Tai Chi for rheumatoid arthritis

Abstract

Background

Rheumatoid arthritis (RA) is a chronic, systemic, inflammatory, autoimmune disease that results in joint deformity and immobility of the musculoskeletal system. The major goals of treatment are to relieve pain, reduce inflammation, slow down or stop joint damage, prevent disability, and preserve or improve the person's sense of well‐being and ability to function. Tai Chi, interchangeably known as Tai Chi Chuan, is an ancient Chinese health‐promoting martial art form that has been recognized in China as an effective arthritis therapy for centuries. This is an update of a review published in 2004.

Objectives

To assess the benefits and harms of Tai Chi as a treatment for people with rheumatoid arthritis (RA).

Search methods

We updated the search of CENTRAL, MEDLINE, Embase, and clinical trial registries from 2002 to September 2018.

Selection criteria

We selected randomized controlled trials and controlled clinical trials examining the benefits (ACR improvement criteria or pain, disease progression, function, and radiographic progression), and harms (adverse events and withdrawals) of exercise programs with Tai Chi instruction or incorporating principles of Tai Chi philosophy. We included studies of any duration that included control groups who received either no therapy or alternate therapy.

Data collection and analysis

We used standard methodological procedures expected by Cochrane.

Main results

Adding three studies (156 additional participants) to the original review, this update contains a total of seven trials with 345 participants. Participants were mostly women with RA, ranging in age from 16 to 80 years, who were treated in outpatient settings in China, South Korea, and the USA. The majority of the trials were at high risk of bias for performance and detection bias, due to the lack of blinding of participants or assessors. Almost 75% of the studies did not report random sequence generation, and we judged the risk of bias as unclear for allocation concealment in the majority of studies. The duration of the Tai Chi programs ranged from 8 to 12 weeks.

It is uncertain whether Tai Chi‐based exercise programs provide a clinically important improvement in pain among Tai Chi participants compared to no therapy or alternate therapy. The change in mean pain in control groups, measured on visual analog scale (VAS 0 to 10 score, reduced score means less pain) ranged from a decrease of 0.51 to an increase of 1.6 at 12 weeks; in the Tai Chi groups, pain was reduced by a mean difference (MD) of ‐2.15 (95% confidence interval (CI) ‐3.19 to ‐1.11); 22% absolute improvement (95% CI, 11% to 32% improvement); 2 studies, 81 participants; very low‐quality evidence, downgraded for imprecision, blinding and attrition bias.

There was very low‐quality evidence, downgraded for, blinding, and attrition, that was inconclusive for an important difference in disease activity, measured using Disease Activity Scale (DAS‐28‐ESR) scores (0 to 10 scale, lower score means less disease activity), with no change in the control group and 0.40 reduction (95% CI ‐1.10 to 0.30) with Tai Chi; 4% absolute improvement (95% CI 11% improvement to 3% worsening); 1 study, 43 participants.

For the assessment of function, the change in mean Health Assessment Questionnaire (HAQ; 0 to 3 scale, lower score means better function) ranged from 0 to 0.1 in the control group, and reduced by MD 0.33 in the Tai Chi group (95% CI ‐0.79 to 0.12); 11% absolute improvement (95% CI 26% improvement to 4% worsening); 2 studies, 63 participants; very low‐quality evidence, downgraded for imprecision, blinding, and attrition. We are unsure of an important improvement, as the results were inconclusive.

Participants in Tai Chi programs were less likely than those in a control group to withdraw from studies at 8 to 12 weeks (19/180 in intervention groups versus 49/165 in control groups; risk ratio (RR) 0.40 (95% CI 0.19 to 0.86); absolute difference 17% fewer (95% CI 30% fewer to 3% fewer); 7 studies, 289 participants; low‐quality evidence, downgraded for imprecision and blinding.

There were no data available for radiographic progression. Short‐term adverse events were not reported by group, but in two studies there was some narrative description of joint and muscle soreness and cramps; long‐term adverse events were not reported.

Authors' conclusions

It is uncertain whether Tai Chi has any effect on clinical outcomes (joint pain, activity limitation, function) in RA, and important effects cannot be confirmed or excluded, since all outcomes had very low‐quality evidence. Withdrawals from study were greater in the control groups than the Tai Chi groups, based on low‐quality evidence. Although the incidence of adverse events is likely to be low with Tai Chi, we are uncertain, as studies failed to explicitly report such events. Few minor adverse events (joint and muscle soreness and cramps) were described qualitatively in the narrative of two of the studies. This updated review provides minimal change in the conclusions from the previous review, i.e. a pain outcome.

Plain language summary

Tai Chi for people with rheumatoid arthritis

Cochrane researchers conducted a review of the effect of Tai Chi for people with rheumatoid arthritis (RA). Literature searches to September 2018 found seven studies with 345 people, ranging in age from 16 to 80 years, in China, South Korea, and the USA. Some attended classes at a hospital, which ranged from 8 to 12 weeks long; others participated in a self‐guided Tai Chi program.

What is rheumatoid arthritis ?

Rheumatoid arthritis is a disease in which the body's immune system attacks its own healthy tissues, causing redness, pain, swelling, and heat around the joints, primarily in the hands and feet. At present, there is no cure for RA, so treatments aim to relieve pain and stiffness, and improve the ability to move.

What is Tai Chi?

Tai Chi combines deep breathing and relaxation with slow and gentle movements. In older people, Tai Chi has been shown to decrease stress, increase muscle strength in the lower body, and improve balance, posture, and the ability to move. It is not known whether Tai Chi provides the same benefits for people with RA.

What happened to people with rheumatoid arthritis who did Tai Chi?

Pain (measured on a visual analog scale (VAS) at 12 weeks)

‐ People who did Tai Chi rated their pain 2.15 points lower (better) on a scale of 0 to 10, compared to the change in the control group (22% absolute improvement). The quality of the evidence was very low, due to a low number of participants and concerns about study design (2 studies, 81 participants).

‐ People who did not do Tai Chi reported a mean change in pain that ranged from 0.5 lower to 1.6 points higher.

Disease Activity (measured with the Disability Activity Scale (DAS‐28‐ESR) at 12 weeks)

‐ People who did Tai Chi scored 0.4 points lower (better) on a scale of 0 to 10 for disease activity compared to the control group (4% absolute improvement). The quality of the evidence was very low, due to concerns about study design and a high number of withdrawals (1 study, 43 participants).

‐ People who did not do Tai Chi reported no change in disease activity.

Function (measured by the Health Assessment Questionnaire (HAQ) at 12 weeks)

‐ People who did Tai Chi scored 0.33 points lower (better) on a scale of 0 to 3 for function compared to the control group (11% absolute improvement). The quality of the evidence was very low, due to concerns about study design and a high number of withdrawals (2 studies, 63 participants).

‐ People who did not do Tai Chi reported a mean change in function ranging from no change to 0.1 points higher.

Overall withdrawals

‐ 17/100 fewer people withdrew from the Tai Chi groups at 12 weeks (17% absolute improvement). The quality of the evidence was low, due to the low number of participants and concerns about study design (7 studies, 289 participants).

We found no studies that looked specifically at radiographic progression, short‐term or long‐term adverse events, although two studies described some joint and muscle soreness and cramps in the text.

What is the bottom line?

We are uncertain whether Tai Chi improves pain, disease activity, or function in people with RA. It is also not clear how much, how intense, and for how long Tai Chi should be done to see benefits.

Summary of findings

Summary of findings for the main comparison. Tai Chi for Rheumatoid Arthritis.

Tai Chi exercise programs compared to no Tai Chi exercise
Patient or population: people with rheumatoid arthritis Settings: outpatient groups in China, South Korea, and the USA Intervention: exercise programs with Tai Chi instruction, or incorporating Tai Chi philosophy Comparison: no Tai Chi exercises or alternative exercises
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments (95% CI)
	Assumed risk	Corresponding risk
Change in pain Visual Analog Scale (VAS) at 12 weeks (0 to 10 scale; lower score means less pain)	The change in mean VAS ranged across control groups from ‐0.51 to 1.6	Mean pain was reduced by 2.15 in the Tai Chi groups (3.19 lower to 1.11 lower)**	MD ‐2.15 (‐3.19 to ‐1.11)**	81 (2 studies)	⊕⊝⊝⊝ very low a,b,c	Absolute difference = 22% lower (11% lower to 32% lower) NNTB = 3 (2 to 5) Relative percent change = 26% (14% to 39%; calculated using baseline mean from Lee 2006 = 8.2)
Change in measure of disease activity Disease Activity Score (DAS‐28‐ESR) at 12 weeks (0 to 10 score; lower score means less disease activity)	Change in mean DAS‐28‐ESR was 0 in the control group	Mean DAS‐28‐ESR was reduced by 0.4 in the Tai Chi group (1.10 lower to 0.30 higher)	MD ‐0.40 (‐1.10 to 0.30)	43 (1 study)	⊕⊝⊝⊝ very low a,b,c,d	Absolute difference = 4% lower (11% lower to 3% higher) NNTB = n/a Relative percent change = 11% (‐9% to 31%; baseline mean = 3.5)
Change in function Health Assessment Questionnaire (HAQ) at 12 weeks (0 to 3 score; lower score means better function)	The change in mean HAQ ranged from 0 to an increase of 0.1 in the control group.	Mean HAQ scores were reduced by 0.33 (0.79 lower to 0.12 higher)	MD ‐0.33 (‐0.79 to 0.12)	63 (2 studies)	⊕⊝⊝⊝ very low a,b,c,d	Absolute difference = 11% lower (26% lower to 4% higher). NNTB = n/a Relative percent change = 94% (‐34% to 225%; calculated using baseline mean from Shin 2015 = 0.35)
Radiographic progression	N/A	N/A	N/A	N/A	N/A	Outcome not reported in included studies
Short‐term serious adverse events	N/A	N/A	N/A	N/A	N/A	Outcome not reported by group in included studies – some general types of adverse events (i.e. joint and muscle soreness, cramps) described in two studies (Kirsteins 1991a; Kirsteins 1991b)
Long‐term adverse effects or toxicity	N/A	N/A	N/A	N/A	N/A	Outcome not reported in included studies
Withdrawals – overall (at 8 to 12 weeks)	297 per 1000	119 per 1000 (56 to 255)	RR 0.40 (0.19 to 0.86)	289 (7 studies)	⊕⊕⊝⊝ low b,c	Absolute difference = 17% fewer (30% fewer to 3% fewer) NNTB=6 (4 to 18) Relative percent change = 60% (14% to 81%)

*The basis for the assumed risk is the median control group risk across studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

** Back‐translated from SMD to a (0 to 10) scale by multiplying SMD by standard deviation of control group at baseline in Lee 2006 (SD = 2.26).
 
