Registration

This systematic study was registered with (PROSPERO) for two main reasons: (1) avoid occurrence of unplanned duplication of systematic reviews; (2) and demonstrate the transparent review process for minimizing bias of studies [22].

Search strategy

Five electronic databases (Google scholar, PubMed, Physiotherapy Evidence Database [PEDro], and Cochrane Library) were utilized for literature search by entering the following terms in multiple combinations: “multiple sclerosis,” “demyelinating disease,” “Tai chi,” “Tai chi Chuan,” “Tai ji,” “Tai chi Chuan,” “Tai chi/Qigong,” “mind-body Qigong,” “Chinese mind-body exercise,” “traditional Chinese exercise,” “mindfulness-based exercise,” and “Tai ji Qigoing.” After conducting the electronic search, manual searches were subsequently performed through reference lists of the relevant publications.

Eligibility criteria

Studies were included in this review if they met the following criteria: 1) peer-reviewed studies were published between 1985 and April, 2016; 2) participants aged 18 years or above, diagnosed with MS according to applicable diagnostic criteria [23, 24]; 3) Tai Chi was used as an exercise intervention for MS patients; 4) studies must include at least one of health outcome measures (e.g., balance, strength, mobility, flexibility, pain, fatigue, depression, or quality of life. To gain a comprehensive understanding about the effectiveness of Tai Chi in MS patients, there were no restrictions about types of studies. However, review articles, conference abstracts, magazine articles, monographs, and videos were excluded.

Study selection

Two review authors (LYZ and HRW) independently performed screening process, involving two phases for obtaining eligible studies. Prior to the beginning of the screening process, the review authors first named three folders for all retrieved studies: “relevant”, “possibly relevant”, and “irrelevant”. In phase 1, the two review authors independently examined the titles and abstracts of all retrieved studies and placed them into the differently named folders. Based on percent agreement in two-rater model, the inter-rater reliability about the eligible studies within the relevant folder was calculated [25]. In phase 2 a third author emerged to discuss disagreement with the first two review authors so that they can reach an agreement about the eligible studies.

Data extraction

For each eligible study, two review authors (LYZ and HRW) utilized pre-determined summary tables to independently extract the important information. Table 1 and Table 2 assess the methodological quality of experimental studies and observational studies, respectively. Table 3 includes author name and year of publication, study purpose, study design, place of study, sample (attribution%), age of patients, disease duration, and stage in disease progression. Table 4 includes author name and year of publication, intervention frequency and duration, outcome measures, results, conclusion, adverse events/follow-up, and effect size. A third party (ZJX) appeared to ensure that all detailed information was extracted.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Study quality assessment for experimental studies.

https://doi.org/10.1371/journal.pone.0170212.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Study quality assessment for observational studies.

https://doi.org/10.1371/journal.pone.0170212.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Summary of TC for patients with multiple sclerosis (study purpose, study design, place of study, sample size (attribution%), age of patients, disease duration, and stage in disease progression.

https://doi.org/10.1371/journal.pone.0170212.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Summary of Tai Chi for patients with multiple sclerosis (intervention frequency and duration, outcome measures, results, conclusion, and adverse events/follow-up).

https://doi.org/10.1371/journal.pone.0170212.t004

Assessment of risk bias

The methodological quality of all eligible studies was systematically evaluated using Delphi quality criteria for assessment of experimental trials [26] and observational studies [27, 28]. With regard to the quality criteria for assessment of experimental studies, it included nine questions assessing risk bias of the eligible studies: random assignment, concealment of allocation, similar baseline, eligibility criteria, blinding of outcome assessor, blinding of subject, blinding of therapist, points measures and variability, and intention-t-treat analysis. With regard to the quality criteria for assessment of observational studies, quality assessment tools were determined based on types of observational studies. For example, quality assessment tool for uncontrolled, pretest/posttest design, included 12 questions to examine four types of risk bias for each eligible study: research question, eligibility criteria and study population, study participant representative of clinical populations of interest, all eligible participants enrolled, sample size, intervention clearly described, outcome measures clearly described, blinding of outcome assessors, follow-up rate, statistical analysis, multiple outcome measures, and group-level interventions and individual-level outcome efforts. For the experimental studies, if a study met seven to nine criteria, overall quality of the study was considered to be high, indicating little or no risk of bias; if a study met four to six criteria, overall quality of the study was considered to be acceptable, indicating some risk of bias; if a study only met one to three criteria, the overall quality of the study was considered to be low. For the observational studies, if a study met nine to twelve criteria, overall quality of the study was considered to be high with little or no risk of bias; if a study met five to eight criteria, overall quality of the study was considered to be acceptable with some risk of bias; if a study only met one to four criteria, overall quality of the study was considered to be poor. Two review authors (LYZ and MZ) independently performed the methodological quality assessment using the same quality assessment tools. A third review author (ZJX) calculated the interrater reliability about the results of the study quality assessment.

