Systematic Review and Meta-Analysis Medicine ® OPEN Subacute stroke physical rehabilitation evidence in activities of daily living outcomes A systematic review of meta-analyses of randomized controlled trials ∗ Alejandro García-Rudolph, PhDa,b,c, , David Sánchez-Pinsach, MSca,b,c, Eloy Opisso Salleras, PhDa,b,c, Josep María Tormos, PhDa,b,c Abstract Background: Stroke is a leading cause of disabilities worldwide. One of the key disciplines in stroke rehabilitation is physical therapy which is primarily aimed at restoring and maintaining activities of daily living (ADL). Several meta-analyses have found different interventions improving functional capacity and reducing disability. Objectives: To systematically evaluate existing evidence, from published systematic reviews of meta-analyses, of subacute physical rehabilitation interventions in (ADLs) for stroke patients. Methods: Umbrella review on meta-analyses of RCTs ADLs in MEDLINE, Web of Science, Scopus, Cochrane, and Google Scholar up to April 2018. Two reviewers independently applied inclusion criteria to select potential systematic reviews of meta-analyses of randomized controlled trials (RCTs) of physical rehabilitation interventions (during subacute phase) reporting results in ADLs. Two reviewers independently extracted name of the 1st author, year of publication, physical intervention, outcome(s), total number of participants, and number of studies from each eligible meta-analysis. The number of subjects (intervention and control), ADL outcome, and effect sizes were extracted from each study. Results: Fifty-five meta-analyses on 21 subacute rehabilitation interventions presented in 30 different publications involving a total of 314 RCTs for 13,787 subjects were identified. Standardized mean differences (SMDs), 95% confidence intervals (fixed and random effects models), 95% prediction intervals, and statistical heterogeneity (I2 and Q test) were calculated. Virtual reality, constraint- induced movement, augmented exercises therapy, and transcranial direct current stimulation interventions resulted statistically significant (P < .05) with moderate improvements (0.5  SMD  0.8) and no heterogeneity (I2 = 0%). Moxibustion, Tai Chi, and acupuncture presented best improvements (SMD > 0.8) but with considerable heterogeneity (I2 > 75%). Only acupuncture reached “suggestive” level of evidence. Conclusion: Despite the range of interventions available for stroke rehabilitation in subacute phase, there is lack of high-quality evidence in meta-analyses, highlighting the need of further research reporting ADL outcomes. Abbreviations: ADLs = activities of daily living, AMSTAR = assessment of multiple systematic reviews, CI = confidence interval, CIM = constraint-induced movement, FES = functional electrical stimulation, GBD = global burden of disease, MD = mean difference, PI = prediction interval, PICOS = population, intervention, comparison, outcome, study design, RCT = randomized controlled trial, rPMS = repetitive peripheral magnetic stimulation, rTMS = repetitive transcranial magnetic stimulation, RTT = repetitive task training, SMD = standard mean difference, tDCS = transcranial direct current stimulation. Keywords: activities of daily living, meta-analysis, rehabilitation, stroke, subacute interventions, umbrella review Editor: Weimin Guo. 1. Introduction This research was partially funded by EU H2020 PRECISE4Q - Personalized Medicine by Predictive Modeling in Stroke for better Quality of Life (Grant Stroke is the 2nd cause of mortality and the 3rd cause of long- Agreement 777107 – Research and Innovation Action). term disability worldwide with 33 million stroke survivors.[1] The authors have no conflicts of interest to disclose. Mortality is declining, yet prevalence is stable, meaning there are Supplemental Digital Content is available for this article. more survivors with long-term disability.[2] Using Global Burden a of Disease incidence rates, between 2015 and 2035, it is projected Department of Research and Innovation, Institut Guttmann, Institut Universitari de Neurorehabilitació adscrit a la UAB, Badalona, Barcelona, b Universitat Autònoma de that the number of stroke survivors in the European Union will Barcelona, Bellaterra (Cerdanyola del Vallès), c Fundació Institut d’Investigació en rise from 3,718,785 in 2015 to 4,631,050 in 2035, an increase of Ciències de la Salut Germans Trias i Pujol, Badalona, Barcelona, Spain. 25%.[3] In the United States, approximately every 40 seconds ∗ Correspondence: Alejandro García-Rudolph, Department of Research and someone experiences a stroke, as death rates have declined, Innovation, Institut Guttmann – Hospital de Neurorehabilitació, Cami Can Ruti s/n stroke has become the leading cause of long-term disability.[4] 08916, Badalona, Barcelona, Spain (e-mail: agarciar@guttmann.com). Interdisciplinary complex rehabilitation interventions are Copyright © 2019 the Author(s). Published by Wolters Kluwer Health, Inc. assumed to represent the mainstay of poststroke care.[5] One of This is an open access article distributed under the terms of the Creative Commons Attribution-Non Commercial License 4.0 (CCBY-NC), where it is permissible to the key disciplines in interdisciplinary stroke rehabilitation is download, share, remix, transform, and buildup the work provided it is properly cited. physical therapy which is primarily aimed at restoring and The work cannot be used commercially without permission from the journal. maintaining activities of daily living (ADL).[6] In this work, we Medicine (2019) 98:8(e14501) consider physical therapy as “therapeutic modalities frequently Received: 20 September 2018 / Received in final form: 4 January 2019 / used in physical therapy specialty by physical therapists or Accepted: 21 January 2019 physiotherapists to promote, maintain, or restore the physical and http://dx.doi.org/10.1097/MD.0000000000014501 physiological well-being of an individual” (Mesh, MEDLINE 1 García-Rudolph et al. Medicine (2019) 98:8 Medicine Subject Heading). Although stroke patients at an early stage 2.2. Inclusion and exclusion criteria depend on a stroke unit in the acute hospital, their functional Studies were included if they met the following criteria, recovery and long-term health status are more affected by subacute established by using the PICOS (Population, Intervention, (1–6 months) rehabilitation hospital.