 CI: Confidence interval; RR: Risk ratio; n/a: data not available

GRADE Working Group grades of evidence
 High quality: Further research is very unlikely to change our confidence in the estimate of effect.
 Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
 Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
 Very low quality: We are very uncertain about the estimate

^a Downgraded one level due to imprecision: low number of participants

^b Downgraded one level due to study limitations: high withdrawal rate lowered the confidence in the estimate of effect in one or more studies

^c Downgraded one level due to study limitations: participants and outcome assessors not blinded

^d Downgraded one level due to imprecision: 95% CI estimate includes both null effect and appreciable benefit or harm

Background

Description of the condition

Rheumatoid arthritis (RA) is a chronic, systemic, inflammatory autoimmune disease (Lee 2011). RA is the most common form of autoimmune inflammatory arthritis in adults; it has a disease prevalence of < 1% (Cross 2014; Kvien 2004). The disease is often progressive, and results in pain, stiffness, and swelling of joints. In late stages, joint deformity and joint destruction develop, which lead to disability and sometimes, loss of independence (Odegard 2005; Yelin 2007).

Because of the complex, systemic nature of the disease, the treatment for RA is also complex, and involves a variety of approaches. The major goals are to relieve pain, reduce inflammation, slow down or stop joint damage, prevent disability, and preserve or improve the person's sense of well‐being and ability to function. Commonly used approaches include use of pharmacological treatment interventions including: non‐steroidal anti‐inflammatory drugs (NSAIDs) for symptom‐relief; glucocorticoids that improve signs and symptoms and slow disease progression, as a short‐term treatment for flares and a bridge to definitive therapy; and traditional or biologic disease‐modifying anti‐rheumatic drugs (DMARDs), which improve signs and symptoms, delay disease progression, improve quality of life and function, and are usually used long‐term or life‐long. While pharmacological approaches are the main‐stay of RA treatment, non‐pharmacological approaches also play a significant role in its management (Vliet Vlieland 2011).

Preservation of muscle strength and endurance is also very important in individuals with RA. Muscle atrophy can quickly occur in people with RA because of disuse, splinting, or the inflammatory process. Deterioration of functional capacity is associated with a decline in independence (Li 2001).

Maintenance of near normal range of motion (ROM) in joints is important in preserving functional ability. One way to preserve muscle strength and endurance, as well as to maintain ROM, is through an exercise routine designed to meet the specific needs of each person. Exercise therapy is aimed at preserving and restoring overall functional ability by maintaining healthy and strong muscles, preserving joint mobility, and maintaining flexibility, balance, strength, endurance, and aerobic capacity. A Cochrane Review of dynamic exercise therapy demonstrated significant benefits on muscle strength, aerobic capacity, and ROM, but it excluded the studies of Tai Chi (Van den Ende 2003).

Description of the intervention

Tai Chi is a traditional Chinese martial art that was developed in the 13th century (Koh 1981). Although it has been widely practiced in China for centuries as an art form, religious ritual, relaxation technique, exercise, and a self‐defence method, it was only during the early 1980s that scientists began to investigate the potential health benefits of Tai Chi (Forge 1997). A variety of health‐related benefits of Tai Chi for older adults have since been documented, including stress reduction (Jin 1992), improved agility and balance (Tse 1991), posture control (Wolf 1997), and lower extremity strength (Duncan 2001). The benefits of Tai Chi can impede the decline of the musculoskeletal system that occurs with aging, and the associated deterioration in functional capacity, and increased risk of falls and hip fractures (Li 2001). A further benefit of Tai Chi, is that compared to most drugs, it is a relatively inexpensive, low technological treatment (Wolf 1996).

How the intervention might work

Practiced internationally as a health‐promoting exercise, Tai Chi combines deep diaphragmatic breathing and relaxation with slow and gentle movements, both isometric and isotonic, while maintaining good postures. Isometric exercises allow the individual to exercise specific muscle groups while avoiding joint motion. Isotonic exercises contract muscles in a way that causes joint movement. The philosophy behind Tai Chi practice is interconnected with Chinese medical theory. Good health is a result of the body's vital energy, the C'hi, circulating freely throughout the body. Illness occurs as a result of blockage of C'hi. Tai Chi is believed to promote the free flow of C'hi if practiced with the correct posture and sufficient relaxation.

Tai Chi involves stepping with full weight‐bearing on both lower extremities, but has a gentler heel‐strike than walking, because of slow and deliberate foot placement. It is unique for its slow, graceful movements with low impact, low velocity, and minimal orthopedic complications. When practiced correctly, the movements of Tai Chi flow imperceptibly from one into another. Although Tai Chi is performed in slow, relaxed, and continuous movements, it demands a considerable amount of work by the leg muscles, because forms are completed with knees bent in a squat‐like position. In addition to the weight‐shifting feature, Tai Chi has a number of therapeutic elements including: (a) small to large degrees of motion; (b) knee flexion; (c) straight and extended head and trunk; (d) combined rotation of head, trunk, and extremities; and (e) asymmetrical diagonal arm and leg movements (Wolf 1993). In terms of metabolic demands, Tai Chi is approximately equivalent to walking at a speed of 6 km/h (Jin 1992), and produces an average increase in heart rate of 50%, using one type of short form (Zhuo 1984). In addition to the physical benefits, Tai Chi uses 'top down' processes, such as cognition and visualization, to provide a total 'mind‐body' exercise (Wayne 2013).

Why it is important to do this review

Although Tai Chi has been recognized in China as an effective therapy for arthritis for centuries (Lam 1998), very few systematic reviews report evidence of its effectiveness on rheumatic disease (Han 2004; Lee 2007). We conducted this review according to the guidelines recommended by the Cochrane Musculoskeletal Group Editorial Board (Ghogomu 2014); it provides updated evidence to our previous report (Han 2004). The previous review found no detrimental effects of Tai Chi on disease activity of RA, and an important benefit in lower‐extremity range of motion outcomes, but did not assess effects on self‐reported pain.

Objectives

To assess the benefits and harms of Tai Chi as a treatment for people with rheumatoid arthritis (RA).

Methods

Criteria for considering studies for this review

Types of studies

We included both randomized controlled trials (RCTs) and controlled clinical trials (CCTs) in this review. We included studies reported as full text, those published as abstract only, and unpublished data; we had no language restriction.

A controlled clinical trial refered to a study that compared one or more intervention groups to one or more comparison (control) groups. This classification was also applied if the method of randomization was not specifically stated, or the method of allocation was known, but not considered strictly random.

We allowed comparisons with no therapy or other active therapy.

Types of participants

This review included adults with a diagnosis of RA, as defined by the American College of Rheumatology (ACR) Classification Critieria (Arnett 1988).

Types of interventions

We included only trials of exercise programs with Tai Chi instruction, or exercises that incorporated principles of Tai Chi training, which were compared to a non‐exercise or alternative exercise method.

Types of outcome measures

Major outcomes

We used major outcome measures for RA trials for Cochrane Reviews (as described below) due to their relevance to people with RA for interventions assessed in this condition (Felson 1995):

1. American College of Rheumatology improvement criteria or improvement in pain– 50% (ACR50), defined as 50% improvement in both tender and swollen joint counts and 50% improvement in at least three of the following five variables: patient global assessment, physician global assessment, pain score, Health Assessment Questionnaire (HAQ) score, and acute phase reactant (Erythrocyte Sedimentation Rate (ESR) or C‐Reactive Protein (CRP (Chung 2006; Felson 1995; Ghogomu 2014)). In the absence of ACR50, we considered ACR20 or ACR70, i.e. 20% or 70% improvement in the criteria listed above, or pain, whichever had the most consistent data. Minimal clinically important difference (MCID) on pain is ≥ 2 on a 0 to 10 point pain scale (Farrar 2000; Farrar 2000a; Farrar 2001). If pain was chosen as the major outcome, outcome data on ACR 20/50/70 was reported under minor outcomes.

2. RA disease remission – defined as Disease Acivity Score (DAS) < 1.6 or DAS‐28 < 2.6 (Fransen 2005; Prevoo 1996)

3. Function – measured by HAQ score or modified HAQ, calculated as score changes (Fries 1980; Pincus 1983), and the proportion achieving MCID on HAQ ≥ 0.22 (Wells 1993)

4. Radiographic progression – as measured by the Larsen, Sharp, or modified Sharp scores (Larsen 1977; Sharp 1971; Van der Heijde 1989)

5. Short‐term serious adverse events (SAEs (Ioannidis 2004))

6. Long‐term adverse effects or toxicity (i.e. cancer)

7. Withdrawals due to adverse events (Ioannidis 2004); in its absence, we considered overall withdrawals.

Minor outcomes

In addition to the major outcomes, we included minor outcome measures of effectiveness from OMERACT 1993 for potential analysis. OMERACT measures for effectiveness include:

1. Number of tender joints per person

2. Number of swollen joints per person

3. Physician global assessment

4. Patient global assessment

5. Acute phase reactants

6. Range of motion

7. Grip strength

Search methods for identification of studies

Electronic searches

The original review searched MEDLINE (1966 to September 2002) and the Cumulative Index of Nursing and Allied Health Literature (CINAHL) databases (1982 to September 2002) using the same search strategy (Appendix 1). We also searched the databases of the Beijing Chinese Academy of Traditional Medicine, the Chinese Biomedical Database, and the Cochrane Controlled Trials Register to December 2003.