Data synthesis

Initially, review authors (LYZ and QF) intended to use Revman 5.3 software within the Cochrane Collaboration for data synthesis [29]. Due to the small number of randomized controlled studies, and the heterogeneity of outcome measures in the eligible studies, a meta-analysis was not performed. Therefore, the review authors (LYZ, ZJX, GD, and QF) carried out a qualitative synthesis based on the eligible studies. For those experimental studies, if mean and standard deviation were reported, effect sizes (Cohen’s d) were calculated based on the standardized formulas. A value of the effect size represents the magnitude of TC intervention that is interpreted as follows: 0.2 = small, 0.5 = medium, and 0.8 = large [30].

Study selection

A total of 89 studies were retrieved from both electronic (n = 64) and manual (n = 25) searches. Only 44 studies remained after removing 45 duplicates based on titles and abstracts. This was followed by using eligible criteria to exclude 34 articles due to the following reasons: irrelevant topic (n = 12), abstract (n = 4), review articles (n = 7), and no TC intervention (n = 11). Ten studies were finally considered eligible for full-text critical appraisal, including seven experimental studies (two randomized controlled design [31, 32] and five quasi-experimental design [19–21, 33, 34]) and three observational studies (-posttest design [35–37]). The inter-rater reliability for the selection of the eligible studies was 83.3%. The flowchart showing the study selection for this systematic review is presented as below (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow chart showing the study selection.

https://doi.org/10.1371/journal.pone.0170212.g001

Methodological quality assessment for eligible studies

The inter-rater reliability for methodological quality assessment within the ten eligible studies was 90%. A third party emerged to resolve a disagreement between the first two review authors (LYZ and HRW) about one eligible study. The methodological quality assessment for experimental studies and observational studies are presented in Table 1 and Table 2, respectively. With regard to the study quality assessment for seven experimental studies, of the two studies, one with acceptable [31] quality and another one with high quality [32] were presented, whereas the remaining five eligible studies demonstrated low overall study quality [19–21, 33, 34]. More specifically, the low overall study quality within the experimental studies is mainly attributed to the following reasons such as selection bias (non-randomized, unconcealed assignment), performance bias (absence of blinding of therapist), measurement bias (absence of blinding of participants and assessor), and attrition bias (lacking intention-to-treat analysis). With regard to the study quality assessment for three observational studies with pretest-posttest design, although one study with high quality [36] and two other studies with acceptable quality [35, 37] were presented, all three studies lacked large sample size [35–37]; two studies lacked blinding of outcome assessor [35, 37]; one study did not meet the requirement of “group-level interventions and individual-level outcome efforts” [35].

Study characteristics

The characteristics of the ten eligible studies are presented in Table 3 and Table 4. These eligible studies were conducted in five different countries (Iran, Germany, USA, India, and UK) and then published between 1999 and 2015, using three main designs (randomized controlled, quasi-experimental, and pretest-posttest) to examine the effect of TC on various health outcomes (e.g., balance, coordination, mobility, flexibility, strength, fatigue, depression, pain, and QOL) in MS patients. Study sample size ranged from 12 to 36 and attrition percentage ranged from zero to 52.6. Study participants ranged in age between 20 and 60 with an average disease duration ranging from less than six to 21.6 years. With regard to the stage of disease progression, study participants presented either Expanded Disability Status Scale (EDSS) score of less than five or Activities of Daily Living Questionnaire (ADL) score ranging from one to twenty-eight. With regard to the intervention frequency and duration, TC intervention sessions ranged from 30 to 90 minutes for two or three sessions weekly in most of the studies when the TC intervention program ranged from two to six months. For the health outcome measures, a variety of reliable and valid assessment tools were utilized, including Multiple Sclerosis Quality of Life Questionnaire, 36-item Short Form Health Status Survey, a visual analogue scale for pain, Patient-determined Disease Steps Questionnaire, 5-item Modified Fatigue Impact Scale, Center for Epidemiological Studies Depression Scale, Fatigue Scale of Motor and Cognitive Function, Questionnaire of Life Satisfaction, Dynamic Gait Index, Functional Lateral-Forward Reach test, Time Up and Go, Activities-specific Balance Confidence, Berg Balance Scale, and Profile of Mood States.