[7] Comparison, Outcome, Study Design)[19] strategy presented in Several meta-analyses have found different interventions Appendix - Supplementary Table 1, http://links.lww.com/MD/ improving functional capacity and reducing disability.[7] A recent C839. umbrella review[8] summarizes the effects of exercise therapy on Study design: meta-analyses (quantitative analysis) of RCTs in functional capacity in patients considering 22 different chronic subacute rehabilitation phase (1–6 months after onset). diseases. Eighty-five meta-analyses were included, nevertheless Study population: stroke patients in subacute rehabilitation only 11 of them reported outcomes on stroke of which 5 refer to phase >18 years. ADLs, in our work we aim to update (only 6 of the 11 studies are Outcomes: ADL outcome scales, regarded as continuous dated after 2013) and extend that body of knowledge. scaled, where usually higher scores indicate a good outcome, for Veerbeek and colleagues[9] retrieved randomized controlled example, functional independence measure,[20] Barthel index,[21] trials (RCTs) regarding physical therapy in stroke. In their modified Rankin scale,[22] Frenchay activities index,[23] River- analysis, most database searches (90%) were performed by mid- mead ADL Assessment,[24] Katz index of independence in 2011. In this work, we aim to link interventions specifically to ADL,[25] motor activity log,[26] and modified Barthel index.[27] ADLs and include new therapeutic approaches in which physical Eligible articles were required to be meta-analyses of RCTs, have exercise is combined with innovative treatments enhancing outcome measures related to ADLs, compare physical therapy neuroplasticity, such as transcranial direct current stimulation[10] with no treatment or usual care, have adult participants, as or repetitive peripheral magnetic stimulation.[11] defined by the Cochrane Collaboration,[28] and have stroke A nonspecific systemic inflammatory response occurs after patients as defined by World Health Organization.[29] Meta- both ischemic and hemorrhagic stroke, either as part of the analyses were excluded if intervention did not clearly take place process of brain damage or in response to complications such as during subacute rehabilitation phase (considering the following deep venous thrombosis.[12] Inflammation may be important criteria as in related research[30] acute: <1 month after stroke, both before, in predisposing to a stroke, and afterwards, where it subacute: 1 to 6 months after stroke and chronic: more than takes part in the mechanisms of cerebral injury and repair.[13,14] 6 months after stroke), they did not report the number of studies Clinically, the susceptibility of the patients to stroke and the or participants in experimental or control groups, they assessed subsequent prognosis are influenced by such inflammatory postsurgical recovery or site-specific musculoskeletal conditions processes.[15–17] As stroke patients with systemic inflammation such as patellofemoral pain syndrome (although physical therapy have been reported to exhibit clinically poorer outcomes,[18] in may be a standard treatment). We excluded meta-analyses this work, we will report mentions to inflammatory processes in that did not present study-specific data (effect size and 95% the identified meta-analyses. confidence intervals [CIs]). To the best of our knowledge, no attempts of reviewing the existing literature through an umbrella review in stroke rehabilitation has been conducted. Umbrella review offers the 2.3. Data extraction possibility to analyze the strength of evidence and extent of We extracted the name of the 1st author, year of publication, potential biases in the association between physical interventions physical intervention, ADL outcome(s), and number of studies in subacute rehabilitation and ADL outcomes. from each eligible meta-analysis. From each individual study in a meta-analysis, we extracted the 1st author, year of publication, total number of subjects assigned to the intervention and to the 2. Methods control groups, mean value and standard deviation of ADL outcome, and maximally adjusted effect size measurements 2.1. Literature search (mean difference [MD] or standardized MD [SMD]) along with According to the Joanna Briggs Institute Umbrella Review the corresponding 95% CI. Two investigators (AGR, DSP) Methodology,[19] literature search was conducted independently independently searched the literature, assessed the eligibility of by 2 reviewers (AGR, DSP) in MEDLINE (2000 through April the retrieved papers, and extracted the data using a standard pro- 2018), Web of Science (through April 2018), Scopus (through form. Disagreements were resolved by consensus with a third April 2018), Cochrane database of systematic reviews, and senior investigator (EOS or JMT). Google Scholar (up to mid-April 2018). The search strategy included combinations of multiple search terms for 2 themes: 2.4. Assessment of methodological quality of included Stroke and interventions (rehabilitation). The keywords used to search for studies for this review are listed in Appendix Search studies strategy, http://links.lww.com/MD/C839. All meta-analyses Both reviewers independently assessed the methodological registered in these databases that reported a systematic electronic quality of each review, using the AMSTAR (Assessment of search of literature for a defined period of time were included. Multiple Systematic Reviews) appraisal tool.[31] Disagreements Bibliographies of identified articles and manual search of relevant were resolved by consensus with a 3rd senior investigator (EOS journals for additional references was conducted. The most or JMT) (see Appendix Supplementary Table 3, http://links.lww. updated or complete publication was used when more than an com/MD/C839 for details). article was present for a single study. In addition, separate meta- analyses on multiple outcomes presented in a single article were 2.5. Data analysis assessed separately. Gray literature search was conducted using different internet search engines and websites: such as Google Estimation of summary effect: The summary effect size and its CIs Scholar. by 95% were estimated using both fixed effects and random 2 García-Rudolph et al. Medicine (2019) 98:8 www.md-journal.com effects models for each meta-analysis,[32] by using the Cochrane The total number of subjects in the included RCTs was 13,787 Collaboration’s Review Manager V.5.3. (7167 cases and 6,620 controls) and the median number of Assessment of heterogeneity: Heterogeneity between studies participants was 171 (range: 30–1136). In Appendix Supple- was assessed with Cochran Q test[33] and the I2 statistic.