We incorporated multiple terms and spellings for Tai Chi and rheumatoid arthritis in the updated search strategy in May 2013 (Appendix 2) and September 2018 (Appendix 3). The May 2013 search included the Cochrane Library (to April 2013), MEDLINE (1946 to May 2013), Embase (1947 to May 2013), the Allied and Complementary Medicine (AMED) database (1985 to May 2013), and the Conference Proceedings Citation Index‐Science (CPCI‐S) (1990 to May 2013). The September 2018 search included MEDLINE (1946 to September 2018), Embase (1947 to September 2018), Cochrane Central Register of Controlled Trials (to August 2018) and search of trial registries ClinicalTrials.gov and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) portal (apps.who.int/trialssearch) through September 2018.

We applied no language restrictions.

Searching other resources

We manually searched reference lists of all the trials selected through the electronic search (both original studies and reviews) to identify additional trials.

Data collection and analysis

Selection of studies

Two review authors independently screened titles and abstracts of all of the potentially‐relevant studies we identified during the searches, and coded them as 'retrieve' (eligible or potentially eligible, or unclear) or 'do not retrieve'. We retrieved the full‐text study reports or publications, and two review authors independently screened the full text to identify studies for inclusion, and identified and recorded reasons for excluding the ineligible studies. We resolved any disagreement through discussion, or consulted a third person. We identified and excluded duplicates and collated multiple reports of the same study so that each study, rather than each report, was the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram and record the characteristics of excluded studies (prisma‐statement.org/PRISMAStatement/Default.aspx).

Data extraction and management

We used a data collection form for study characteristics and outcome data, which was piloted. Two independent reviewers (AH and VR for original review; AM and NS for update) extracted the following study characteristics:

1. Methods: study design, total duration of study, details of any 'run‐in' period, number of study centres and location, study setting, withdrawals, and date of study

2. Participants: N, mean age, age range, sex, disease duration, severity of condition, diagnostic criteria, important baseline data; inclusion criteria and exclusion criteria

3. Interventions: intervention, comparison, concomitant medications, and excluded medications

4. Outcomes: major and minor outcomes specified and collected, and time points reported

5. Characteristics of the design of the trial as outlined below in the 'Assessment of risk of bias in included studies' section

6. Notes: funding for trial and notable declarations of interest by trial authors

We extracted the number of events and number of participants per treatment group for dichotomous outcomes, and means, standard deviations, and number of participants per treatment group for continuous outcomes. We resolved disagreements by consensus. One review author (AM) transferred data into the Review Manager 5 file (Review Manager 2014). We double‐checked that data were entered correctly by comparing the data presented in the systematic review with the study reports.

When possible, we abstracted change from baseline values instead of final values, unless final values were the only ones presented. In order to make proper comparisons, we abstracted unadjusted, per protocol values when available. If studies presented multiple time points, we abstracted all time points in order to be able to make the closest time point comparison with other studies.

Assessment of risk of bias in included studies

Two review authors (AH and VR for original review; AM and NS for update) independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We resolved disagreements by discussion. We assessed the risk of bias according to the following domains.

1. Random sequence generation (selection bias)

2. Allocation concealment (selection bias)

3. Blinding of participants and personnel (performance bias)

4. Blinding of outcome assessment (detection bias)

5. Incomplete outcome data (attrition bias)

6. Selective outcome reporting (reporting bias)

7. Other bias

We graded each potential source of bias as high, low, or unclear risk, and provided a quote from the study report, together with a justification for our judgment in the 'Risk of bias' table. We summarised the 'Risk of bias' judgements across different studies for each of the domains listed.

When considering treatment effects, we took into account the risk of bias for the studies that contributed to that outcome.

We presented the figures generated by the 'Risk of bias' tool to provide summary assessments of the risk of bias.

Measures of treatment effect

We analysed dichotomous data as risk ratios (RR), or Peto odds ratios (OR) when the outcome was a rare event (approximately less than 10%), and used 95% confidence intervals (CIs). We analysed continuous data as mean difference (MD) or standardized mean difference (SMD), and entered data presented as a scale with a consistent direction of effect across studies.

We used SMDs when different scales were used to measure the same conceptual outcome (e.g. disability), and back‐translated to a typical scale (e.g. 0 to 10 for pain) by multiplying the SMD by a typical among‐person standard deviation (e.g. the standard deviation (SD) of the control group at baseline from the most representative trial), as per Chapter 12 of theCochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011b).

In the Effects of interventions section, and the comments column of the 'Summary of findings' table, we provided the absolute percent difference, the relative percent change from baseline, and the number needed to treat for an additional beneficial outcome (NNTB), which we provided only when the outcome showed a statistically significant difference.

For dichotomous outcomes, we calculated the NNTB from the control group event rate and the relative risk, using the Visual Rx NNT calculator (Cates 2008). We calculated the NNTB for continuous measures using the Wells calculator (available at the CMSG Editorial office, musculoskeletal.cochrane.org/).

For dichotomous outcomes, we calculated the absolute risk difference using the risk difference statistic in RevMan 5, and expressed the result as a percentage. For continuous outcomes, we calculated the absolute benefit as the improvement in the intervention group minus the improvement in the control group, in the original units, expressed as a percentage.

We calculated the relative percent change for dichotomous data as the risk ratio ‐ 1, and expressed it as a percentage. For continuous outcomes, we calculated the relative difference in the change from baseline as the absolute benefit divided by the baseline mean of the control group, expressed as a percentage.

Clinical Relevance

Clinical improvement was defined as a 15% improvement relative to a control. A relative percentage difference (RPD) as defined by the Philadelphia Panel on Rehabilitation Interventions of greater than 15% indicates clinical importance (Wells 2001).

Unit of analysis issues

Where multiple trial arms were reported in a single trial, we would have included only the relevant arms. If two comparisons (e.g. drug A versus placebo and drug B versus placebo) were combined in the same meta‐analysis, we would have halved the control group to avoid double‐counting. However, no included studies provided issues with unit of analysis.

Dealing with missing data

For dichotomous outcomes (e.g. ACR20 and number of withdrawals), we used the number of participants randomised to the group as the denominator.

For continuous outcomes (e.g. mean change in pain score), we calculated the MD based on the number of participants analyzed at that time point. If the number of participants analyzed was not presented for each time point, we used the number of randomised participants in each group at baseline.

We did not need to compute missing standard deviations from another statistic, such as standard errors, CIs or P values, using the methods recommended in the Cochrane Handbook for Systematic Reviews of Interventions, Chapter 7 (Higgins 2011).

Assessment of heterogeneity

We assessed clinical and methodological diversity in terms of participants, interventions, outcomes, and study characteristics for the included studies by observing these data from the data extraction tables, to determine whether a meta‐analysis was appropriate. We assessed statistical heterogeneity by visual inspection of the forest plot, looking for obvious differences in results between the studies, and using the I² and Chi² statistical tests.

As recommended in theCochrane Handbook for Systematic Reviews of Interventions we interpreted an I² value of 0% to 40% to mean it might not be important; 30% to 60% to mean it may represent moderate heterogeneity; 50% to 90% to mean it may represent substantial heterogeneity; and 75% to 100% to represent considerable heterogeneity. As further recommended in the Cochrane Handbook for Systematic Reviews of Interventions, we noted that the importance of I² depended on: (i) magnitude and direction of effects and (ii) strength of evidence for heterogeneity. A Chi² test with a P value ≤ 0.10 indicated evidence of statistical heterogeneity (Deeks 2011).

Assessment of reporting biases

We planned to create and examine a funnel plot to explore possible small study biases, however, there were not enough included studies to effectively test for reporting bias.

To assess outcome reporting bias, we planned to check trial protocols against published reports. For studies published after 1 July 2005, we screened the Clinical Trial Register at the International Clinical Trials Registry Platform of the WHO for the published trial protocol. We evaluated whether selective reporting of outcomes was present.

Data synthesis

We pooled outcomes and treatments where appropriate, and used a random‐effects model to analyze data for a conservative approach. In cases where heterogeneity was moderate to substantial, we also performed fixed‐effects model to see if the interpretation would be different.

GRADE and 'Summary of findings' tables

We created a 'Summary of findings' table that included the following outcomes: pain, disease activity (measured by DAS‐28‐ESR), function (measured by HAQ), radiographic progression, short‐term serious adverse events, long‐term adverse effects or toxicity, and withdrawals overall at 12 weeks.

Two people(AM, JS) independently assessed the quality of the evidence. We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of the body of evidence as it related to the studies that contributed data to the meta‐analyses for the prespecified outcomes. We assessed the quality of evidence as high, moderate, low, or very low. We considered the following criteria for upgrading the quality of evidence, if appropriate: large effect, dose‐response gradient, and plausible confounding effect. We used the methods and recommendations described in sections 8.5 and 8.7, and chapters 11 and 12, of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011; Schünemann 2011a; Schünemann 2011b). We used GRADEpro GDT software to prepare the 'Summary of findings' tables (GRADEpro GDT 2015). We justified all decisions to down‐ or up‐grade the quality of studies using footnotes, and we provided comments to aid the reader's understanding of the results where necessary. We provided the number needed to treat for an additional beneficial outcome (NNTB) or the number needed to treat for an additional harmful outcome (NNTH), and the absolute and relative per cent change in the comments column of the 'Summary of findings' tables.

Subgroup analysis and investigation of heterogeneity

No subgroup analysis was planned or performed.

Sensitivity analysis

We performed a sensitivity analysis to determine the robustness of the treatment effect for any outcomes showing moderate to considerable heterogeneity (greater than 30%), by comparing random‐effects methods to fixed‐effect methods. We had further planned to stratify studies by risk of bias, but the number of studies included in the review did not make this analysis possible

Results

Description of studies

See Figure 1 for flow diagram of search results.