Tai chi for quality of life

Recently more attention has been paid to QOL assessment in MS patients because it covers multiple dimensions of life (physical wellbeing, material wellbeing, social wellbeing, emotional wellbeing, development, and activity). QOL assessment can better evaluate disease progression and symptomatic management when compared to assessment of fatigue, depression, or physical disability, which only reflects a small portion of life experience in MS patients [38–40]. Five studies examining the effect of TC on QOL in MS patients were included in this review [20, 21, 33, 35, 37].

Azimzadeh, Hosseini, and Tabrizi [33] examined the effect of TC on QOL in 34 female MS patients, recruited from the Multiple Sclerosis Society. With a non-equivalent, randomized, and controlled trial, the 34 MS patients either received TC training (n = 16) or usual treatment (n = 18). The researchers used the Multiple Sclerosis Quality of Life Questionnaire (MSQOL-54 scale), which is a reliable and valid disease-specific QOL instrument [38]. After a 12-week intervention period, MS patients in the TC group demonstrated significant improvements on subscales of quality of life such as pain, emotional well being, energy, social function, health distress, overall quality of life, physical health composite score, and mental health composite, as well as total quality of life (p < 0.05). However, the researchers did not report disease duration and stage in disease progression.

Tavee, Rensel, Planchon, Butler, and Stone [21] explored if an 8-week meditation-based TC was more effective for QOL in MS patients by using a prospective, nonrandomized, and controlled trial. To obtain relatively large statistical power, researchers combined two different types of patients, MS and peripheral neuropathy (PN). 39 participants (19 with MS and 17 with PN) volunteered to participate in the meditation-based TC and 25 patients (11 with MS, 14 with PN) were assigned to the control group. The final analysis included 22 (10 MS, 12 PN) in the Tai chi group and 18 (7MS, 12 PN) in the control group after 21 withdrew from the study because of unavailable transportation, loss of interest, hospitalization, and deteriorating inflammation. For the MS and PN patients combined, statistical analyses were used. Separate statistical analyses for only MS were also conducted. After 8 weeks, participants in the TCgroup showed significant improvements in SF-36 scores for overall summed physical health (p = .011 MS), mental health (p = .02 combined group), vitality (p = 0.005 combined group), and physical role (p = 0.003 combined group), whereas no improvement in the control was observed.

Burschka, Keune, Oy, Oschmann, and Kuhn [20] designed a randomized controlled study to explore if TC could be used as a therapeutic exercise to improve QOL (Questionnaire of Life Satisfaction) in mildly disabled MS patients (relapsing-remitting multiple sclerosis = 27, secondary progressive = 4, and clinically isolated syndrome = 1). After 6 withdrew from the study (schedule conflicts = 5 and health problem = 1), 32 MS patients (Expanded Disability Status Scale, EDSS < 5) were used for final analysis, including 15 in the TC group (two 90-minute sessions weekly for 6 months while keeping the usual healthcare) and 17 in the control groups (keeping usual healthcare). Although the researchers failed to report use of intention-to-treat analysis, a significant improvement on QOL was reported in the TC group (p = 0.012), with an increase of 16.8 QOL from mean score of baseline (215.77) to posttest (232.57), when compared to MS patients in the control group demonstrating a decrease of 10.65 QOL from mean score of baseline (204.46) to posttest (193.81) (p = 0.29).

Long-term healthcare for MS patients is needed in order to improve the quality of their lives. However, expensive medicine restricts the majority of MS patients to meet the need for symptomatic management. Searching for affordable therapeutic exercises is essential for MS patients to ease their economic burden. Therefore, Husted, Pham, Hekking, and Niederman [35] examined the therapeutic effect of TC on improving QOL in 19 MS patients, with a pretest-posttest design. The MS patients experienced two 1-hour sessions weekly for 8 weeks. The QOL was measured at baseline and after the 8-week intervention, using a generic Medical Outcomes Study 36-item Short-form Health Survey. Improvements were observed on the subscales (vitality, social function, mental health, and ability to perform physical and emotion roles) of the SF-36.

