Background: Stroke survivors often experience impaired mobility and physical functions. Tai Chi and Qigong have been shown to have physical and psychological benefits for stroke patients. Purpose: To summarize the evidence on Tai Chi and Qigong for improving mobility in stroke survivors, specifically the ability to walk, dynamic balance, and activities of daily living (ADL). Methods: Independent searches of 16 electronic databases in English, Korean, and Chinese from their inception until December 2021 were conducted by two research teams. Methodological quality was assessed using Cochrane's risk of bias tool 2.0. Comprehensive Meta-Analysis 3.0 software was used to calculate effect sizes with subgroup analysis and to assess heterogeneity and publication bias. Results: The meta-analysis included 27 randomized trials (18 with Tai Chi and 9 with Qigong) on stroke survivors (N = 1,919). None of the studies were considered at high risk of bias, about 70% had some concerns, and 30% were considered low risk. Meta-analysis of 27 randomized controlled trials with random-effects models indicated that Tai Chi and Qigong effectively improved mobility, specifically on the ability to walk (Hedges'g = 0.81), dynamic balance (Hedges'g = 1.04), and ADL (Hedges'g = 0.43). The effects of Tai Chi and Qigong were significant for short-term and long-term programs (Hedges'g 0.91 vs. 0.75), and when compared with active controls and no treatment group (Hedges'g 0.81 vs. 0.73). Conclusion: Tai Chi and Qigong performed for 12 weeks or less were effective in improving the mobility of stroke survivors. Further studies are warranted to assess whether Tai Chi and Qigong work best as an adjunct to rehabilitation, an effective alternative to rehabilitation or as a maintenance strategy, and whether the results could be further optimized by assessing different schools of Tai Chi and Qigong, different types of stroke patients, and different points in the post-stroke recovery process. PROSPERO registration number: This study has been registered on the UK National Institute for Health Research (
Stroke is a major worldwide cause of death and disability. Despite the declining stroke incidence, the aging population and accumulating risk factors contribute to an increased lifetime risk of stroke [[
Complex factors prevent stroke survivors from successfully surviving, and most risk factors are lifestyle-related and largely modifiable [[
Tai Chi and Qigong (TCQ) are grounded in the principles of traditional Chinese medicine and have been described as equivalent in terms of essential forms and principles. Qigong is considered the ancient root of all traditional Chinese medicine practices [[
Recovery of mobility function is considered one of the main goals of stroke rehabilitation [[
Therefore, this study aimed to use a meta-analysis combined with recent clinical trials not included in previous reviews written in multi-languages. In addition, subgroup analyses were conducted based on outcome measures, duration of intervention, and control conditions to identify TCQ as an effective alternative intervention among stroke survivors.
This review was registered in the PROSPERO database (PROSPERO Register code: CRD42020220277,
This systematic review and meta-analysis included randomized controlled trials (RCTs) published in English, Korean, and Chinese. Participants were stroke patients with either cerebrovascular infarction or hemorrhage who had been discharged from hospitals and managed at rehabilitation or community health centers. The intervention group was those who applied Tai Chi or Qigong alongside conventional medication. There were no limitations on Tai Chi or Qigong intervention types, durations, or settings. The control group consisted of those who used other forms of physical activity, conventional or no treatment. Pretest and posttest data were also included.
The primary outcome of this review was mobility, defined as the ability to move freely and easily, which consisted of the objective measures of the ability to walk, dynamic balance, and ADL.
A literature search was conducted between the year of inception and December 2021 on the following databases: Embase, PubMed, Cochrane Library, CINAHL, ScienceDirect, Ovid, DDOD for English, Research Information Sharing Service (RISS), Korean Studies Information Service System, National Digital Science Library, DBPIA, KoreaScholar, National Assembly Library of Korea, and Chinese National Knowledge Infrastructure (CNKI), Wanfang Data, and VIP for Chinese Science Journals Database.