1.

Study flow diagram – 2018 update

Results of the search

We identified two randomized controlled trials (Van Deusen 1987; Wang 2008), and five controlled clinical trials (Jianjiang 1999; Kirsteins 1991a; Kirsteins 1991b; Lee 2006; Shin 2015), involving 345 participants, from the literature search. One trial provided a description of two separate CCTs (Kirsteins 1991a; Kirsteins 1991b).

Included studies

Descriptions of the included studies can be found in the 'Characteristics of included studies' table. The seven studies, with a total of 345 participants with rheumatoid arthritis, were conducted in the United States, China, and Korea. All studies compared a form of Tai Chi exercise, either alone (Kirsteins 1991a; Kirsteins 1991b; Lee 2006; Shin 2015; Wang 2008); or in combination with other treatments (Jianjiang 1999; Van Deusen 1987), to a control group that either received no exercise (Jianjiang 1999; Kirsteins 1991a; Kirsteins 1991b; Lee 2006), received health or exercise information to take home (Shin 2015; Van Deusen 1987), or received an alternative exercise therapy (Wang 2008). Some of the control group participants did receive information on self‐range of motion exercise or RA education (Kirsteins 1991a; Kirsteins 1991b; Van Deusen 1987; Wang 2008) but none of the control groups performed Tai Chi exercises during the study period. Participants ranged in age from 15 to 80 years; 83% of those studied were women. Two studies included women only (Lee 2006; Shin 2015). Disease duration varied from less than one year to more than 35 years among participants. Duration of Tai Chi exercise instruction and outcome assessment ranged from 8 to 12 weeks; a couple of studies had a second follow‐up at 24 weeks (Van Deusen 1987; Wang 2008). Most studies were performed in tertiary care hospital settings or private rheumatology practices.

Excluded studies

An abstract by Zhu and colleagues was awaiting further assessment at the time of the original review (Zhu 1999a). We were not able to locate this study for further assessment for this review.

A total of seven studies were excluded due to lack of a non‐intervention control group (Bang 2014; Dogra 2015; Hammond 2013; Hun 2013; Lee 2012; Uhlig 2005); or our inability to extract data for participants with RA only (Callahan 2016).

Risk of bias in included studies

We assessed the studies identified for this updated review using standard Cochrane methodology for risk of bias. Risk of bias is summarized below, and can be seen graphically in Figure 2 and Figure 3.

2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies

3.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study

Allocation

The majority of the studies provided insufficient information on the allocation of treatments. We judged one study at low risk of selection bias because it described a process using opaque envelopes and appropriate random sequence generation (Wang 2008). We judged random sequence generation at high risk of bias for four studies (Jianjiang 1999; Kirsteins 1991a; Kirsteins 1991b; Shin 2015), and unclear for two (Lee 2006; Van Deusen 1987).

Blinding

We judged all studies at high risk of performance bias. It was not possible to blind participants in the Tai Chi treatment and control arms. Most studies did not sufficiently describe their procedure for blinding of outcomes assessors; however, two of the studies described blinding of outcomes assessors, and since they contributed mostly objective outcomes, we considered them at low risk (Kirsteins 1991a; Kirsteins 1991b).

Incomplete outcome data

We assessed four studies at high risk for attrition bias due to a large number of withdrawals during the course of the study (Kirsteins 1991a; Lee 2006; Shin 2015; Van Deusen 1987).

Selective reporting

We assessed most studies at low risk for reporting bias because outcomes described in the methods section were reported in the results.

Other potential sources of bias

Other potential sources of bias included: two sequential studies in which a few of the participants from the first study were also in the second study, possibly in opposite arms (Kirsteins 1991a; Kirsteins 1991b), and concerns about lack of validity of the measurement tools for exercise‐rest self report in another study (Van Deusen 1987).

Effects of interventions

See: Table 1

A total of 345 participants provided data of which 180 were assigned to Tai Chi and 165 to other treatments.

Tai Chi versus no Tai Chi

Major outcomes

A summary of the results for the major outcomes is listed in Table 1.

Change in pain

Two trials (81 participants) measured pain as change from baseline to end of treatment (Lee 2006; Wang 2008).

Results from two studies showed a statistically and clinically meaningful improvement in pain in the Tai Chi group compared to the no Tai Chi control group (mean difference (MD) ‐2.15 cm on VAS, 95% confidence interval (CI) ‐3.19 to ‐1.11; Analysis 1.1; very low‐quality evidence; absolute reduction in pain score 22%. 95% CI 11% to 32%; number needed to for an additional beneficial outcome (NNTB) 3, 95% CI 2 to 5). We are uncertain if Tai Chi improves self‐reported pain.

1.1. Analysis.

Comparison 1 Major efficacy: clinical and functional outcomes, Outcome 1 Change in pain, Visual Analog Scale (VAS) at 12 weeks.

We back‐translated the standardized mean difference (SMD) by multiplying the SMD by the standard deviation of the control group at baseline from the most representative trial (Lee 2006).

Change in measure of disease activity

Very low‐quality evidence from one trial showed a reduced DAS‐28‐ESR score among the Tai Chi group compared to the control group (MD ‐0.40, 95% CI ‐1.10 to 0.30; 1 RCT, 43 participants; Analysis 1.2; absolute difference ‐4%, 95% CI ‐11% to 3%; Shin 2015). This may not be an important difference. We are uncertain whether Tai Chi improves disease activity as measured by the DAS‐28‐ESR.

1.2. Analysis.

Comparison 1 Major efficacy: clinical and functional outcomes, Outcome 2 Change in disease activity (DAS‐28‐ESR) at 12 weeks.

Change in function

Two studies measured change from baseline, using the Health Assessment Questionnaire (HAQ; lower score denotes better function). The Tai Chi groups reported better function than the no Tai Chi groups (MD ‐0.33, 95% CI ‐0.79 to 0.12; 2 studies, 63 participants; Analysis 1.3; very low‐quality evidence; absolute difference ‐11%, which might be clinically meaningful (Shin 2015; Wang 2008). We are uncertain whether Tai Chi improves function.

1.3. Analysis.

Comparison 1 Major efficacy: clinical and functional outcomes, Outcome 3 Change in function, Health Assessment Questionnaire (HAQ) at 12 weeks.

In sensitivity analysis for possible heterogeneity (I²=82%), fixed‐effect analysis showed a reduced HAQ score among the Tai Chi groups when compared to the control groups (MD ‐0.20, 95% CI ‐0.34 to ‐0.06), which was statistically significant but might not be clinically meaningful.

Radiographic progression

None of the studies reported this outcome.

Short‐term serious adverse events

None of the studies reported this outcome.

Long‐term adverse effects or toxicity

None of the studies reported this outcome.

Withdrawals ‐ overall

None of the studies reported on withdrawals due to adverse events.

All seven studies (345 participants) provided data on withdrawals from any cause (Jianjiang 1999; Kirsteins 1991a; Kirsteins 1991b; Lee 2006; Shin 2015; Van Deusen 1987; Wang 2008). The Tai Chi groups reported 19 out of 180 (11%) withdrawals; the control groups (no Tai Chi) reported 49 out of 165 (30%) withdrawals. Pooled analysis showed the Tai Chi groups had fewer withdrawals than the no Tai Chi groups (risk ratio (RR) 0.40, 95% CI 0.19, 0.86; Analysis 2.1; low‐quality evidence; absolute difference 17% fewer, 95% CI ‐30% to ‐3%); however, it may not be clinically meaningful.

2.1. Analysis.

Comparison 2 Major safety: withdrawals, Outcome 1 Withdrawals Overall.

In sensitivity analysis for possible heterogeneity (I²=41%), fixed effects analysis showed a similar RR of 0.36 (95% CI 0.22 to 0.58) among the Tai Chi group when compared to the control group, which was statistically significant and clinically meaningful.

Minor outcomes

ACR20

ACR20 was reported in one study (Wang 2008, N=20) and found to be higher in the Tai Chi group (RR=11.0 (95% CI 0.69 to 175.86); Analysis 3.1) with an absolute difference of 50% (95% CI 18% to 82%), with very low quality of evidence. While not a statistically significant result, this difference could be clinically meaningful. We are uncertain whether Tai Chi improves symptoms of RA.

3.1. Analysis.

Comparison 3 Minor efficacy: clinical and functional outcomes, Outcome 1 ACR20 at 12 weeks.

Number of tender joints

Three studies (73 participants) measured joint tenderness using the Ritchie index (Kirsteins 1991a; Kirsteins 1991b; Wang 2008). Meta‐analysis found MD ‐0.01 (95% CI ‐0.32 to 0.29; Analysis 3.2). This may not have been clinically important.

3.2. Analysis.

Comparison 3 Minor efficacy: clinical and functional outcomes, Outcome 2 Number of tender joints.

Number of swollen joints

Four studies (113 participants) measured swollen joints (Kirsteins 1991a; Kirsteins 1991b; Shin 2015; Wang 2008). Meta‐analysis found MD 0.76 joints (95% CI ‐1.27 to 2.79; Analysis 3.3). This may not have been clinically important.

3.3. Analysis.

Comparison 3 Minor efficacy: clinical and functional outcomes, Outcome 3 Number of swollen joints.

Physician global assessment

None of the studies reported this outcome.

Patient global assessment

None of the studies reported this outcome.

Acute phase reactants

None of the studies reported this outcome.

Range of motion

One study (33 participants) measured ankle plantar flexion (MD 24 degrees, 95% CI 3.34 to 44.66; Analysis 4.1), and lower extremity flexion (MD 34 degrees, 95% CI 10.79 to 57.21; Analysis 4.2; Van Deusen 1987). We are uncertain whether Tai Chi improves these range of motion outcomes.

4.1. Analysis.

Comparison 4 Minor efficacy: range of motion, Outcome 1 Ankle plantar flexion (degrees).