Emmerik, Jones, Busa, Remeius, and Averill [37] carried out a pretest-posttest trial examining the effect of TC on Multiple Sclerosis Impact Scale-29 (MSIS-29) measuring QOL, with a small sample size of 12 MS patients (age 48.5 ±10.8 years, height 1.66 ±0.08 m, mass 68.6 ±19.8 kg). The MS patients experienced nine 1-hour TC sessions within three weeks. Only seven MS patients were included for final analysis after five withdrew from the study. The total psychosocial wellbeing of the MSIS-29 was associated with a significant improvement (p = 0.032).

Tai chi for physical function

Impaired balance and mobility are two main symptoms of MS patients. More than 90% of MS patients were reported to have balance dysfunctions [41, 42]. The detrimental effect of balance dsyfunction was associated with significantly higher frequency of falling while performing daily activities, which worsens the social and economic burdens in patients [43, 44]. Therefore, researchers should focus more on the usefulness of TC for improving balance, mobility (dynamic balance), coordination, flexibility, and leg strength, because these abilities are important for people to perform voluntary locomotion movements.

Burschka et al. [20] examined a six-month TC training program versus usual treatment on balance and coordination in MS patients. Balance tests (static and dynamic) consisted of 14 tasks with an order of increasing difficulty. For the static balance assessment, MS patients were asked to perform single leg stances in different conditions (descriptions of the different conditions were not reported); the dynamic balance was measured by walking forward, making a turn, and walking back toward the start line on a wooden floor. Similarly, a coordination test consisted of 10 tasks, with an order of increasing difficulty. For the balance and coordination tests, MS patients were awarded one point if he or she completed each task. A maximum of 14 points for the balance test and 10 points for the coordination test can be given to each MS patient. MS patients were asked to complete the two tests at baseline and after a six-month intervention. Significant improvements on both balance (p = 0.031; effect size = 0.79) and coordination performance (p = 0.003; effect size = 0.83) were observed in the TC group, whereas slightly decreased mean scores emerged in the group receiving usual treatment, from pretest (6.88) to posttest (6.53) in the balance test, and from pretest (4.94) to posttest (4.82) in the coordination test.

Researchers examined the effectiveness of a combined exercise (TC and mindful practice) versus TC on balance, gait, and mobility in 16 patients with RRMS, with a randomized, controlled, paralleled study [31]. Eight RRMS patients in Group 1 experienced twenty 60-minute sessions (20-minute mental practice, followed by 40-minute TC) over 10 to 20 weeks. Eight in Group 2 completed twenty 40-minute TC sessions over the same duration. The study included Dynamic Gait Index (DGI) (measuring mobility and dynamic balance), Functional Balance Reach (FR) test at lateral and forward directions, Timed Up and Go (TUG) (measuring gait speed, dynamic balance and mobility), and Activities-specific Balance Confidence (ASBC). The method for data analysis was not reported. Although both intervention arms demonstrated significant improvements in DGI (effect size = 0.17), FRF (effect size = 0.3), FRL (effect size = 1.2), TUG (effect size = 0.55), and ASBC (effect size = 1.02), no significant differences existed between the two groups. Researchers concluded that TC is beneficial for improving balance and functional mobility in RRMS patients.

Non-equivalent control, pretest-posttest design was used in two different studies, examining the effectiveness of TC on balance, as measured by the Berg Balance Scale in female MS patients [19, 34]. Azimzadeh, Hosseini, Nourozi, and Davidson [19] equally assigned 36 MS patients into two groups: an intervention group experiencing two 45 to 60-minute group-based TC sessions weekly for 12 weeks and the control group receiving counseling sessions and physical therapy as usual services. Mohali, Ebrahimi, Hassan, Khoshraftar, and Shoeibi [34] recruited 30 MS patients from the MS Society and assigned them into two groups: an intervention group receiving an 8-week TC training with three sessions weekly and the control group only receiving usual treatment. Although both studies did not report the statistical significance, improvements (effect size = 0.15) on average balance scores between baseline and post-test were observed in both (p = 0.003) [19] and (p < 0.001) [34], whereas no significant changes were observed in the control groups.