[34] For mentary Table 1, http://links.lww.com/MD/C839, we present for interpreting I2, we follow related research criteria[34]: I2 = 0% no each intervention the total number of cases and controls, almost heterogeneity, I2 = 25% low heterogeneity and when I2 exceeds 80% of subjects have participated in 7 of the 21 interventions 50% or 75%, the heterogeneity is considered substantial or (acupuncture, robotic, CIM, tDCS, moxibustion, repetitive task considerable, respectively. training [RTT], and mirror therapy). Estimation of prediction intervals: For the summary random For results that were not presented as SMD in the original effects, we estimated the 95% prediction interval (PI),[35] we meta-analysis, the Cochrane Collaboration’s Review Manager therefore can report the range of effects across study settings, V.5.3 was used to convert the outcomes to SMD to allow visual providing a more complete picture for clinical practice. comparison of the results in a forest plot (presented in Grading the evidence: For each rehabilitation intervention, the Supplementary Material, http://links.lww.com/MD/C839 along data were analyzed qualitatively based on the SMDs of each with forest plots for fixed and random effect sizes for all 55 meta- included meta-analysis. The SMDs were evaluated using forest analyses). plot and graded as small (SMD < 0.5), moderate (SMD 0.5–0.8), and large (SMD > 0.8) effect sizes.[36] The significance of the 3.3. Summary effect size results (P < .05) was judged based on the 95% CIs of the SMDs. Publication bias: Studies published in peer-reviewed journals The evaluation of the level of significance for both random and are much more likely to report statistically significant results than fixed effect calculations, the number of studies, sample size (cases are studies that report a nonsignificant conclusion, especially for and controls), the 95% PI and heterogeneity (I2 and Q test level of smaller studies. Publication bias is assessed checking for significance) are reported in Table 1 for the resulting 55 meta- asymmetry in funnel plots.[37] analyses, grouped by rehabilitation interventions. Thirty of the This study was approved by Institut Guttmann Local Ethics 55 meta-analyses (54%) reported nominally statistically signifi- Committee. cant findings (P < .05) in both fixed and random effect sizes denoted with † in Table 1. Sixteen (77%) of the 21 interventions 2.6. Patient involvement reported statistically significant results in both fixed and random effects in at least 1 meta-analysis. No patients were involved in setting the research question or Bilateral training, imagery training, repetitive transcranial outcome measures, nor were they involved in developing plans magnetic stimulation (rTMS), repetitive peripheral magnetic for design or implementation of the study. stimulation (rPMS), and Nintendo Wii training did not report statistically significant results in any of their meta-analyses. Based 3. Results only on the fixed effects model, 44 meta-analyses (80%) gave nominally statistically significant findings (P < .05). Wii, bilateral 3.1. Literature review training, and rPMS did not report statistically significant results The search retrieved 251 published systematic reviews evaluating in any of their meta-analyses. Thirty-one meta-analyses (57%) rehabilitation interventions. Of these, 88 reviews met the abstract gave a nominally significant findings based on the random effects inclusion criteria and were selected for closer scrutiny. Full texts model. Wii, bilateral training, imagery, rTMS, and rPMS did not of these articles were retrieved and both reviewers performed the report statistically significant results in any of their meta- final selection. Overall, 11 reviews published in Cochrane Library analyses. Only Wii, bilateral training, and rPMS did not report database and 19 published in other academic journals were statistically significant results neither in random nor in fixed included. A PRISMA (Preferred Reporting Items for Systematic effects models. At a stricter threshold of P < .001, 27 (49%) and Reviews and Meta-Analyses) of the study selection process is 13 (24%) meta-analyses produced significant summary results provided in Figure 1. using the fixed and random effects models, respectively. At P < In Appendix Excluded, http://links.lww.com/MD/C839, we 10 6, 23 (42%) and 6 (11%) meta-analyses were significant, present a list of a selection of excluded publications, all dated respectively. In general, the magnitude of the observed summary ≥2015 and most of them addressing interventions already estimates were small to moderate with 30% of the estimates included in the final selection that had to be left out for the yielding an effect size >0.8 as shown in Figure 2. reasons presented in Figure 1. Negative effects summaries were reported in 4 meta-analyses from 3 interventions (acupuncture [2], CIM [1], and rPMS [1]). 3.2. Description of meta-analyses 3.4. Heterogeneity between studies A total of 55 meta-analyses on 21 subacute rehabilitation interventions presented in 30 different publications[1,10,11,38– Q test was significant at P  0.10 in 24 of the 55 meta-analyses 65] (Table 1) involving a total of 314 RCTs were included in this (44%). Eleven meta-analyses (20%) showed no heterogeneity umbrella review. Meta-analyses included a median of 5 studies (I2 = 0%) for several interventions (robotic, Wii, tDCS, CIM, (ranging from 1 to 19). Robotic training and CIM (constraint mCIM, virtual reality, RTT, bilateral, and augmented). Thirteen induced movement) are the interventions with the highest meta-analyses (24%) showed low heterogeneity (I2 < 50%) in number of meta-analyses (10 each), followed by Acupuncture robotic, tDCS, acupuncture, CIM, virtual reality, functional and transcranial direct current stimulation (tDCS) with 6 and 5, electrical stimulation (FES), RTT, mirror therapy, and circuit respectively, then modified CIM and Virtual Reality with 3, interventions. Substantial heterogeneity (50%  I2  75%) is the rest of interventions have been studied in 1 or 2 meta present in 11 (20%) meta-analyses addressing 5 interventions analyses. (robotic, tDCS, CIM, mCIM, Tai Chi). Considerable heteroge- 3 García-Rudolph et al. Medicine (2019) 98:8 Medicine Records idenfied through Idenficaon database searching n =251 Arcles screened aer duplicates Arcles excluded aer removed tle and abstract review (n = 84) Screenin (n =172) Full-text arcles assessed Full-text arcles excluded, Eligibility for eligibility n =88 (n =58) No quantave analysis No RCT No physical therapy No ADL outcome No sub-acute phase No case/control data Studies included in No meta-analysis Included quantave synthesis (meta-analysis) (n =30) Cochrane reviews: 11 Other revies:19 Figure 1. PRISMA Flow diagram of the study selection process.