4.2. Analysis.

Comparison 4 Minor efficacy: range of motion, Outcome 2 Lower extremity flexion: hip, knee, ankle dorsal flexion (degrees).

Van Deusen 1987 also measured shoulder flexion (MD 21 degrees, 95% CI ‐17.56 to 59.56; Analysis 4.3), and shoulder internal and external rotation (MD 42 degrees, 95% CI ‐7.97 to 91.97; Analysis 4.4). The results were inconclusive and may not have been clinically important.

4.3. Analysis.

Comparison 4 Minor efficacy: range of motion, Outcome 3 Shoulder flexion (degrees).

4.4. Analysis.

Comparison 4 Minor efficacy: range of motion, Outcome 4 Shoulder internal and external rotation (degrees).

Grip strength

Two studies (50 participants) measured grip strength (Kirsteins 1991a; Wang 2008). The results were inconclusive and may not have been clinically important (MD ‐3.12, 95% CI ‐9.99 to 3.76); Analysis 3.4). One study (Kirsteins 1991b), for which grip strength was reported previously, was removed from this analysis due to different measurement technique as described in the study.

3.4. Analysis.

Comparison 3 Minor efficacy: clinical and functional outcomes, Outcome 4 Grip strength.

50‐foot walking test

Three studies (68 participants) measured the amount of time it took to walk 50 feet (Kirsteins 1991a; Kirsteins 1991b; Wang 2008. While we did not plan to measure this outcome, we felt it was an important part of the functional picture. The results were inconclusive, and may not have been clinically important (MD 0.17 seconds, 95% CI ‐1.06 to 1.40; Analysis 3.5).

3.5. Analysis.

Comparison 3 Minor efficacy: clinical and functional outcomes, Outcome 5 50 foot walk (seconds).

Short‐term adverse events

While no numeric data were presented, the authors of two of the studies report that approximately one‐third of the participants complained of soreness in the knee, shoulder, or lower back during the first three weeks of the studies (Kirsteins 1991a; Kirsteins 1991b). The pain eventually subsided for all participants, and they were able to continue the program without altering practice time or quality of movement. The only exception was one participant, who complained of knee pain. The authors also reported that there were complaints of soreness in the quadriceps, shoulders, or both, after the three 90‐minute sessions during the course of the intervention; they did not report the number of participants with this complaint. Because these joint and muscle pain complaints diminished during the course of the study and did not cause withdrawals, they were not considered to be important by the investigators of the study. One participant from one study complained of leg cramps and ankle soreness after the first class, and therefore switched from the test group to the control group upon his physicians's advice (Kirsteins 1991b). The participant was not examined, therefore it remains unclear whether this was due to a joint flare‐up.

Discussion

Summary of main results

This systematic review summarized the results of seven controlled trials on the effectiveness and safety of Tai Chi for people with rheumatoid arthritis (RA (Table 1)). The Tai Chi programs ran from eight (Van Deusen 1987), to 12 weeks (Lee 2006; Wang 2008).

Because of very low‐quality evidence, we are uncertain if Tai Chi improves self‐reported pain better than no Tai Chi, measured on a visual analog scale (VAS), despite a mean difference (MD) of ‐2.15 (95% confidence interval (CI) ‐3.19 to ‐1.11), which may be clinically important. Because of very low‐quality evidence and inconclusive results, we are uncertain whether Tai Chi improves disease activity, measured by the Disease Activity Score (DAS‐28‐ESR; (MD ‐0.40, 95% CI ‐1.10 to 0.30)), or function, measured by the Health Assessment Questionnaire (HAQ; (MD ‐0.33, 95% CI ‐0.79 to 0.12)). Improvement in function might be clinically meaningful, since the between group difference of 0.33 exceeded the minimum clinically important difference (MCID) of 0.22 for a change in an individual's score (Wells 1993).

In terms of harms, apart from reporting on overall study withdrawals, which were less likely among Tai Chi participants, this review did not find evidence on the safety of Tai Chi exercise for people with RA. While there was some qualitative description of adverse effects, safety outcomes of interest were not quantitatively reported. The safety of Tai Chi is well‐reported in other studies (Wayne 2014), which have shown that exercise does not exacerbate the clinical disease activity of people with RA (Hansen 1993; Nordemar 1981). Unlike the other studies that used non‐weight bearing exercises, Tai Chi is a weight‐bearing exercise; therefore, it has a potential advantage over traditional exercise programs of stimulating bone formation and strengthening connective tissues (Gerber 1986).

In addition to the results from the major outcomes, the finding from one small trial that Tai Chi might benefit shoulder and ankle range of motion (ROM) should be considered in a broader context. Range of motion is not part of the OMERACT core set, partially due to lack of consensus about its validity (methods of measuring range of motion have poor validity and reliability) and the lack of importance to both people with RA and clinicians. Less than full joint range of motion is required to perform many activities of daily life (Magee 1992). Therefore, improvement in range of motion may not always translate into improved function in daily activities.

Overall completeness and applicability of evidence

The number of trials that met the criteria for inclusion was fairly low (N = 7), and apart from reported overall withdrawals, all outcomes only contained data from one or two studies. In addition, there were significantly more withdrawals in the control groups of these studies (17% absolute risk difference). It is uncertain whether this difference in withdrawals is due to greater enjoyment and perceived benefit of participants in the Tai Chi group. This interpretation is supported by one CCT (N = 46) that found statistically significantly greater enjoyment and perceived benefit of Tai Chi by study participants (Van Deusen 1987). It is possible that tolerability of Tai Chi was better than control, demonstrated by a lower overall withdrawal rate from Tai Chi groups than from the control groups. This is an important finding that demonstrates that the intervention may be tolerated well by participants.

Few studies examined the improvement in RA symptoms and disease activity (usually one to two studies) when comparing Tai Chi to a comparator of no Tai Chi, different exercises, or a behavioral intervention, so we can not be confident that Tai Chi makes a clinically meaningful improvement in these important outcomes over a comparator. No studies reported using the ACR50, which was a main outcome measure of interest, and only one study reported using the DAS‐28‐ESR. There might be an improvement in self‐reported pain and physical function (measured by HAQ) with Tai Chi, however this was also limited by the number of studies providing this evidence, and the uncertainty of the results.

This evidence, with its limitations, has applicability to the groups on which it is based, primarily, women with RA. Tai Chi might improve pain and have less withdrawals versus comparator among people with RA. Tai Chi might be an option for people with RA, who are interested in supplementing their usual care with a non‐pharmacological intervention. More robust evidence is needed to evaluate its role in the management of RA.

Quality of the evidence

The quality of the evidence was low or very low for all major outcomes, mostly downgraded for concerns about imprecision and high risk of bias, due to blinding and participant withdrawals. Double‐blinding is acknowledged as one of the most important factors in reducing bias, but was not conducted for any of the seven included trials (Schulz 1995). Therefore, participants in the experimental and control groups were aware of their status. The experimental group may have had greater improvements due to both participant and evaluator bias. It is not feasible to blind participants in exercise studies to the fact they are participating in an exercise intervention. However, it is possible to blind evaluators to participants group allocation, was done in two of the included studies, and this should be done in future studies (Kirsteins 1991a; Kirsteins 1991b).

The dose of Tai Chi intervention was not clear in the trials included in this review. In the Kirsteins studies, test participants were instructed to practice the exercises on their own for around 20 minutes/day, as tolerated, and keep a daily log (Kirsteins 1991a; Kirsteins 1991b). However, there was poor compliance with the participant practice logs, and less than one‐third of the participants submitted them. In Van Deusen 1987, the experimental participants were encouraged to practice the range of motion (ROM) dance sequence and relaxation techniques during daily periods of rest on a daily basis at home in addition to any specific exercises recommended by their physician or therapist. It is unclear whether the participants followed these recommendations, and if so, for how many minutes a day Tai Chi exercises were practiced.

Furthermore, the control group in the Van Deusen 1987 trial was given a brochure, which explained the ROM dance Program and the research project. Although they did not receive any specific instructions regarding home rest and exercise, their knowledge of the program and exercises may have produced beneficial effects on the RA of these participants. This would make it difficult to detect differences attributable solely to Tai Chi in the experimental group. Other studies also gave control group participants information about exercise (Shin 2015), or provided education sessions about stretching and wellness (Wang 2008).

Overall, poor reporting of the content of some Tai Chi interventions, unknown quality of Tai Chi instruction, and questionable fidelity to reported protocols by both the intervention and control groups, could lead to uncertainty of comparability between studies. Due to the small number of studies and overall number of people in the studies, we were not able conduct any relevant subgroup analyses.

Potential biases in the review process

For the updated review, we did not search the Chinese databases or the Allied and Complementary Medicine Database (AMED), as was done in the original review. Therefore, there may have been studies that were missed. In addition, there was one study identified in the original review for which we were still not able to find the full text (Zhu 1999a).

Agreements and disagreements with other studies or reviews

This review did not differ substantially from other systematic (Lee 2007; Siegel 2017), and literature reviews (Uhlig 2012; Wang 2012), which showed moderate effect, albeit with limited evidence, on some health outcomes for people with RA who used Tai Chi, including pain, physical function, and quality of life, with little report of harm.

The finding of improvement on a pain VAS with Tai Chi is similar to other studies that have reported improvement in pain in other conditions with Tai Chi, including low back pain (Cho 2014), fibromyalgia (Jones 2012; Wang 2018), osteoarthritis (Yan 2013; Ye 2014), and arthritis in general (Callahan 2016).

Authors' conclusions

Implications for practice.

Based on very low‐quality evidence, this systematic review suggests that Tai Chi has uncertain effects on the overall disease activity of rheumatoid arthritis (RA), in terms of self‐reported pain, disease activity, and physical function.

Withdrawals from study were greater in the control groups than the Tai Chi groups, a result that was significant and clinically meaningful though based on low quality evidence. Adverse events were described qualitatively in the narrative of two of the studies, and included reports of joint and muscle soreness, and cramps in a small number of participants.