To reduce publication bias, dissertations and conference proceedings were also searched by setting the search terms included in the title and abstract. In addition, the thesis and dissertation database were searched by DODD in English and Research Information Sharing Service (RISS) in the Korean database. In the Chinese database, the thesis and dissertation were automatically searched together during the general search. When duplicate studies were found, peer-reviewed articles were chosen over dissertations or conference proceedings.
We divided the search team into Korean, English, and Chinese teams in pairs. Each researcher independently searched their assigned databases to agree on a decision during the screening process and to minimize the number of missing studies. The search team used consistent search strategies. First, a search strategy was established using English MeSH and Emtree search terms. The Chinese database search strategy was then based on confirming it with the Chinese team and translating it into Chinese terms. In the Korean database, MeSH terms and corresponding keywords in Korean were used. All search strategies were established by sharing among the research teams, and progress was checked and shared through research meetings. Additional manual searches were conducted on Google Scholar and the references in the retrieved articles. MeSH and Emtree search terms were used, with Boolean operators used to combine these terms. S1 Table lists the search strategies for the English databases.
CNKI terms were the main terms for searching Chinese databases, including "Taichi (太极)" OR "Qigong (气功)" OR "Taiji Quan (太极拳)" OR "Baduan jin (八段锦)" (S2 Table). All articles selected from each database were screened for duplicates and managed using EndNote X9.3. Microsoft Excel was used to summarize the following characteristics of the included studies: year of publication, language, population, intervention type, outcome variables, and comparison groups.
Two teams of reviewers independently assessed the bias risk of each selected English and Chinese study according to Cochrane's risk of bias tool 2.0 (Cochrane RoB 2.0) [[
Comprehensive Meta-Analysis (version 3, Biostat, USA) was used to combine the effect sizes and assess heterogeneity and publication bias. The analyzed data included the sample description (sample size, and Tai Chi or Qigong intervention), duration (duration of one session, frequency, and intervention length), mobility outcome measure (the ability to walk, dynamic balance, and ADL), and control condition (active or no-treatment). As a mobility outcome, the ability to walk was measured using Timed Up and Go (TUG), gait (a subscale of the Short Physical Performance Battery [SPPB]), Dynamic Gait Index, functional ambulation category, computerized gait analysis, 10-m walk tests, 6-min walk tests, and 2-min step tests. Dynamic balance was measured using the Berg Balance Scale (BBS), and ADL was measured using the Barthel Index (BI) or modified Barthel Index (MBI) in all included studies.
Two independent researchers extracted and confirmed all available statistical data for the included studies and then entered them into an Excel spreadsheet. Quantitative data extraction included pre- and post-test data means and standard deviations, the mean and standard deviation of each group's changed scores, and the sample size of each group. Hedges' g was used to quantify effect sizes and 95% confidence intervals (CIs). A random-effects model was used for the analysis so that the results of this analysis could be compared with similar studies [[
Heterogeneity was tested using 95% prediction intervals (PIs) computed using the CMA prediction intervals program [[
As shown in Fig 1, selecting literature for the qualitative and quantitative analyses was performed in line with the Cochrane guidelines and reported using PRISMA [[
DIAGRAM: Fig 1 Flow diagram of the study selection process.
Bias risk was assessed using Cochrane RoB 2.0 to determine the bias level of each domain via an algorithm [[
In the domains of bias risk due to missing outcome data and outcome measurements, there were some concerns about only one study (3.7%), with most studies considered a low bias risk. Some of the study outcome measurements were conducted with blinding (k = 13), not specified (k = 8), or no blinding (k = 6), but mostly objective or valid measures were used as study outcomes. Regarding the bias risk in the selection of the reported results, there were some concerns about three studies (3.7%) and most studies (96.3%) had a low bias risk. The overall assessment indicated that there were some concerns about 70.4% of the studies, 29.6% had a low bias risk, and no high bias risk (Fig 2).
Graph: Fig 2 Assessment of the bias risk in the included studies.
The main characteristics of the 27 RCTs (18 using Tai Chi [[
Graph
Table 1 Characteristics of the included studies.