Two research groups conducted similar studies about the usefulness of TC on balance performance in secondary progressive MS patients with two different study designs: pretest-posttest [36] and randomized-controlled-paralleled design [32]. Mills, Allen, and Carey-Morgan [36] initially assigned 12 MS patients into intervention training receiving six individual TC sessions, followed by 30 minutes of home-based TC exercises guided by a videotape per day for the rest of the 2-month intervention period. Four withdrew because of bereavement, loss of interest, or incomplete symptom diaries. The results of the final analysis including the eight MS patients showed a significant improvement on balance performance between baseline (5.63) and posttest (11.88) (p < 0.05). Mills and Allen [32] utilized mindfulness-based TC to help with symptom management in eight MS patients. Balance assessment was performed at baseline, after six TC training sessions, and a 3-month follow-up. Although the description of statistical analysis was not reported, the TC group was associated with a significant improvement on balance performance (effect size = 1.48), and five MS patients still maintained this balance performance at the 3-month follow-up assessment.

Apretest-posttest design was used in two different studies, examining the effectiveness of short-term TC training on lower limb-related functions, including balance (static balance and mobility), neural drive, flexibility, and leg strength [35, 37]. Husted et al. [35] used 25 feet walking distance (measuring functional balance and mobility) and hamstring flexibility tests as outcome measures in the study. After an 8-week TC training, walking speed of MS patients was 21% faster than the baseline walking speed and post-intervention flexibility was 28% greater when compared to the baseline flexibility performance. In the study of a 3-week TC training for benefits in balance, gait, leg strength, and the neural drive was conducted by [37]. MS patients were asked to perform multiple tests, including sensorimotor (plantar sensation and toe taps measuring neural drive), Chair Rise test (leg strength), and average center of pressure velocity (COP) and total excursion and time to contact (TTC) for postural stability. Multiple significant improvements were observed in plantar sensation (p = .02), neural drive (p = 0.024), leg strength (p = 0.025), and dynamic balance (p = 0.022). Postural sway velocity of Tandem Stance test did not demonstrate significant improvement, but significant improvements in the Tandem Stance test on TTC at Antero-Posterior (p = .005) and Medio-lateral (p = .05) directions were observed.

Tai chi for fatigue

Fatigue is defined as a subjective perception of physical or mental energy deficiency while performing voluntary movements in daily life [45]. Fatigue is one of the most commonly reported symptoms in patients with MS, affecting roughly four-fifths of MS patients [46]. Patients with multiple sclerosis do not only experience fatigue, but multiple motor functions may also be deteriorated by MS-related fatigue, including balance and mobility [47, 48]. TC has been proven to provide fatigue relief in patients with breast cancer [49] and rheumatoid arthritis [50], therefore it may be an alternative method for fatigue management in MS patients.

Five studies [20, 21, 32, 36, 37]were found that examined the effect of TC on fatigue in patients with MS. Fatigue was measured using three different assessment tools, including the 5-item Modified Fatigue Impact Scale (MFIS-5) [21], the Fatigue Scale of Motor and Cognitive Function (FSMC) [20], the Fatigue Severity Scale (FSS) [37], the Profile of Mood States (POMS) [36], and the Symptom Rating Questionnaire (SRQ) [32]. The results of the five studies were not consistent. Of those, TC training was proved to have a significant beneficial effect on fatigue relief in two studies [21, 36], but this benefit was not observed in three studies [20, 32, 37].

Tai chi for depression and pain

Two studies assessed the effectiveness of TC on depression [20, 36]. Burschka et al. [20] used the Center for Epidemiological Studies Depression (CES-D) scale, reporting a significant reduction in depression (p = 0.007) in MS patients in the TC group, in comparison to the control group. A significant reduction was also observed in depression as measured by the POMS between pre-test (6.25) and post measure (3.00; p < 0.04) in [36]. Pain is a relatively common symptom, following MS patients about half of the time over the course of their disease [50]. Pain management is highly reliant on conventional medicine, but side effects are often unavoidable. Tavee et al. [21] conducted a study examining if TC is beneficial for pain relief in patients with MS, and reported a significant reduction in pain (p = 0.031), as measured by the Visual Analogue Scale for Pain (VAS).