Implications for research.

We identified several research gaps that have implications for future research. Studies of Tai Chi in RA (trials and longitudinal studies) need to report standardized RA efficacy outcomes, including ACR50, ACR20, disease activity, and remission, using validated measures, such as DAS‐28, functional ability using the Health Assessment Questionnaire (HAQ) or similar outcome measures, quality of life, pain, and radiographic progression. Longitudinal, observational studies of Tai Chi in RA that report harms data (serious adverse events, withdrawals, etc.) on larger groups or people followed for a longer duration can improve our understanding of the benefit/risk ratio of Tai Chi. Therefore, well‐designed, large trials are needed to elucidate the benefits/harms of Tai Chi in RA, focusing on these important outcomes. As placebo‐controlled exercise trials are not feasible, more trials of Tai Chi vs. other exercises are needed, and in addition, longer‐term outcomes need to be assessed. More diverse participant groups by age, sex, race/ethnicity and baseline RA disease activity and functional status should be examined in these studies, and results analyzed by important subgroups. This will help understand whether and how the effect of Tai Chi is impacted by important biological and other factors.

Randomized trials comparing Tai Chi to another form of physical exercise (Yoga, aerobic training, stretching exercises, pool therapy, etc.) in RA are recommended. This would allow estimation of effect size of Tai Chi versus an active comparator. Also, when designing future studies, it is important to consider the quality of the measurement and reporting of the dose (frequency, intensity, and duration) of Tai Chi interventions. In addition, in order to minimize detection bias, consistent blinding of outcomes assessors is recommended, since participants cannot be blinded to the intervention.

Most current studies of Tai Chi were limited mostly to 8 to 12 weeks, which may not have allowed the observation of the full effect of Tai Chi, and provides no insight into sustainability of such an intervention, or durability of the benefit past three months. Future trials should be of longer duration, preferably six months or longer, which also assess whether the benefits of Tai Chi noted at a shorter duration of three to six months are sustained over longer periods of time. Extension studies, after the end of the comparative trial phase, can also improve our understanding of the likelihood of people continuing the intervention beyond a short period of time.

What's new

Date	Event	Description
9 December 2018	New citation required but conclusions have not changed	Conclusions from previous review not changed; additional major outcome added, new author team
9 December 2018	New search has been performed	Updated search, 3 new studies added

Appendices

Appendix 1. Original search strategy

Identifying arthritis patients:
 1. exp ARTHRITIS/
 2 exp RHEUMATOID ARTHRITS/
 3. 1 or 2

Identifying intervention:
 4 exp Tai Ji/ or tai chi.mp.
 5 tai chi.tw.
 6 tai chi chuan.tw.
 7 tai‐chi.tw.
 8 or/4‐7

Identifying trials:
 9. clinical trial.pt.
 10. randomized controlled trial.pt.
 11. tu.fs.
 12. dt.fs.
 13. random$.tw.
 14. placebo$.tw.
 15. ((sing$ or doubl$ or tripl$) adj (masked or blind$)).tw.
 16. sham.tw.
 17. or/9‐16
 18. 3 and 8 and 17
 19. from 18 keep 1‐2

Appendix 2. Updated search strategy (2013)

Database: Cochrane Library Issue 4 of 12, Apr 2013

Search Name: Tai chi for treating rheumatoid arthritis

ID Search

#1 tai ji

#2 tai chi

#3 tai chi chuan

#4 MeSH descriptor: [Tai Ji] explode all trees

#5 #1 or #2 or #3 or #4

#6 MeSH descriptor: [Arthritis, Rheumatoid] explode all trees

#7 arthritis

#8 #6 or #7

#9 #5 and #8

Database: Ovid MEDLINE(R) In‐Process & Other Non‐Indexed Citations and Ovid MEDLINE(R) <1946 to Present>

Search Strategy:

‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐

1 exp Arthritis/ (192765)

2 exp arthritis, rheumatoid/ (96470)

3 (arthritis adj2 rheumat$).tw. (74796)

4 (felty$ adj2 syndrome).tw. (666)

5 (caplan$ adj2 syndrome).tw. (112)

6 rheumatoid nodule.tw. (228)

7 (sjogren$ adj2 syndrome).tw. (10808)

8 still$ disease.tw. (1536)

9 or/1‐8 (210347)

10 exp Tai Ji/ (524)

11 tai chi.tw. (737)

12 tai ji.tw. (10)

13 tai chi chuan.tw. (113)

14 Ai Chi.tw. (6)

15 Taijiquan.tw. (9)

16 Taijizhang.tw. (0)

17 or/10‐16 (806)

18 9 and 17 (61)

Database: Embase Classic+Embase <1947 to 2013 May 23>

Search Strategy:

‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐

1 exp arthritis, rheumatoid/ (153370)

2 (arthritis adj2 rheumat$).tw. (109141)

3 (felty$ adj2 syndrome).tw. (906)

4 (caplan$ adj2 syndrome).tw. (237)

5 rheumatoid nodule.tw. (327)

6 (sjogren$ adj2 syndrome).tw. (14665)

7 still$ disease.tw. (2185)

8 or/1‐7 (179590)

9 exp Tai Chi/ (1144)

10 tai chi chuan.tw. (139)

11 tai ji.tw. (13)

12 tai chi.tw. (983)

13 Taijiquan.tw. (28)

14 Taijizhang.tw. (0)

15 taichi.mp. (14)

16 taichi.tw. (13)

17 or/9‐16 (1385)

18 8 and 17 (58)

Database: AMED (Allied and Complementary Medicine) <1985 to May 2013>

Search Strategy:

‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐

1 exp Arthritis/ (4428)

2 exp Arthritis rheumatoid/ (1761)

3 ((rheumatoid or reumatoid or revmatoid or rheumatic or reumatic or revmatic or rheumat$ or reumat$ or revmarthrit$) adj3 (arthrit$ or artrit$ or diseas$ or condition$ or nodule$)).tw. (2538)

4 (felty$ adj2 syndrome).tw. (1)

5 (caplan$ adj2 syndrome).tw. (0)

6 (sjogren$ adj2 syndrome).tw. (31)

7 (sicca adj2 syndrome).tw. (8)

8 still$ disease.tw. (3)

9 or/1‐8 (5279)

10 exp Tai chi/ (217)

11 tai chi chuan.tw. (55)

12 Taijiquan.tw. (13)

13 tai ji.tw. (7)

14 tai chi.tw. (278)

15 taichi.tw. (1)

16 Taijizhang.tw. (0)

17 or/10‐16 (292)

18 9 and 17 (19)

Conference Proceedings Citation Index –Science (CPCI‐S) – 1990 to present

Topic=(tai chi or tai ji) AND Topic=(arthritis)

Timespan=All years. Databases=CPCI‐S.

Appendix 3. Updated search strategy (2018)

Search Strategies

C1 ‐ Database: Ovid MEDLINE(R) and Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations and Daily <1946 to September 24, 2018>

Search Strategy:

1 exp Arthritis/ (240447)

2 exp arthritis, rheumatoid/ (106361)

3 (arthritis adj2 rheumat$).tw. (97534)

4 (felty$ adj2 syndrome).tw. (711)

5 (caplan$ adj2 syndrome).tw. (120)

6 rheumatoid nodule.tw. (262)

7 (sjogren$ adj2 syndrome).tw. (14026)

8 still$ disease.tw. (2072)

9 or/1‐8 (268174)

10 exp Tai Ji/ (907)

11 tai chi.tw. (1418)

12 tai ji.tw. (30)

13 tai chi chuan.tw. (176)

14 Ai Chi.tw. (17)

15 Taijiquan.tw. (14)

16 Taijizhang.tw. (0)

17 or/10‐16 (1553)

18 9 and 17 (105)

19 limit 18 to ed=20130524‐20180925 (43)

C1 ‐ Database: Embase Classic+Embase <1947 to 2018 September 24>

Search Strategy:

1 exp arthritis, rheumatoid/ (201800)

2 (arthritis adj2 rheumat$).tw. (151508)

3 (felty$ adj2 syndrome).tw. (905)

4 (caplan$ adj2 syndrome).tw. (207)

5 rheumatoid nodule.tw. (362)

6 (sjogren$ adj2 syndrome).tw. (19271)

7 still$ disease.tw. (3034)

8 or/1‐7 (236843)

9 exp Tai Chi/ (2437)

10 tai chi chuan.tw. (214)

11 tai ji.tw. (59)

12 tai chi.tw. (2006)

13 Taijiquan.tw. (39)

14 Taijizhang.tw. (0)

15 taichi.mp. (45)

16 taichi.tw. (42)

17 or/9‐16 (2817)

18 8 and 17 (89)

19 limit 18 to dd=20130524‐20180925 (18)

C1 ‐ Database: EBM Reviews – Cochrane Central Register of Controlled Trials <August 2018>

Search Strategy:

1 exp Arthritis/ (12641)

2 exp arthritis, rheumatoid/ (5172)

3 (arthritis adj2 rheumat$).tw. (9639)

4 (felty$ adj2 syndrome).tw. (3)

5 (caplan$ adj2 syndrome).tw. (0)

6 rheumatoid nodule.tw. (3)

7 (sjogren$ adj2 syndrome).tw. (507)

8 still$ disease.tw. (29)

9 or/1‐8 (17984)

10 exp Tai Ji/ (288)

11 tai chi.tw. (770)

12 tai ji.tw. (16)

13 tai chi chuan.tw. (80)

14 Ai Chi.tw. (13)

15 Taijiquan.tw. (71)

16 Taijizhang.tw. (0)

17 or/10‐16 (878)

18 9 and 17 (36)