Study, year Country Setting Language Participants Intervention Comparison Outcome measure Safety monitoring E (M:F) C (M:F) Disease-related characteristics Mean age (years) Type Intensity (per week) Duration (weeks) Au-Yeung, 2009 Hong Kong Community English 59 (33:26) 55 (33:22) Stroke time >6months 63.4±10.7 Tai Chi, Sun style 1h×1 (+3h self-practice) 12 Stretching and education TUG Harness during test Liu, 2009 China Community Chinese 24 (14:10) 24 (11:13) Unilateral hemiplegia 52.1±14.1 Tai Chi, Yang style 30min×1 (+6×self-training) 12 Home rehabilitation training BBS Caregiver support Bai, 2011 China Hospital Chinese 30 (22:8) 30 (20:10) Ischemic: 61.7% Hemorrhage: 38.3% 53.7±4.5 Qigong (Baduanjin) 20min×14 (2 sessions/day) 6 Balance training BBS NR Cai, 2011 China Community Chinese 30 (20:10) 30 (23:7) Stroke time ≥6 months 60.3±10.5 Sitting Qigong (Baduanjin) 30min×4/5 12 No treatment BI NR Taylor-Piliae, 2012 USA Hospital (OPD) English 16 (10:6) 12 (7:5) Stroke time ≥3months 69.3±11.0 Tai Chi, Yang style 1h×3 12 No treatment SPPB (gait), 2-min step test No AEs Yang, 2013 China Hospital Chinese 50 (35:15) 50 (31:19) First stroke 54.3±13.8 Tai Chi, not specified 45min×7 4 Balance training BI, BBS Protect from falls Taylor-Piliae, 2014 USA Community English 53 (34:19) AC: 44 (20:24) UC: 48 (23:25) Ischemic: 66% Hemiparesis: 73% 69.9±10.0 Tai Chi, Yang style 1h×3 12 SilverSneakers or no treatment SPPB (gait), 2-min step test Monitored Kim, 2015 Korea Hospital English 11 (7:4) 11 (6:5) Stroke 53.5±11.5 (E) 55.2±10.2 (C) Tai Chi, Yang style 1h×2 6 Physical therapy Dynamic gait index, 10MWT, TUG NR Zheng, 2015 China Hospital Chinese 51 (27:24) 55 (31:24) Ischemic stroke 59.0±13.0 Tai Chi, not specified 30min×14 (2 sessions/day) 48 Routine rehabilitation training ADL (NI) NR Fu, 2016 China Hospital Chinese 30 (19:11) 30 (18:12) Stroke time ≤3months 59.7±7.6 Tai Chi, Yang style 40min×6 8 No treatment BBS, 10MWS Assist patients Wang, 2016 China Community Chinese 14 (9:5) 16 (14:2) Stroke 45–75 Tai chi, Yun Shou style 20min×5 12 Balance training BBS, gait analysis NR Yang, 2016 China Hospital Chinese 26 (17:9) 21 (14:7) Stroke time ≤3months 51.4±15.6 Tai Chi, not specified 40min×3 8 No treatment BBS NR Zhang, 2016 China Hospital Chinese 31 (17:14) 31 (18:13) Stroke time 1–3months 55.1±4.8 Qigong (Baduanjin) 20min×10 (2 sessions/day) 12 Balance training BBS NR Chan, 2018 Hong Kong Community English 15 (9:6) AC: 17 (10:7) UC: 15 (8:7) Stroke 63.0 ± 7.0 Tai Chi, Yang style 1h×2 12 Conventional exercise or no treatment Gait analysis No AEs Chen, 2018 China Hospital English 8 8 Stroke time ≥3 months 50–75 Tai Chi, not specified NI 24 No treatment BBS NR Xie, 2018 China Hospital English 120 (83:37) 124 (99:35) Ischemic: 74.2% Hemorrhage: 25.8% 60.9±8.7 Tai Chi, Yang style 1h×5 12 Balance training BBS, TUG, MBI No AEs Zhang, 2018 China Hospital English 45 (27:18) 45 (25:20) Ischemic: 60.0% Hemorrhage: 40% 