[19] neity (I2 > 75) is shown in 15 (27%) meta-analyses in 8 hydrotherapy, mirror therapy, rPMS, rTMS, tCDS, and virtual interventions (robotic, moxibustion, acupuncture, CIM, imagery, reality. ElectroAcu, Tai Chi, rTMS). Table 2 shows the number of statistical significant studies at the 4 levels of heterogeneity of small (SMD < 0.5), moderate (SMD 0.5–0.8), and large (SMD > 3.6. Prediction intervals 0.8) effect sizes. We calculated 95% PI, as presented in Table 1; the null value was excluded in only 10 meta-analyses (from a total of 55). 3.5. Assessment of methodological quality of included Figure 3 shows the forest plot of the meta-analyses with significant (P < .05) summary fixed and random effects with studies (AMSTAR) random effects 95% PI excluding null value, along with the Twenty-nine (94%) of the included publications are rated high number of studies (treatments and controls) implemented in quality (mean value 9.61, see Appendix Supplementary Table 3, forest plot R package.[66] http://links.lww.com/MD/C839). All meta-analyses provided an In Appendix Supplementary Figure 1, http://links.lww.com/ “a priori” design, performed a literature search in at least 2 MD/C839, we included the funnel plots (all of them presenting electronic databases, reported duplicate study selection and made visually acceptable symmetry) for the 4 meta-analyses with more it possible to replicate the literature search, but 20% did not than 1 study from Figure 3 to visually assess symmetry (these report the inclusion of grey literature, 7% did not report conflict funnel plots are extracted from Supplementary Material, http:// of interest, and 20% did not assess the scientific quality of the links.lww.com/MD/C839, where we present all funnel plots for included studies. Appendix Supplementary Table 4, http://links. all selected meta-analyses). lww.com/MD/C839, shows top rated interventions: circuit, FES, For publications with P < .05 and number of studies ≥9, 4 García-Rudolph et al. Medicine (2019) 98:8 www.md-journal.com Table 1 Summary effect calculations, studies, sample size (cases and controls), 95% prediction interval (PI) and heterogeneity. Studies Fixed effects Random effects ∗ Intervention Publication (cases/controls) (P) (P) PI (95%) I 2 (P ) ∗ Robotic Mehrholz et al (2015)[38] 6 (94/101) 0.49 [0.20, 0.78] 0.49 [0.20, 0.78] [0.02, 0.96] 0 (.55) (.001) (.001) Mehrholz et al (2015)[38] 12 (272/ 0.39 [0.21, 0.58] 0.32 [ 0.05, 0.68] [ 1.16, 1.8] 73 (<.0001) 250) (<.00001) (.09) Norouzi et al (2012)[39] 3 (55/51) 0.96 [0.55, 1.38] 0.81 [ 0.07, 1.70] [ 4.24, 5.86] 77 (.01) (<.00001) (.07) Norouzi et al (2012)[39] 3 (32/30) 0.02 [ 0.49, 0.53] 0.00 [ 0.62, 0.62] [ 2.77, 2.77] 31 (.23) (.95) (1.00) Veerbeek et al (2017)[1] 17 (234/ 0.33 [0.13, 0.53] 0.27 [ 0.05, 0.59] [ 0.98, 1.52] 56 (.002) 193) (.001) (.09) Veerbeek et al (2017)[1] 14 (184/ 0.35 [0.12, 0.58] 0.24 [ 0.17, 0.65] [ 1.37, 1.85] 64 (.0005) 146) (.003) (.25) Veerbeek et al (2017)[1] 17 (234/ 0.33 [0.13, 0.33] 0.27 [ 0.05, 0.59] [ 0.98, 1.52] 56 (.002) 193) (.001) (.09) Veerbeek et al (2017)[1] 13 (185/ 0.34 [0.11, 0.57] 0.22 [ 0.20, 0.65] [ 1.45, 1.89] 67 (.0003) 146) (.004) (.30) Veerbeek et al (2017)[1] 12 (158/ 0.08 [ 0.16, 0.33] (.51) 0.04 [ 0.27, 0.35] [ 0.84, 0.92] 32 (.13) 121) (.80) † Veerbeek et al (2017)[1] 5 (76/72) 0.82 [0.48, 1.17] 0.85 [0.32, 1.38] [ 0.98, 2.68] 54 (.07) (<.00001) (.002) Sling Chen et al (2016)[40] 1 (40/40) 0.24 [ 0.20, 0.68] 0.24 [ 0.20, 0.68] NA NA (.29) (.29) † Chen et al (2016)[40] 1 (48/48) 1.12 [0.69, 1.55] 1.12 [0.69, 1.55] NA NA (.00001) (.00001) Wii Cheok et al (2015)[41] 2 (20/17) 0.27 [ 0.38, 0.93] 0.27 [ 0.38, 0.93] [ 8.23, 8.77] 0 (.75) (.41) (0.41) tDCS Elsner et al (2016)[10] 9 (232/ 0.24 [0.03, 0.44] 0.24 [0.03, 0.44] [ 0.05, 0.53] 0 (.52) 164) (.02) (.02) Triccas et al (2015)[42] 5 (129/71) 0.24 [ 0.06, 0.54] 0.32 [ 0.08, 0.72] [ 0.85, 1.49] 33 (.20) (.11) (.12) Triccas et al (2015)[42] 3 (102/76) 0.19 [ 0.12, 0.50] 0.54 [ 0.27, 1.34] [ 3.93, 5.01] 70 (.04) (.23) (.19) † Triccas et al (2015)[42] 3 (65/65) 0.38 [0.03, 0.73] 0.38 [0.02, 0.74] [ 0.92, 1.68] 5 (.35) (.03) (.04) † Banchetti et al (2017)[43] 3 (37/52) 0.51 [0.07, 0.95] 0.51 [0.07, 0.95] [ 0.88, 1.9] 0 (.39) (.02) (.02) Moxibustion† Han et al (2018)[44] 14 (459/ 0.90 [0.76, 1.04] 1.07 [0.64, 1.50] [ 0.95, 3.09] 88 (<.00001) 434) (<.00001) (<.00001) Acupuncture Lee et al (2013)[45] 2 (75/75) 0.17 [ 0.49, 0.15] 0.16 [ 0.50, 0.18] [ 5.26, 4.94] 10 (.29) (.31) (.34) † Lee et al (2013)[45] 2 (62/61) 0.78 [0.41, 1.15] 0.78 [0.40, 1.17] [ 4.83, 6.39] 7 (.30) (<.00001) (<.00001) Lee et al (2013)[38] 2 (75/75) 1.29 [0.92, 1.66] 1.34 [ 0.56, 3.25] (.17) [ 40.98, 43.66] 96 (<.00001) (<.00001) † Li et al (2014)[46] 12 (574/ 0.81 [0.68, 0.94] 1.36 [0.71, 2.01] [ 1.56, 4.28] 95 (<.00001) 562) (<.00001) (<.00001) † Yang et al (2016)[47] 9 (309/ 1.20 [1.01, 1.30] 1.27 [0.54, 2.00] (.0007) [ 1.88, 4.42] 94 (<.00001) 307) (<.00001) † Peng et al (2018)[48] 6 (347/ 0.97 [0.81, 1.14] 1.31 [0.57, 2.05] (.005) [ 1.76, 4.38] 76.7 (<.0001) 345) (<.00001) CIM Corbetta et al (2010)[49] 9 (151/ 0.18 [ 0.06, 0.42] 0.21 [ 0.08, 0.50] [ 0.55, 0.97] 31 (.18) 127) (.14) (.16) Peurala et al (2012)[50] 2 (116/ 1.27 [0.98, 1.55] 0.98 [0.01, 1.94] (.05) [ 19.58, 21.54] 83 (.02) 115) (<.00001) Peurala et al (2012)[50] 2 (29/27) 1.78 [1.10, 2.47] 3.75 [ 1.05, 8.55] (.13) [ 101.79, 109.3] 91 (.0007) (<.00001) Peurala et al (2012)[50] 5 (77/94) 0.59 [0.27, 0.91] 0.59 [0.17, 1.00] [ 0.69, 1.87] 41 (.15) (.0003) (.006) Peurala et al (2012)[50] 2 (116/ 1.16 [0.87, 1.44] 0.87 [-0.07, 1.81] (.07) [ 19.01, 20.75] 82 (.02) 115) (<.00001) † Peurala et al (2012)[50] 1 (20/20) 0.82 [0.17, 1.46] 0.82 [0.17, 