Cyprus
Luis MT Jesus24 , Maria Kambanaros25 , Eun Kyoung
26. Kangwon National University Hospital, Republic of
Kang26 , Eman M Khedr27 , Anthony ­Pak-Hin Kong28 , Korea
Tarja Kukkonen29 , Marina Laganaro10 , Matthew A Lambon 27. Assiut University Hospital, Egypt
Ralph30 , Ann Charlotte Laska31 , Béatrice Leemann32, 28. University of Hong Kong, Hong Kong
Alexander P Leff33 , Roxele R Lima34 , Antje Lorenz35 , 29. Tampere University Hospital, Finland
Brian MacWhinney36 , Rebecca Shisler Marshall37 , 30. University of Cambridge, UK
31. Karolinska Institutet, Sweden
Flavia Mattioli38 , İlknur Maviş39 , Marcus Meinzer40 ,
32. Hôpitaux Universitaires de Genève, Switzerland
Reza Nilipour41 , Enrique Noé42 , Nam-Jong Paik43 , 33. University College London, UK
Rebecca Palmer11 , Ilias Papathanasiou44 , Brigida 34. Educational Association Bom Jesus—IELUSC,
Patricio45 , Isabel Pavão Martins46 , Cathy Price33 , Brazil
Tatjana Prizl Jakovac47 , Elizabeth Rochon48 , Miranda 35. Humboldt University Berlin, Germany
L Rose49 , Charlotte Rosso50 , Ilona Rubi-Fessen51 , 36. Carnegie Mellon University, USA
37. University of Georgia, USA
Marina B Ruiter52 , Claerwen Snell53 , Benjamin
38. Azienda Socio Sanitaria Territoriale, Italy
Stahl54 , Jerzy P Szaflarski55 , Shirley A Thomas56 , 39. Anadolu University, Turkey
Mieke van de Sandt-Koenderman57 , Ineke van der 40. University Medicine Greifswald, Germany
Meulen58 , Evy Visch-Brink58 , Linda Worrall6 , 41. University of Social Welfare and Rehabilitation
Heather Harris Wright.59 Sciences, Iran
42. NEURORHB-Hospitales Vithas, Spain
1. Glasgow Caledonian University, UK 43. Seoul National University College of Medicine,
2. University of Edinburgh, UK Republic of Korea
3. The Jikei University School of Medicine, Japan 44. University of Patras, Patras, Greece
4. University of Oslo, Sunnaas Rehabilitation 45. School of Polytechnic Institute of Porto, Portugal
Hospital, Norway 46. Universidade de Lisboa, Portugal
5. MAHSC, University of Manchester, UK 47. University of Zagreb, Croatia
6. The University of Queensland, Australia 48. University of Toronto, Canada
7. University of Muenster, Germany 49. La Trobe University, Australia.
8. St Mauritius Rehabilitation Centre, and RWTH 50. Sorbonne Université, and Hôpital Salpetriere, France
Aachen University, Germany 51. RehaNova Rehabilitation Hospital and University
9. Eastern Kentucky University, USA of Cologne, Germany
10. University Hospital and University of Geneva, 52. Sint Maartenskliniek and Radboud University,
Switzerland Netherlands
11. University of Sheffield, UK 53. Warrington and Halton NHS Foundation Trust, UK
12. NHS Improvement, London, UK 54. Charité Universitätsmedizin Berlin, Germany
13. Marche Polytechnic University, Italy 55. University of Alabama at Birmingham, USA
14. University of Verona, Italy 56. University of Nottingham, UK
15. University of Bordeaux, France 57. University Medical Center Rotterdam, Netherlands
16. Edith Cowan University, Australia 58. Erasmus University Medical Center, Netherlands
17. University of Glasgow, UK 59. North Carolina University, USA.

International Journal of Stroke, 17(10)