63.7±6.8 Tai Chi, Yang style 40min×5 48 No treatment BI NR Ding, 2019 China Hospital (OPD) Chinese 57 (33:24) 56 (31:25) Ischemic: 46.9% Hemorrhage: 53.1% 55.4±4.7 (E) 56.3±3.2 (C) Qigong (Baduanjin) 20min×5 4 Balance training BBS Monitored Xie, 2019 China Hospital Chinese 30 (13:7) 20 (12:8) Ischemic: 50% Hemorrhage: 50% 51.1±12.9 (E) 53.9±13.0 (C) Qigong (Baduanjin) 50min×5 3 No treatment BI, BBS, 6MWT Monitored Fan, 2020 China Hospital Chinese 43 (29:14) 43 (30:13) Ischemic: 61.6% Hemorrhage: 38.4% 63.4±5.0 (E) 63.8±5.3 (C) Tai Chi, not specified 1.5h×3 12 No treatment BBS, TUG, 6MWT Monitored Wang, 2020 China Hospital Chinese 30 (16:14) 30 (17:13) Stoke 55.1±6.3 (E) 55.9±6.2 (C) Qigong (Baduanjin) 12min×5 4 No treatment MBI NR Yang, 2020 China Hospital Chinese 30 (18:12) 30 (14:16) Ischemic: 73.3% Hemorrhage: 26.7% 64.0±3.9 (E) 62.9±4.7 (C) Tai Chi, Yang style 30min×2 6 No treatment BI Assist patients Yu, 2020 China Community English 35 (21:14) 36 (20:16) Stroke time ≥3months Ischemic: 57.7% Hemorrhage: 42.3% 63.0±8.9 (E) 58.7±9.7 (C) 7 step forms Tai Chi from 24-form Tai Chi 40min×3 12 Conventional rehabilitation program Gait analysis, BBS No AEs Zheng, 2020 China Community English 24 (19:5) 24 (22:2) Stroke time >3months (first ever stoke) Ischemic: 47.9% Hemorrhage: 52.1% 61.6±9.2 (E) 62.8±6.4 (C) Qigong (Baduanjin) 40min×3 24 Routine rehabilitation treatment MBI No AEs Zheng, 2021 China Hospital English 30 (24:6) 30 (19:11) Stroke time <2months (first ever stoke) Ischemic: 78.3% Hemorrhage: 21.7% 63.5±10.4 (E) 67.2±9.2 (C) Qigong (Liuzijue) 45min×1 3 Conventional respiration training BBS, MBI NR Song, 2021 Korea Hospital (OPD) English 18 (10:8) 16 (11:5) Ischemic: 58.8% Hemorrhage: 41.2% 58.7±17.1 (E) 57.1±10.7 (C) Modified Tai Chi 50min×2 24 Symptom management program BBS, FAC, K-MBI No AEs Yuen, 2021 Hong Kong Hospital (OPD) English 29 (15:14) 29 (14:15) Stroke time >3months (first ever stoke) Ischemic: 62.1% Hemorrhage:37.9% 63.1±10.6 (E) 62.0±13.1 (C) Qigong (Baduanjin) 50min×3 16 Stretching training TUG, MBI No AEs
1
2 Abbreviations: M, males; F, females; MBI, modified Barthel Index; BBS, Berg Balance Scale; TUG, Timed Up and Go; BI, Barthel Index; SPPB, Short Physical Performance Battery; E, experimental group; C, control group; 10MWT, 10-m walking test; ADL, activities of daily living; AC, active control group; UC, no-treatment group; OPD, outpatient department; 10MWS, 10-m maximum walking speed; 6MWT, 6-min walking test; FAC, Functional Ambulation Category; K-MBI, Korean Version of the Modified Barthel Index; AEs, adverse events; NR, not reported
Stroke diagnoses consisted of ischemic or hemorrhagic infarction, including either hemiparesis or hemiplegia. Most studies did not include stroke duration as an inclusion criterion, with patients recruited within 3 months after the diagnosis in four studies, longer than 3 months in five studies, and longer than 6 months in two studies.