1.46] NA NA (.01) (.01) (continued ) 5 García-Rudolph et al. Medicine (2019) 98:8 Medicine Table 1 (continued). Studies Fixed effects Random effects ∗ Intervention Publication (cases/controls) (P) (P) PI (95%) I 2 (P ) † Peurala et al (2012)[50] 5 (77/94) 0.67 [0.35, 0.99] 0.68 [0.18, 1.18] (.007) [ 1.05, 2.41] 58 (.05) (<.00001) Peurala et al (2012)[50] 1 (18/12) 0.53 [ 1.27, 0.21] 0.53 [ 1.27, 0.21] NA NA (.16) (.16) Peurala et al (2012)[50] 4 (69/46) 0.07 [ 0.33, 0.47] 0.30 [ 1.51, 0.91] (.63) [ 5.74, 5.14] 88 (<.0001) (<.00001) † Peurala et al (2012)[50] 4 (51/71) 0.50 [0.13, 0.87] 0.50 [0.13, 0.87] [ 2.47, 3.47] 0 (.91) (.008) (.008) mCIM† Shi et al (2011)[51] 3 (45/43) 0.45 [0.03, 0.88] 0.45 [0.03, 0.88] [ 0.89, 1.79] 0 (.80) (.04) (.04) † Shi et al (2011)[51] 6 (87/86) 0.73 [0.41, 1.05] 0.79 [0.22, 1.37] (.007) [ 1.17, 2.75] 64 (.02) (<.00001) † Shi et al (2011)[51] 6 (87/86) 0.81 [0.49, 1.13] 0.86 [0.36, 1.37] [ 0.77, 2.49] 54 (.05) (<.00001) (.0007) VR† Laver et al (2012)[52] 3 (60/41) 0.81 [0.39, 1.22] 0.81 [0.39, 1.22] [ 0.5, 2.12] 0 (.55) (.0002) (.0002) † Laver et al (2015)[53] 8 (136/ 0.43 [0.18, 0.69] 0.43 [0.18, 0.69] [0.06, 0.8] 2 (.41) 117) (.0009) (.0001) † Laver et al (2015)[53] 8 (80/73) 0.44 [0.11, 0.76] 0.44 [0.11, 0.76] [ 0.03, 0.91] 0 (.57) (.009) (.009) FES† Eraifej et al (2017)[54] 5 (32/33) 1.23 [0.65, 1.80] 1.24 [0.46, 2.03] (.002) [ 1.21, 3.69] 43 (.14) (<.00001) RTT† French et al (2010)[55] 5 (175/ 0.29 [0.07, 0.51] 0.28 [0.01, 0.55] [ 0.49, 1.05] 30 (.22) 150) (.01) (.04) † French et al (2016)[56] 9 (273/ 0.28 [0.10, 0.45] 0.28 [0.10, 0.45] [0.03, 0.53] 0 (.47) 254) (.002) (.002) Bilateral Lee et al (2017)[57] 2 (61/56) 0.26 [ 0.10, 0.63] 0.26 [ 0.10, 0.63] [ 4.48, 5] 0 (.54) (.16) (.16) Augmented† Veerbeek et al (2011)[58] 2 (66/72) 0.54 [0.20, 0.88] 0.54 [0.20, 0.88] [ 3.87, 4.95] 0 (.60) (.002) (.002) Imagery Fernandes et al (2017)[59] 5 (81/81) 0.50 [0.17, 0.83] 0.67 [ 0.20, 1.54] [ 2.89, 4.23] 85 (<.0001) (.003) (.13) ElectroAcu† Cai et al (2017)[60] 7 (262/ 0.73 [0.55, 0.91] 0.80 [0.41, 1.20] [ 0.72, 2.32] 78 (<.00001) 260) (<.00001) (<.00001) Tai Chi† Lyu et al (2018)[61] 7 (199/ 1.29 [1.06, 1.52] 1.44 [0.78, 2.09] [ 1.22, 4.1] 88 (<.00001) 192) (<.00001) (.0001) † Lyu et al (2018)[61] 2 (81/85) 0.92 [0.60, 1.25] 0.97 [0.47, 1.47] [ 8.38, 10.32] 54 (.14) (<.00001) (.0001) Hydrotherapy† Mehrholz et al (2011)[62] 1 (16/15) 1.90 [1.03, 2.76] 1.90 [1.03, 2.76] NA NA (<.00001) (<.00001) Mirror therapy† Thieme et al (2018)[63] 19 (333/ 0.46 [0.30, 0.63] 0.48 [0.30, 0.65] [0.07, 0.89] 15 (.27) 289) (<.00001) (<.00001) rTMS Hao et al (2013)[64] 2 (93/90) 0.91 [0.58, 1.24] 1.90 [ 1.04, 4.84] (.21) [ 63.68, 67.48] 98 (<.00001) (<.00001) Circuit† English et al (2017)[65] 2 (147/ 0.32 [0.09, 0.55] 0.30 [0.03, 0.58] [ 4.08, 4.68] 17 (.27) 149) (.006) (.03) rPMS Momosaki et al (2017)[11] 1 (31/32) 0.11 [ 0.60, 0.38] 0.11 [ 0.60, 0.38] [0.02, 0.96] NA (.66) (.66) Several lines for the same publication mean different meta-analyses reported in the same publication (e.g., in Verbeek 2017 RT-UL groups were divided into 5 subgroups: shoulder/elbow, whole arm [shoulder/ elbow/wrist and/or hand], elbow, elbow/wrist/hand, wrist/hand and hand). FES = functional electrical stimulation; mCIM = modified constraint induced movement; NA = not applicable; tDCS = transcranial direct current stimulation; VR = virtual reality; RTT = repetitive task training; rTMS = repetitive transcranial magnetic stimulation; rPMS = repetitive peripheral magnetic stimulation; wii = Nintendo wii. ∗ P-value of the Q test. † Statistically significant findings (P < .05).  we performed regression analysis with meta R package[67] on is explained by the number of participants in the experimental the number of participants in experimental groups and the year group. of publication covariates, as in similar previous umbrella  we performed sensitivity analysis with meta R package[67] to reviews.[68] We identified 2 studies [47,63] with P-value <.05 for explain high heterogeneity (publications with P < .05 and the total participants in experimental group covariate. As number of studies ≥9). Therefore, we analyzed Han et al[44] shown in Appendix Supplementary Table 5, http://links.lww. (89%), Li et al[46] (95%), and Yang et al[47] (94%). We com/MD/C839, 2.28% and 8.21%, respectively, of effect size obtained our best results for Li et al,[46] when omitting Wu 6 García-Rudolph et al. Medicine (2019) 98:8 www.md-journal.com 4 3,5 3 Random effects summary 2,5 2 1,5 1 0,5 0 -0,5 0 2 4 6 8 10 12 -1 1/Standard Error Figure 2. Association of the meta-analysis summary random effects estimate with the inverse of the standard error. study (2011), heterogeneity declines from 95% to 89%. In the interventions (e.g., virtual reality technologies). But as shown in other 2 publications, best reductions are only 1% to 2%. page 83 of Supplementary Material, http://links.lww.com/MD/  we performed subgroup analysis (by using the Cochrane C839, no significant subgroup differences were identified. We Collaboration’s Review Manager V.5.3) on those publications proceeded similarly with other publications[44,47] with no with P < .05 and number of studies ≥9 grouping by year of significant differences either. publication. For example, in Thieme et al,[63] we performed Details of meta-regression, sensitivity, and subgroups analysis subgroup analysis with 3 groups: are presented in Supplementary Material, http://links.lww.com/ ^ Group 1: 2008 to 2012 (6 studies) MD/C839. ^ Group 2: 2013 to 2014 (7 studies) In Appendix Supplementary Figure 2, http://links.lww.com/ ^ Group 3: 2015 to 2018 (6 studies) MD/C839, we included a work design diagram showing our whole Differences could be identified within the 3 periods because of analysis process, taking as starting point the outcome of PRISMA technology evolution in the administration of mirror therapy flow diagram of the study selection process presented in Figure 1. Table 2 Summary of the number of studies for small, moderate, and large effect sizes grouped by heterogeneity level for the different intervention. Intervention No Heterogeneity 0% Low < 50% Substantial 50-75% Considerable >75% Robot assisted 6 5 tDCS 9 3 5 Moxibustion 14 Acupuncture 2 27 CIM 4 5 5 mCIM 3 6 6 Virtual Reality 3 8 8 FES 5 RTT 9 5 Augmented 2 Electroacu 7 Tai Chi 2 7 Mirror therapy 19 Circuit 2 SMD > 0.8 SMD 0.5 - 0.8 SMD < 0.5 No effects 7 García-Rudolph et al. Medicine (2019) 98:8 Medicine Figure 3. Forest plot for meta-analysis with significant (P < .05) summary fixed and random effects with random effects 95% prediction interval excluding null value. 