Data and analyses

Comparison 1. Major efficacy: clinical and functional outcomes.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Change in pain, Visual Analog Scale (VAS) at 12 weeks	2	81	Std. Mean Difference (IV, Random, 95% CI)	‐0.95 [‐1.41, ‐0.49]
2 Change in disease activity (DAS‐28‐ESR) at 12 weeks	1	43	Mean Difference (IV, Random, 95% CI)	‐0.4 [‐1.10, 0.30]
3 Change in function, Health Assessment Questionnaire (HAQ) at 12 weeks	2	63	Mean Difference (IV, Random, 95% CI)	‐0.33 [‐0.79, 0.12]

Comparison 2. Major safety: withdrawals.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Withdrawals Overall	7	345	Risk Ratio (M‐H, Random, 95% CI)	0.40 [0.19, 0.86]

Comparison 3. Minor efficacy: clinical and functional outcomes.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 ACR20 at 12 weeks	1	20	Odds Ratio (M‐H, Fixed, 95% CI)	21.0 [0.97, 453.91]
2 Number of tender joints	3	73	Mean Difference (IV, Random, 95% CI)	‐0.74 [‐3.12, 1.65]
3 Number of swollen joints	3	70	Mean Difference (IV, Random, 95% CI)	2.59 [‐0.18, 5.35]
4 Grip strength	2	50	Mean Difference (IV, Random, 95% CI)	‐3.12 [‐9.99, 3.76]
5 50 foot walk (seconds)	3	68	Mean Difference (IV, Random, 95% CI)	0.17 [‐1.06, 1.40]

Comparison 4. Minor efficacy: range of motion.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Ankle plantar flexion (degrees)	1	33	Mean Difference (IV, Random, 95% CI)	24.0 [3.34, 44.66]
2 Lower extremity flexion: hip, knee, ankle dorsal flexion (degrees)	1	33	Mean Difference (IV, Random, 95% CI)	34.0 [10.79, 57.21]
3 Shoulder flexion (degrees)	1	33	Mean Difference (IV, Random, 95% CI)	21.0 [‐17.56, 59.56]
4 Shoulder internal and external rotation (degrees)	1	33	Mean Difference (IV, Random, 95% CI)	42.00 [‐7.97, 91.97]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Jianjiang 1999.

Methods	Controlled clinical trial with no matching
Participants	Adults (N = 68) with rheumatoid arthritis in China; age range (15 to 56 years); 57% women; disease duration (0.5 to 16 years)
Interventions	Study duration: 8 weeks Control group (N = 33) received Shan Pi Tang decoction Experimental group (N = 35) received Shan Pi Tang decoction combined with physical activity therapy: 1. Health education (RA knowledge and method of exercises); 2. Daily exercises (every morning for one hour; included slow running, walk, gymnastics, Tai Chi); 3. Massage and hot compress applied to affected area every evening for 30 minutes and 15 minutes, respectively.
Outcomes	Recovery rate, markedly effective rate, effective rate, relapse rate (how many participants were tested at posttest), and withdrawal from study
Notes	There were three groups in this study. Group 1 received oral Lei Gong Teng and no exercise therapy. Group 2 received oral Shan Pi Tang decoction and no exercise therapy. Group 3 received oral Shan Pi Tang along with the health education, exercise therapy, massage and hot compress. We included the comparison of group 2 versus group 3 since the difference between these groups is the exercise based on Tai Chi, massage and hot compress.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	controlled clinical trial ‐ "100 patients with RA were randomly divided in A, B, and C groups"
Allocation concealment (selection bias)	Unclear risk	insufficient information
Blinding of participants and personnel (performance bias) All outcomes	High risk	blinding not possible
Blinding of outcome assessment (detection bias) All outcomes	High risk	does not state whether assessors were blinded but participants were not
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	insufficient information
Selective reporting (reporting bias)	Unclear risk	insufficient information
Other bias	Unclear risk	insufficient information

Kirsteins 1991a.

Methods	Controlled clinical trial; enrolled participants were divided into groups by matching their age and performance on a functional assessment survey
Participants	Ambulatory adults (N = 47) from three different rheumatology private practices in the United States, diagnosed with rheumatoid arthritis (American Rheumatism Association functional class II or III) after age 18, who could walk without any assistive devices and were on a stable regimen of medications for a sufficient time for maximal results. All participants were graduates of an Arthritis Foundation Self‐help class and had been instructed in self‐range of motion exercises. Age range (37 to 70 years); 89% women
Interventions	Study duration: 12 weeks with 10 week exercise instruction period Experimental group (N = 25): Tai Chi Chuan exercises for 1 hour session once a week for 10 weeks. Each class consisted of a 15 to 20 minute warm‐up period, followed by a 5 minute rest period. The rest of the class was devoted to a repetition of a series of 15 movements derived from the Yang Style Tai Chi Chian Short Form. The movements, which were adapted for people with RA, involved a combination of stepping and transferring weight with deep breathing, relaxation, and arm movements. The experimental group was also instructed to practice the Tai Chi exercises on their own for approximately 20 minutes a day as tolerated and keep a daily log. Control group (N = 22): no Tai Chi Chuan exercise instruction
Outcomes	Outcomes tested one week before and one week after 10‐week intervention period. Outcomes included: number of tender and swollen joints, time to walk 50 feet, right and left handgrip strength
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	‐ "divided into a test group (25participants) and a control group (22 participants) by matching their age and performance on a functional assessment survey"
Allocation concealment (selection bias)	Unclear risk	insufficient information
Blinding of participants and personnel (performance bias) All outcomes	High risk	blinding not possible between Tai Chi and control arm
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"All the tests were performed by two junior medical school students, who were trained by one of the authors and blinded to the group status of each" ‐ Participants not blinded but study contributed mostly non‐patient reported outcomes
Incomplete outcome data (attrition bias) All outcomes	High risk	missing data proportion (˜ 35%) compared with observed event risk enough to induce clinically relevant bias in intervention effect estimate
Selective reporting (reporting bias)	Low risk	all outcomes described in methods reported in results tables
Other bias	Unclear risk	Insufficient information

Kirsteins 1991b.

Methods	Controlled clinical trial; enrolled participants were divided into groups by matching their age and performance on a functional assessment survey
Participants	Ambulatory adults (N = 28) from three different rheumatology private practices in the United States, diagnosed with rheumatoid arthritis (American Rheumatism Association functional class II or III) after age 18, who could walk without any assistive devices and were on a stable regimen of medications for a sufficient time for maximal results. All participants were graduates of an Arthritis Foundation Self‐Help class and had been instructed in self‐range of motion exercises. Age range (38 to 72 years); 75% women
Interventions	Study duration: 12 weeks with 10 week exercise instruction period Experimental group (N = 18): Tai Chi Chuan exercises for 1 hour session twice a week for 10 weeks. Each class consisted of a 15 to 20 minute warm‐up period, followed by a 5 minute rest period. The rest of the class was devoted to a repetition of a series of 15 movements derived from the Yang Style Tai Chi Chian Short Form. The movements, which were adapted for people with RA, involved a combination of stepping and transferring weight with deep breathing, relaxation, and arm movements. The experimental group was also instructed to practice the Tai Chi exercises on their own for approximately 20 minutes a day as tolerated and keep a daily log. Control group (N = 10): no Tai Chi Chuan exercise instruction
Outcomes	Outcomes tested one week before and one week after 10‐week intervention period. Outcomes included: number of tender and swollen joints, time to walk 50 feet, right and left handgrip strength
Notes	Some of the participants from Kirsteins 1991a also participated in the second study as follows: From the control group of the first study, 4 participated in the experimental group and 2 participated in the control group. From the experimental group of the first study, 9 participated in the experimental group and 4 participated in the control group of the second study
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	‐ "patients were divided into a test group (18 patients) and a control group (10 patients)"
Allocation concealment (selection bias)	Unclear risk	insufficient information
Blinding of participants and personnel (performance bias) All outcomes	High risk	blinding not possible between Tai Chi and control arm
Blinding of outcome assessment (detection bias) All outcomes	Low risk	"All the tests were performed by two junior medical school students, who were trained by one of the authors and blinded to the group status of each"‐ Participants not blinded but study contributed mostly non‐patient reported outcomes
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	missing data proportion (˜ 20%) compared with observed event risk enough to induce clinically relevant bias in intervention effect estimate
Selective reporting (reporting bias)	Low risk	all outcomes described in methods reported in results tables
Other bias	High risk	cross over of participants from one arm to the other from study 1 to study 2

Lee 2006.

Methods	Controlled clinical trial with no matching;
Participants	Ambulatory women (N = 61) with rheumatoid arthritis with more than 6 months treatment history at a hospital setting in Korea; mean age of 50 years in both experimental and control groups; mean disease duration (5.5 years in experimental group; 3.8 years in control group)
Interventions	Study duration: 12 weeks Started with 40 participants in each group before dropouts. Pre/post testing performed on remaining participants. Experimental group (N = 32): Tai Chi instruction once a week for 50 minutes (15 minutes warm up, 20 minutes Tai Chi exercise, 15 minutes cool down). Developed by Dr. Lam in 1997, this Tai Chi exercise regimen is composed of 6 fundamental moves and 6 intense moves Control group (N = 29): no Tai Chi instruction
Outcomes	Pain – visual analog scale (VAS), fatigue scale, activities of daily living, sense of balance
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	insufficient information – randomization method not mentioned
Allocation concealment (selection bias)	Unclear risk	insufficient information – allocation method not mentioned
Blinding of participants and personnel (performance bias) All outcomes	High risk	blinding not possible between Tai Chi and control arm
Blinding of outcome assessment (detection bias) All outcomes	High risk	blinding of outcome assessment not mentioned, but participants not blinded and contributed self‐reported outcome data
Incomplete outcome data (attrition bias) All outcomes	High risk	overall 24% dropout rate; 20% in the intervention arm and 28% in the control arm
Selective reporting (reporting bias)	Unclear risk	all outcomes described in methods reported in results tables
Other bias	Unclear risk	insufficient information

Shin 2015.