The intervention types varied for TCQ, while Yang style Tai Chi (k = 9) and Baduanjin Qigong (k = 8) were mostly applied. The intervention was defined as short term (k = 11) when performed for less than 12 weeks, with ranges of 2–14 sessions and 1–5 hours per week. The other 17 studies applied long-term interventions (12 weeks or longer), with ranges of 12–48 weeks, 2–14 sessions per week, and 20–90 min per session. The comparison groups were either a no-treatment control (received usual or no treatment) or an active control (received rehabilitation training, acupuncture, a national exercise program [e.g., the SilverSneakers app], or balance training).
No adverse events (AEs) occurred during the TCQ intervention in 7 of the 27 included studies [[
The meta-analysis of 27 RCTs with random-effects models indicated that TCQ was effective in improving the mobility of stroke survivors (Hedges' g = 0.81, 95% CI = 0.57 to 1.05). Potential heterogeneity across studies was indicated by a 95% prediction interval (PI) of –0.39 to 2.00 (Fig 3A). Publication bias was suspected based on asymmetric funnel plots and Egger's regression test (p =.018) (S1 Fig).
Graph: Fig 3 Forest plots of the effects of Tai Chi and Qigong on mobility.
- Subgroup analysis by the ability to walk, dynamic balance, and activities of daily living
- Mobility was assessed as an outcome measure using the ability to walk (k = 15), dynamic balance (k = 15), or ADL (k = 12). Hedges' g was calculated for each mobility measure: the ability to walk, dynamic balance, and ADL (Fig 3B).
- The ability to walk was assessed in 15 studies using TUG, gait analysis, 2-min step tests, 10-m walking tests, and 6-min walking tests. The significant effect of TCQ on the ability to walk was indicated by a Hedges' g of 0.43 (95% CI = 0.21 to 0.65; 95% PI = –0.35 to 1.20). No publication bias was suspected based on funnel plots and Egger's regression test (p =.176) (S1 Fig).
- Dynamic balance was measured using BBS in all included studies, and a significant large effect size of TCQ on balance (k = 15) was indicated (Hedges' g = 1.04; 95% CI = 0.65 to 1.43; 95% PI = –0.54 to 2.62), with potential heterogeneity (Fig 3B). Publication bias was suspected based on funnel plots and Egger's regression test (p =.024) (S1 Fig).
- ADL was measured using BI or MBI. The significant effect size of TCQ on ADL (k = 12) was indicated by a Hedges' g of 0.63 (95% CI = 0.34 to 0.91; 95% PI = –0.38 to 1.63). The symmetrical funnel plot and Egger's regression test (p =.110) suggested that publication bias was not present (S1 Fig).
- Subgroup analysis by program duration
- A subgroup analysis was conducted based on program duration, presented as either short term (k = 10) or long term (k = 17). The effect sizes of TCQ on mobility were indicated by Hedges' g values of 0.91 (95% CI = 0.51 to 1.30; 95% PI = –0.47 to 2.28) and 0.75 (95% CI = 0.45 to 1.06; 95% PI = –0.47 to 2.28) for short- and long-term interventions, respectively. The effect of TCQ on mobility was significant regardless of the duration of intervention, but there is no significant difference in the effect sizes between short- and long-term interventions (Q = 0.37, p =.544) (Fig 3C). The prediction interval indicated the presence of heterogeneity among the included studies. No publication bias was considered based on funnel plots and Egger's regression test for the short-term (p =.538), but publication bias was suspected in the long-term with significant Egger's regression test (p =.038) (S1 Fig).