4. Discussion to variability in stroke recovery. Genetic variation in any of these It is recommended [69] that the 95% PI should be routinely components might influence each individual’s capacity for brain reported to allow more informative inferences in meta-analyses. plasticity and could explain the variability in motor rehabilitation It presents the expected range of true effects in similar studies, efficacy. therefore in 95% of cases the true effect of a new study (assessing the impact of a physical intervention in ADLs) will fall within the 4.1. Study limitations PI values. In our case, most PIs contain null values, leading us to We did not publish a protocol for this study. We only included consider that although therapists’ interventions are effective on systematic reviews with meta-analysis (we took this approach as average to increase performance in ADLs most of them may not in recent reviews related to health).[75] be effective because the null value was excluded in only 10 (with 5 of them NA) of the 55 meta-analyses. 5. Conclusion Considering recent umbrella reviews in medical field (e.g., Cancer or Parkinson disease)[70] “highly convincing” evidence Moxibustion, acupuncture, and Tai Chi not only show large requires >1000 cases, P < 10–6 by random effects, 95% PI SMD values and large number of participants, but also the excluding the null and not large heterogeneity (I2 < 50%). highest values of heterogeneity (I2 > 75), when interventions are “Suggestive” evidence requires >1000 cases and P < .001 by grouped according to the four levels of heterogeneity of small random effects. Only acupuncture meta-analysis reached such (SMD < 0.5), moderate (SMD 0.5–0.8), and large (SMD > 0.8) thresholds, leaving the rest of studies included in this work at effect sizes considering the significance of the results (P < .05) the “weak” level, pointing out the need of new RCTs assessing based on the 95% CIs of the SMDs for both fixed and random ADLs. effects. After full-text review of the 30 included studies, we did not find Robot assisted, virtual reality, tDCS, and RTT present small mentions to inflammatory process. Clinical outcomes after stroke and moderate effect sizes but without heterogeneity (I2 = 0). The are highly variable, and reasons for these variations are often total number of participants in the included RCTs was 13,787 unexplained.[71] Recovery after ischemic or hemorrhagic stroke and the median number was 171 with almost 80% of subjects begins immediately after acute onset, and several different levels participating in only 7 of the 21 interventions (acupuncture, of biological responses are involved. Genetic factors also robotic, CIM, tDCS, moxibustion, RTT, and mirror therapy), influence many different aspects of brain function and repair, meaning that 80% of participants are recruited in only 30% of as well as recurrent stroke risk and response to interventions, and interventions. Besides, when categorizing our evidence according can account for unexplained variation in stroke recovery.[71] Up to state-of-the art thresholds only weak levels of evidence are to 30% of ischemic strokes remain unexplained after thorough reached in almost all of the included studies. investigation.[72] Cryptogenic stroke is more common in patients Acupuncture as a complementary therapy has increased with stroke occurring at a young age, defined as age 55 or less. worldwide and has become widely applied to stroke rehabilita- For example, it has been reported that approximately 1% to 2% tion.[76] Similarly, Tai Chi has been applied in stroke rehabilita- of young stroke patients are demonstrated with Fabry disease,[72] tion for over 10 years worldwide.[61] Moxibustion has been less a rare inherited disorder of the metabolism, associated with renal, studied, being noninvasive it might be more applicable in other cardiac, and cerebrovascular complications.[73,74] With such a cultures.[44] Our results encourage further RCTs into them. multitude of molecular events being related to recovery, not Future research could analyze the excess significance bias, surprisingly a number of genes have been suggested as important whether observed number of studies with statistically significant 8 García-Rudolph et al. Medicine (2019) 98:8 www.md-journal.com results is different from the expected number of them[77] [14] Tuttolomondo A, Di Sciacca R, Di Raimondo D, et al. Effects of clinical and laboratory variables and of pretreatment with cardiovascular drugs and extend subgroup analyses considering gender and/or age in acute ischaemic stroke: a retrospective chart review from the GIFA differences. study. Int J Cardiol 2011;151:318–22. [15] Emsley HC, Hopkins SJ. Acute ischaemic stroke and infection: recent and emerging concepts. Lancet Neurol 2008;7:341–53. Acknowledgments [16] McColl BW, Allan SM, Rothwell NJ. Systemic infection, inflammation The authors thank Olga Araujo from Institute Guttmann’s and acute ischemic stroke. Neuroscience 2009;158:1049–61. [17] Di Raimondo D, Tuttolomondo A, Buttà C, et al. Effects of ACE- Centre de Documentació en Neurorehabilitació Santi Beso inhibitors and angiotensin receptor blockers on inflammation. Curr Arnalot for her continuous support in publications retrieval. Pharm Des 2012;18:4385–413. [18] Elkind MS, Cheng J, Rundek T, et al. Leukocyte count predicts outcome after ischemic stroke: the Northern Manhattan Stroke Study. J Stroke Author contributions Cerebrovasc Dis 2004;13:220–7. [19] Aromataris, E, Fernandez, R, Godfrey, C, et al. Methodology for JBI Conceptualization: Eloy Opisso Salleras, Josep Maria Tormos. umbrella reviews. Joanna Briggs Institute Reviewers’ Manual: 2014 Data curation: Alejandro Garcia-Rudolph, David Sanchez- edition/Supplement ( 1-34). Australia: The Joanna Briggs Institute. Pinsach. [20] Hamilton BB, Laughlin JA, Fiedler RC, et al. Interrater reliability of the Formal analysis: Alejandro Garcia-Rudolph. 7-level functional independence measure (FIM). Scand J Rehabil Med Funding acquisition: Josep Maria Tormos. 1994;26:115–9. [21] Wade DT, Hewer RL. Functional abilities after stroke: measurement, Investigation: Alejandro Garcia-Rudolph. natural history and prognosis. J Neurol Neurosurg Psychiatry 1987; Methodology: Alejandro Garcia-Rudolph. 