Methods	Controlled clinical trial with case‐matched controls
Participants	Ambulatory women (N = 56) consecutively recruited from rheumatology department of hospital in South Korea; all satisfied the ACR 1987 revised classification criteria for rheumatoid arthritis; controls matched by age and BMI; 13 participants dropped out of control group so analysis done on remaining 43 participants; mean age (64 years in Tai Chi group, 63 in control group); mean disease duration (10.3 years in intervention group, 15.4 in control group); baseline DAS‐28‐ESR (3.8 in Tai Chi group, 3.5 in control group); baseline HAQ (0.6 in Tai Chi group, 0.4 in control group)
Interventions	Study duration: 12 weeks Intervention group (N = 29): Tai Chi exercise (“Twelve Movement Tai Chi for arthritis”) implemented as a group exercise once a week for 60 min over the course of 3 months at the hospital gymnasium. Control group (N = 14): given information about lifestyle modification and exercise
Outcomes	RA disease activity (DAS‐28‐ESR, RAPID 3); functional disability (HAQ); cardiovascular disease risk factors; atherosclerotic measurements
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	The 56 participants with RA were assigned to two groups by their willingness to participate
Allocation concealment (selection bias)	Unclear risk	insufficient information – allocation method not mentioned
Blinding of participants and personnel (performance bias) All outcomes	High risk	blinding not possible
Blinding of outcome assessment (detection bias) All outcomes	High risk	The practitioners who performed clinical assessment were blinded to exercise group and study phase, but participants not blinded, and study contributed mostly patient‐reported outcomes
Incomplete outcome data (attrition bias) All outcomes	High risk	Among the initial allocated 46 participants, 13 participants in the control group dropped out due to losing interest in the study; dropout rate = 28%
Selective reporting (reporting bias)	Low risk	all outcomes were presented
Other bias	Unclear risk	insufficient information

Van Deusen 1987.

Methods	Randomized controlled trial
Participants	Ambulatory adults (N = 46) with rheumatoid arthritis in the United States who had medical recommendations for home rest and exercise, and no prior range of motion (ROM) dance experience. Demographic data reported at post‐test on 39 participants: age (average 56 years; range 29 to 80 years); 82% women; disease duration (average 10 years; range 0 to 38 years)
Interventions	Study duration: 8 weeks Experimental group (N = 23): taught a series of eight (once per week), 90‐minute, weekly health education classes which included a repetition of a 7‐minute Tai Chi ROM dance sequence accompanied by a poem and music, a guided relaxation experience, and a group discussion. The experimental participants were also encouraged to practice ROM dance sequences daily at home along with relaxation techniques. Control group (N = 23): not taught the classes, but did receive a brochure outlining the ROM dance program and the research project. The control group was not given any instructions regarding home rest and exercise.
Outcomes	outcomes assessed at 8 weeks and 24 weeks: upper and lower range of extremity range of motion, frequency, benefit and enjoyment of exercise of dance program (scale 3 to 15)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	insufficient information – method of randomization not described
Allocation concealment (selection bias)	Unclear risk	insufficient information – not described
Blinding of participants and personnel (performance bias) All outcomes	High risk	blinding not possible between ROM dance arm and control arm
Blinding of outcome assessment (detection bias) All outcomes	High risk	insufficient information – blinding of assessors not described but participants not blinded
Incomplete outcome data (attrition bias) All outcomes	High risk	large drop‐out rate (28%)
Selective reporting (reporting bias)	Low risk	although some outcomes were regrouped, all were reported
Other bias	High risk	measurement tools developed for the study (for exercise‐rest self report) may have lacked rigorous validity, and had not been tested with the population under consideration

Wang 2008.

Methods	Prospective, single‐blind randomized controlled trial
Participants	Ambulatory adults (N = 20) aged 18 and over, with functional class I or II rheumatoid arthritis, in an urban, tertiary care academic hospital in the United States. Average age (48 years in Tai Chi group, 51 in control group); women (80% in Tai Chi group, 70% in control group); disease duration (14 years in Tai Chi group, 15 years in control group); baseline HAQ (0.9 in Tai Chi group, 0.4 in control group); baseline pain VAS (3.2 in Tai Chi group, 1.4 in control group)
Interventions	Study duration: 12 weeks Intervention (N = 10): two 60‐minute Tai Chi sessions conducted each week for 12 weeks; Control (N = 10): two 60‐minute RA education and stretching classes each week for 12 weeks Tai Chi sessions consisted of Tai Chi principles, exercises, breathing techniques, and relaxation. Participants were encouraged to practice Tai Chi at home. Control group did not receive any Tai Chi instruction, but participated in stretching and wellness education.
Outcomes	Outcomes assessed at baseline, 12 weeks, 24‐week telephone follow‐up: ACR20; disease activity (tender joints, swollen joints, HAQ, pain scale, patient global, fatigue, erythrocyte sedimentation rate (ESR), C‐reactive protein (CRP); functional assessments (grip strength, stand time, walking time)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	random assignment using computer‐generated random numbers
Allocation concealment (selection bias)	Low risk	provided in sealed, opaque envelopes
Blinding of participants and personnel (performance bias) All outcomes	High risk	blinding not possible between Tai Chi and control arm
Blinding of outcome assessment (detection bias) All outcomes	High risk	outcomes measures assessed by phone – not sure if blinded; physical and functional assessments blinded, but participants not blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	outcomes data reported for all participants
Selective reporting (reporting bias)	Low risk	all outcomes described in methods reported in results tables
Other bias	Low risk	no additional bias concerns

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Bang 2014	no control group
Callahan 2016	multiple types of arthritis; RA not differentiated
Dogra 2015	cohort study; no control group
Hammond 2013	no control group
Hun 2013	not control group
Lee 2012	no control group
Uhlig 2005	pilot study; no control group
Zhu 1999	unable to find full text

Differences between protocol and review

Search strategy for the updated review did not include searching in Chinese databases or the Allied and Complementary Medicine Database (AMED). Methodological differences in some areas (i.e. risk of bias, 'Summary of findings' table) and outcomes selected were updated to use the most recent guidance from the Cochrane Musculoskeletal Group. Major clinical and functional outcomes include pain (which was included since ACR50 was not reported), measure of disease activity (DAS‐28‐ESR) and measure of function (HAQ); with one major safety outcome, overall withdrawals. Additional outcomes reported in the original review (tender and swollen joints, strength, range of motion), and an additional efficacy measure (ACR20) are included in the narrative as minor outcomes.

Contributions of authors

In the updated review

• AM was primarily responsible for updating this review and revising the analyses and draft of the review, which included selecting trials from the updated search, reviewing, extracting, and analyzing data for the update.

• JS was responsible for reviewing data and analyses, interpreting the findings, and revising the draft of the review.

• GW and PT contributed methodological expertise and commented on early drafts, as well as on the revised updated draft.

In the original review:

• AH was responsible for selecting trials, extracting and analyzing data, and writing the manuscript.

• VR assisted in selecting trials, extracting and analyzing data, and writing the manuscript.

• MJ commented on drafts, provided input into interpretation of analyses and results, and added to the discussion and conclusion of the review. JM developed the search strategy.

• WT conducted the searches in Chinese databases, translated articles written in Chinese, and helped to extract and analyze data from these Chinese articles.

• GW and PT contributed methodological expertise and commented on early drafts.

Sources of support

Internal sources

• Institute for Population Health, University of Ottawa, Canada.

  resources and facility support for P. Tugwell and G. Wells

• Birmingham VA Medical Center, USA.

  resources and facility support for J. Singh

• University of Alabama at Birmingham, USA.

  resource and facility support for A. Mudano and J. Singh

External sources

• No sources of support supplied

Declarations of interest

AM – none

JAS – JAS has received consultant fees from Crealta/Horizon, Medisys, Fidia, UBM LLC, Medscape, WebMD, Clinical Care options, Clearview healthcare partners, Putnam associates, Spherix, the National Institutes of Health and the American College of Rheumatology. JAS owns stock options in Amarin pharmaceuticals and Viking therapeutics. JAS is a member of the executive of OMERACT, an organization that develops outcome measures in rheumatology and receives arms‐length funding from 36 companies. JAS serves on the FDA Arthritis Advisory Committee. JAS is a member of the Veterans Affairs Rheumatology Field Advisory Committee. JAS is the editor and the Director of the UAB Cochrane Musculoskeletal Group Satellite Center on Network Meta‐analysis. JAS previously served as a member of the following committees: member, the American College of Rheumatology's (ACR) Annual Meeting Planning Committee (AMPC) and Quality of Care Committees, the Chair of the ACR Meet‐the‐Professor, Workshop and Study Group Subcommittee and the co‐chair of the ACR Criteria and Response Criteria subcommittee.

GW – none

PT – travel and accommodation for OMERACT meetings – a registered non‐profit independent medical research organization, whose goal is to improve and advance the health outcomes for patients suffering from musculoskeletal conditions. OMERACT receives unrestricted educational grants from the American College of Rheumatology, European League of Rheumatology, and several pharmaceutical companies listed below, which is used to support fellows, international patient groups, and support a major international bi‐annual conference, which results in many peer‐reviewed publications; Amgen, Astra Zeneca, Bristol Myers Squibb, Celgene, EliLilly, Genentech/Roche, Genzyme/Sanofi, Horizon Pharma Inc, Merck, Novartis, Pfizer, PPD, Quintiles, Regeneron, Savient, Takeda Pharmaceutical, UCB Group, Vertex, Forest, Bioiberica. PT is an independent committee member for clinical trial Data Safety Monitoring Boards for FDA approved trials, being conducted by UCB Biopharma GmbH & SPRL, Parexel International, and Prahealth Sciences. He is an independent medical consultation professional services for CHEOR Solutions (Canada) Ltd., Innovative Science Solutions LLC; and an advisory committee member of the Canadian Reformulary Group Inc., a company that reviews the evidence for health insurance companies employer drug plans.

No authors have an interest in Tai Chi or any comparators considered for this review.

New search for studies and content updated (no change to conclusions)