- Subgroup analysis according to comparison groups
- A subgroup analysis was conducted between active control (k = 14) and no-treatment control (k = 15). The effect sizes of TCQ on mobility were indicated by Hedges' g values of 0.81 (95% CI = 0.43 to 1.18; 95% PI = -0.69 to 2.31) and 0.73 (95% CI = 0.47 to 0.99; 95% PI = –0.23 to 1.69) for active control and no treatment control, respectively. The effect sizes of TCQ on mobility were significant regardless of the types of comparison groups, but no significant difference was found between active control and no-treatment control groups (Q = 0.12, p =.730) (Fig 3D). The funnel plot was symmetrical, but significant Egger's regression tests (p =.038) suggested publication bias for active control. There was no publication bias for no-treatment control (p =.687) (S1 Fig).
Meta-analysis of 27 RCTs with 1,919 subjects found TCQ improved mobility, including the ability to walk, dynamic balance, and ADL in stroke patients. The effect size of the random effect model remained significant for different intervention duration, and even when compared with active control groups such as physiotherapy, balance training, or combined exercise programs.
The ability to walk was measured in our included studies using TUG, PPB, gait analysis, and walking tests, with PIs of –0.66 to 1.32. While some concerns about heterogeneity exist, these findings indicate the potential benefits of TCQ in improving the ability to walk (Hedges' g = 0.43) among stroke survivors. A previous meta-analysis involving five RCTs assessed the ability to walk using TUG and SPPB and similarly found a small effect size [[
Our analysis of dynamic balance based on 20 studies indicated the inclusion of TCQ in stroke rehabilitation programs would improve dynamic balance (Hedges' g = 1.04) among stroke survivors for a duration less than 12 weeks. Previous meta-analyses have supported that TCQ affects dynamic balance when performed two or three times weekly for 6–12 weeks [[
However, the effects of TCQ on mobility have varied between outcome measures or intervention dose [[
Multiple clinical studies have considered TCQ safe and feasible interventions for improving mobility in stroke survivors [[
Some strengths and limitations should be considered when interpreting the results. The strengths of this study would be that it drew from three different language databases, including from countries where TCQ is widely practiced and that the quality of the included trials is fairly good—with none at high risk of bias and almost 30% at low risk of bias. In addition, with the exception of walking ability, there appeared to be a low possibility of publication bias when assessed for the different components of mobility.
The main limitation of this study was the heterogeneity among the studies due to the wide range of stroke-related symptoms and rehabilitation stages. The types of intervention, although all based on traditional Chinese principles, were different in timing, intensity, and duration. Moreover, mobility was assessed using various outcome measurement methods. A subgroup analysis was performed, which defined mobility using the ability to walk, dynamic balance, and ADL, yet the PI was still large (from –0.54 to 2.62), indicating heterogeneity across the included studies.
Further studies are warranted to assess whether TCQ works best as an adjunct to rehabilitation, an effective alternative to rehabilitation or as a maintenance strategy, and whether the results could be further optimized by assessing different schools of TCQ, different types of stroke patients, and different points in the post-stroke recovery process. Although TCQ has been considered safe and feasible for stroke survivors, all future trials should have careful safety monitoring plans in place and report adverse events on the findings.
Our review suggests that TCQ was effective in improving mobility including the ability to walk, dynamic balance, and ADL among stroke survivors for programs of both shorter and longer duration. This effect of TCQ remained significant when compared with other alternative interventions. As an effective alternative to rehabilitation, TCQ could effectively applied to stroke survivors to promote functional recovery through improving the ability to walk, dynamic balance, and ADL. The heterogeneity of the included studies should be considered.
S1 Checklist. PRISMA 2009 checklist.
(DOCX)
S1 Fig. Funnel plots of the effects of Tai Chi and Qigong on mobility.
(TIF)
S1 Table. Examples of search strategies in PubMed and Embase.
(DOCX)
S2 Table. Korean and Chinese search terms.
(DOCX)
By Moonkyoung Park; Rhayun Song; Kyoungok Ju; Jisu Seo; Xing Fan; Ahyun Ryu; YueLin Li and Taejeong Jang
Reported by Author; Author; Author; Author; Author; Author; Author; Author