50:177–82. Project administration: Eloy Opisso Salleras. [22] Bonita R, Beaglehole R. Modification of Rankin scale: recovery of motor Resources: Eloy Opisso Salleras. function after stroke. Stroke 1988;19:1497–500. [23] Schuling J, De Haan R, Limburg M, et al. The Frenchay activities index: Software: Alejandro Garcia-Rudolph, David Sanchez-Pinsach. assessment of functional status in stroke patients. Stroke 1993;24: Supervision: Josep Maria Tormos. 1173–7. Writing – original draft: Alejandro Garcia-Rudolph. [24] Whiting S, Lincoln N. An ADL assessment for stroke patients. Br J Occup Writing – review & editing: Alejandro Garcia-Rudolph, Eloy Ther 1980;43:44–6. Opisso Salleras, Josep Maria Tormos. [25] Hartigan I. A comparative review of the Katz ADL and the Barthel Index in assessing the activities of daily living of older people. Int J Older People Nurs 2007;2:204–12. References [26] Uswatte G, Taub E, Morris D, et al. Reliability and validity of the upper- extremity motor activity log-14 for measuring real-world arm use. Stroke [1] Veerbeek JM, Langbroek-Amersfoort AC, van Wegen EE, et al. Effects of 2005;36:2493–6. robot-assisted therapy for the upper limb after stroke. Neurorehabil [27] Shah S, Vanclay F, Cooper B. Improving the sensitivity of the Barthel Neural Repair 2017;31:107–21. Index for stroke rehabilitation. J Clin Epidemiol 1989;42:703–9. [2] Wattchow KA, McDonnell MN, Hillier SL. Rehabilitation interventions [28] Higgins JPT, Green S. Cochrane handbook for systematic reviews for upper limb function in the first four weeks following stroke: a of interventions version 5.1.0 [updated March 2011], 2011. www. systematic review and meta-analysis of the evidence. Arch Phys Med handbook.cochrane.org. Accessed April 2018. Rehabil 2018;99:367–82. [29] Hatano S. Experience from a multicentre stroke register: a preliminary [3] The Burden of Stroke in Europe Report. Overview of stroke burden and report. Bull World Health Organ 1976;54:541–53. care in each EU and SAFE member country (2017). ISBN 978-1-5272- [30] Peurala SH, Karttunen AH, Sjögren T, et al. Evidence for the effectiveness 0857-5 Stroke Alliance for Europe -Stroke Association House 240 City of walking training on walking and self-care after stroke: a systematic Road, London. review and meta-analysis of randomized controlled trials. J Rehabil Med [4] Benjamin E. Heart disease and stroke statistics—2017 update a report 2014;46:387–99. from the American Heart Association. Circulation 2017;135:e146–603. [31] Shea BJ, Bouter LM, Peterson J, et al. External validation of a [5] Langhorne P, Legg L. Evidence behind stroke rehabilitation. J Neurol measurement tool to assess systematic reviews (AMSTAR). PLoS One Neurosurg Psychiatry 2003;74(Suppl 4):iv18–21. 2007;2:1–5. [6] Langhorne P, Bernhardt J, Kwakkel G. Stroke rehabilitation. Lancet [32] DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin 2011;377:1693–702. Trials 1986;7:177–88. [7] Lee SH (ed), Stroke Revisited: Diagnosis and Treatment of Ischemic [33] Cochran WG. The combination of estimates from different experiments. Stroke, Stroke Revisited, Rehabilitation in Sub-acute and Chronic Stage Biometrics 1954;10:101–29. After Stroke Han-Young Jung (2011). [34] Higgins JPT, Thompson SG, Deeks JJ, et al. Measuring inconsistency in [8] Pasanen T, Tolvanen S, Heinonen A, et al. Exercise therapy for functional meta-analyses. Br Med J 2003;327:557–60. capacity in chronic diseases: an overview of meta-analyses of randomised [35] Riley RD, Higgins JP, Deeks JJ. Interpretation of random effects meta- controlled trials. Br J Sports Med 2017;51:1459–65. analyses. BMJ 2011;342:d549. [9] Veerbeek JM, van Wegen E, van Peppen R, et al. What is the evidence for [36] Cohen J. Statistical Power Analysis for the Behavioral Sciences. London: physical therapy poststroke? A systematic review and meta-analysis. Academic Press; 1988. PLoS One 2014;9:e87987. [37] Mlinaric´ A, Horvat M,  Supak Smolčic´ V. Dealing with the positive [10] Elsner B, Kugler J, Pohl M, et al. Transcranial direct current stimulation publication bias: why you should really publish your negative results. (tDCS) for improving activities of daily living, and physical and cognitive Biochem Med 2017;27: functioning, in people after stroke. Cochrane Database of Syst Rev 2016, [38] Mehrholz J, Pohl M, Platz T, et al. Electromechanical and robot-assisted Issue 3. Art. No.: CD009645. DOI: 10.1002/14651858. CD009645. armtraining for improving activities of daily living, armfunction, and pub3. armmuscle strength after stroke. Cochrane Database of Systematic [11] Momosaki R, Yamada N, Ota E, et al. Repetitive peripheral magnetic Reviews 2015, Issue 11. Art. No.: CD006876. doi: 10.1002/14651858. stimulation for activities of daily living and functional ability in people CD006876.pub4. after stroke. Cochrane Database of Syst Rev 2017, Issue 6. Art. No.: [39] Norouzi-Gheidari N, Archambault PS, Fung J. Effects of robot-assisted CD011968. doi: 10.1002/14651858.CD011968.pub2. therapy on stroke rehabilitation in upper limbs: systematic review and [12] De Simoni MG, Milia P, Barba M, et al. The inflammatory response in meta-analysis of the literature. J Rehabil Res Dev 2012;49:479–96. cerebral ischemia: focus on cytokines in stroke patients. Clin Exp [40] Chen L, Chen J, Peng Q, et al. Effect of sling exercise training on balance Hypertens 2002;24:535–42. in patients with stroke: a meta-analysis. PLoS One 2016;11:e0163351. [13] Hedley CA, Emsley, Pippa J. Tyrrell Inflammation and Infection in [41] Cheok G, Tan DM, Low A, et al. Is nintendo wii an effective intervention Clinical Stroke Journal of Cerebral Blood Flow & Metabolism 22:1399- for individuals with stroke? A systematic review and meta-analysis. 1419.© 2002 The International Society for Cerebral Blood Flow and Journal of the American Medical Directors Association 2015;1611: Metabolism. 923–32. 9 García-Rudolph et al. Medicine (2019) 98:8 Medicine [42] Tedesco Triccas L, Burridge JH, Hughes AM, et al. Multiple sessions of [59] Fernandez Guerra Z, Lucchetti ALG, Lucchetti G. Motor imagery transcranial direct current stimulation and upper extremity rehabilitation training after stroke: a systematic review and meta-analysis of in stroke: a review and meta-analysis. Clin Neurophysiol 2016;127: randomized controlled trials. J Neurol Phys Ther 2017;41:205–14. 946–55. [60] Cai, Zhang CS, Liu S et al. Electroacupuncture for poststroke spasticity: a [43] Banchetti PA, Marini C. Efficacy of transcranial direct current systematic review and meta-analysis. Arch Phys Med Rehabil. 2017; 98(12): stimulation as support for neurorehabilitation therapy: a metaanalysis 2578-2589.e4. doi: 10.1016/j.apmr.2017.03.023. Epub 2017 Apr 25. of available studies. Int J Phys Med Rehabil 2017;5:417. [61] Lyu D, Lyu X, Zhang Y, et al. Tai Chi for stroke rehabilitation: a [44] Han C-H, Ma JN, An N, et al. Moxibustion for stroke: systematic systematic review and meta-analysis of randomized controlled trials. review, meta-analysis, and GRADE-based recommendations. Eur J Front Physiol 2018;9:983. Integrative Med 2018;20:115–25. [62] Mehrholz J, Kugler J, Pohl M. Water-based exercises for improving [45] Lee S-J, Shina B-C, Lee MS, et al. Scalp acupuncture for stroke recovery: a activities of daily living after stroke. Cochrane Database of Systematic systematic review and meta-analysis of randomized controlled trials. Eur Reviews 2011, Issue 1. Art. No.: CD008186. doi: 10.1002/14651858. J Integr Med 2012;5:87–99. CD008186.pub2. [46] Li L, Zhang H, Meng S-q, et al. An updated meta-analysis of the efficacy [63] Thieme H, Morkisch N, Mehrholz J, Pohl M, Behrens J, Borgetto B, and safety of acupuncture treatment for cerebral infarction. PLoS One Dohle C. Mirror therapy for improving motor function after stroke. 2014;9:e114057. Cochrane Database of Systematic Reviews 2018, Issue 7. Art. No.: [47] Yang A, Wu HM, Tang JL et al. Acupuncture for stroke rehabilitation. CD008449. doi: 10.1002/14651858.CD008449.pub3. Cochrane Database of Systematic Reviews 2016, Issue 8. Art. No.: [64] Hao Z,Wang D, Zeng Y, Liu M. Repetitive transcranial magnetic CD004131. doi: 10.1002/14651858.CD004131.pub3. stimulation for improving function after stroke. Cochrane Database of [48] Peng L, Zhang C, Zhou L, et al. Traditional manual acupuncture Systematic Reviews 2013, Issue 5. Art. No.: CD008862. doi: 10.1002/ combined with rehabilitation therapy for shoulder hand syndrome after 14651858.CD008862.pub2. stroke within the Chinese healthcare system: a systematic review and [65] English C, Hillier SL, Lynch EA. Circuit class therapy for improving meta-analysis. Clin Rehabil 2018;32:429–39. mobility after stroke. Cochrane Database of Systematic Reviews 2017, [49] Corbetta D, Sirtori V, Moja L, et al. Constraint-induced movement Issue 6. Art. No.: CD007513. doi: 10.1002/14651858.CD007513.pub3. therapy in stroke patients: systematic review and meta-analysis. Eur J [66] Gordon M, Lumley T. Advanced Forest Plot Using ‘grid’ Graphics. Phys Rehabil Med 2010;46:537–44. Available at: https://cran.r-project.org/web/packages/forestplot/forest [50] Peurala SH, Kantanen MP, Sjögren T, et al. Effectiveness of constraint- plot.pdf. Accessed August 20, 2018. induced movement therapy on activity and participation after stroke: a [67] Schwarzer G. meta: An R Package for Meta-Analysis Available at: systematic review and meta-analysis of randomized controlled trials. Clin https://cran.rproject.org/web/packages/meta/meta.pdf. Accessed August Rehabil 2012;26:209–23. 20, 2018. [51] Shi YX, Tian JH, Yang KH, et al. Modified constraint-induced [68] O’Sullivan JW, Muntinga T, Grigg S, et al. Prevalence and outcomes of movement therapy versus traditional rehabilitation in patients with incidental imaging findings: umbrella review. BMJ 2018;361:k2387. upper-extremity dysfunction after stroke: a systematic review and meta- [69] IntHout J, J.P.A. , Rovers MM, et al. Plea for routinely presenting analysis. Arch Phys Med Rehabil 2011;92:972–82. prediction intervals in meta-analysis BMJ Open 2016;6:e010247. [52] Laver K, George S, Thomas S, et al. Cochrane review: virtual [70] Bellou V, Belbasis L, Tzoulaki I, et al. Environmental risk factors and reality for stroke rehabilitation. Eur J Phys Rehabil Med 2012;48: Parkinson’s disease: an umbrella review of meta-analyses. Parkinsonism 523–30. Relat Disord 2016;23:1–9. [53] Laver KE, George S, Thomas S et a Virtual reality for stroke [71] Pearson-Fuhrhop KM, Burke E, Cramer SC. The influence of genetic rehabilitation. Cochrane Database of Systematic Reviews 2015, Issue factors on brain plasticity and recovery after neural injury. Curr Opin 2. Art. No.: CD008349. doi: 10.1002/14651858.CD008349.pub3. Neurol 2012;25:682–8. [54] Eraifej J, Clark W, France B, et al. Effectiveness of upper limb functional [72] Feldt-Rasmussen U. Fabry disease and early stroke. Stroke Res Treat electrical stimulation after stroke for the improvement of activities of 2011;2011:615218. daily living and motor function: a systematic review and meta-analysis. [73] Tuttolomondo A, Pecoraro R, Simonetta I, et al. Neurological Syst Rev 2017;6:40. complications of Anderson-Fabry disease. Curr Pharm Des 2013;19: [55] French B, Thomas LH, Leathley M, et al. Does repetitive task training 6014–30. Review. improve functional activity after stroke? A Cochrane systematic review [74] Tuttolomondo A, Pecoraro R, Simonetta I, et al. Anderson-Fabry disease: and meta-analysis. J Rehabilit Med 2010;42:9–14. a multiorgan disease. Curr Pharm Des 2013;19:5974–96. [56] French B, Thomas LH, Coupe J et al. Repetitive task training for [75] Dinu M, Pagliai G, Casini A, et al. Mediterranean diet and multiple improving functional ability after stroke. Cochrane Database of health outcomes: an umbrella review of meta-analyses of observational Systematic Reviews 2016, Issue 11. Art. No.: CD006073. doi: studies and randomised trials. Eur J Clin Nutr 2018;72:30–43. 10.1002/14651858.CD006073.pub3. [76] Gu J, Wang Q, Wang XG, et al. Assessment of registration information [57] Lee Y, Kim MY, Park JH, et al. Comparison of the effects of bilateral and on methodological design of acupuncture RCTs: a review of 453 unilateral training after stroke: a meta-analysis. NeuroRehabilitation registration records retrieved from WHO International Clinical Trials 2017;40:301–13. Registry Platform. Evid Based Complement Alternat Med 2014;2014: [58] Veerbeek JM, Koolstra M, Ket JC, et al. Effects of augmented exercise 614850. therapy on outcome of gait and gait-related activities in the first 6 months [77] Ioannidis JP, Trikalinos TA. An exploratory test for an excess of after stroke: a meta-analysis. Stroke 2011;42:3311–5. significant findings. Clin Trials 2007;4:245–53. 10