A systematic review and meta-analysis of the effects of Qigong and Tai Chi for depressive symptoms

Accepted Manuscript Title: A Systematic Review and Meta-analysis of the Effects of Qigong and Tai Chi for Depressive Symptoms Author: Xin Liu Justin Clark Dan Siskind Gail M. Williams Gerard Byrne Jiao L. Yang Suhail A. Doi PII: DOI: Reference:

S0965-2299(15)00082-5 http://dx.doi.org/doi:10.1016/j.ctim.2015.05.001 YCTIM 1455

To appear in:

Complementary Therapies in Medicine

Received date: Revised date: Accepted date:

30-9-2014 11-5-2015 24-5-2015

Please cite this article as: Liu X, Clark J, Siskind D, Williams GM, Byrne G, Yang JL, Doi SA, A Systematic Review and Meta-analysis of the Effects of Qigong and Tai Chi for Depressive Symptoms, Complementary Therapies in Medicine (2015), http://dx.doi.org/10.1016/j.ctim.2015.05.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

TITLE PAGE

Title:

ip t

A Systematic Review and Meta-analysis of the Effects of Qigong and Tai Chi for Depressive Symptoms

cr

Authors: Xin Liu1,2,3, Ph.D., Justin Clark4, Dan Siskind3,5,6, PhD, Gail M. Williams5, Ph.D., Gerard Byrne3,7, M.D., Ph.D., Jiao L. Yang8, MPsy, Suhail A. Doi9, FRCP, Ph.D. 1

The Third Clinical School, Zhejiang Chinese Medical University, Hangzhou, China Wuhan Sport University, Wuhan, China 3 School of Medicine, The University of Queensland, Brisbane, Australia 4 Drug Arm Australia Library, Brisbane, Australia 5 School of Population Health, The University of Queensland, Brisbane, Australia 6 Metro South Addiction and Mental Health Service, Brisbane, Australia 7 Mental Health Service, Royal Brisbane & Women's Hospital, Brisbane, Australia 8 Centre of Psychological Health, Department of Culture and Art Teaching, University of International Business and Economics, Beijing, China 9 Research School of Population Health, The Australian National University, Canberra, Australia

M

an

us

2

Ac ce pt e

d

Correspondence: Dr. Xin Liu Translational Research Institute, School of Medicine, The University of Queensland 199 Ipswich Road, Woolloongabba Brisbane, QLD 4012, Australia Tel: 61-7-3176 2051 Fax: 61-7-3176 5663 E-mail: [email protected] Word Count: 3,446 (main text) Number of Pages: 45 Number of Tables: 5 Number of Figures: 5

1

Page 1 of 47

Abstract

ip t

Background: Qigong and Tai Chi are the two most popular traditional Chinese exercises, known as mind-body movement therapies. Previous studies suggest that Qigong and Tai Chi may be beneficial in reducing depressive symptoms. This was the first study to systematically review and compare the effects of Qigong and Tai Chi on depressive symptoms.

us

cr

Methods: A systematic search of six electronic databases was undertaken through to February 2014, for randomized controlled trials (RCTs) which reported depressive symptoms measured by a depressive symptom rating scale. The standardized mean difference in depressive symptoms score between Qigong or Tai Chi and a control group (at the end of follow-up) was extracted as a primary outcome. The secondary outcome was the standardized mean gain in symptom score (SMG) relative to the baseline from individual arms of the RCTs for various forms of care including Qigong, Tai Chi, usual care, other exercise, education and miscellaneous interventions.

M

an

Results: Thirty studies with a total of 2328 participants (823 males and 1505 females) were included. A significant effect was found for the Qigong interventions (Cohen’s d -0.48 95% CI 0.48 to -0.12; SMG -0.52, 95% CI -0.79 to -0.26). There was no significant effect seen for Tai Chi for the primary endpoint. No mean change in symptom scores were seen for Tai Chi, usual care, other exercises, education and the ‘miscellaneous’ group in pre-post assessment in single arms. The Qigong results were found to be robust in sensitivity analyses.

.

Ac ce pt e

d

Conclusions: Qigong appears to be beneficial for reducing depressive symptom severity. However, given the low quality of the included studies and the documented evidence of publication bias, these results should be viewed cautiously.

2

Page 2 of 47

1

Introduction

Depression is a serious medical condition, with an estimated 350 million people of all ages affected worldwide.1 Depression is associated with a range of physical illnesses, including stroke,

ip t

diabetes, cancer, arthritis, osteoporosis, and obesity,3,4 and the onset of depression incrementally worsens the latter health problems.5-7 Interventions to tackle this problem are therefore important and previous studies have indicated an important role for physical activity in the management of

us

cr

depression.8

Specialized physical activity interventions such as Qigong and Tai Chi have been shown to be effective in reducing depressive symptoms in previous studies.9-13 Qigong and Tai Chi are the

an

two most popular traditional Chinese exercises, known as mind and body movement therapies. Qigong is a traditional Chinese medical exercise and Tai Chi is a type of Chinese martial art. Qigong came into being earlier than Tai Chi, and has been recognized as a ‘medical’ exercise in

M

the Traditional Chinese Medicine field for thousands of years.14 There are many styles of Qigong, such as Daoyin Qigong, WuQinXi (Five Animals Qigong) and BaDuanJin (Eight-

d

section Brocades Qigong). In contrast, Tai Chi was created by Mr. Chen Wang Ting (a retired

Ac ce pt e

General from army) as martial arts more than 300 years ago (in 1600’s), named Chen style Taijiquan.15 Apparently, Qigong is a health/medial exercise, consisting of movements with health functions and with mind activity playing a predominant role in training. Therefore, Qigong focuses more on ‘inside’ energy flow than Tai Chi. Qigong can be practiced as a 'static' (sitting, lying or standing) or ‘dynamic’ (moving) style, and is generally easier to master than Tai Chi. Tai Chi can be practiced as standing and mainly moving style (not sitting or lying) and was initially a martial art, consisting of movements with fighting functions and therefore it focuses on ‘outside’ defending and attacking intention while practicing. However, in line with cultural development, it has now evolved into many different gentle styles, such as ‘Yang style’, ‘Sun style’, ‘Wu style’, ‘24 forms’, ‘48 forms’ and ‘42 forms’, and its function/intention has also changed accordingly. Therefore, people, especially the aging group, nowadays practice it for health at a slow pace.

3

Page 3 of 47

Although some reviews on the effects of Qigong and Tai Chi have also been conducted,9-13 they were not conclusive because they were either not quantitative reviews or had several limitations. For example, there has just been only one previous meta-analysis on Qigong in relation to depressive symptoms that examined 10 randomized controlled trials (RCTs) with quantitative

ip t

data but pooled only subsets of 2, 3 or 4 RCTs (by type of control group)9 and the results were only suggestive of a beneficial effect. When it comes to Tai Chi, one meta-analysis12 combined both RCTs and observational studies while another meta-analysis combined just four RCTs and

cr

253 participants13 with both suggesting a significant effect of Tai Chi on depression scores. However, this was inconclusive given the small numbers. A subsequent meta-analysis that

us

combined Tai Chi and Qigong RCTs found no effect on depression scores, though this was primarily a study in cancer patients.10 The latter body of evidence has also been backed up by a

an

comprehensive review on the topic of meditative movements which suggests that the current data is inconclusive with regards to depressive symptoms.11 Finally, none of the reviews assessed and compared the difference between Qigong and Tai Chi in terms of the effect on depressive

M

symptoms.

d

This study was therefore undertaken to 1) perform a more rigorous review and analysis of the

Ac ce pt e

effect of Qigong and Tai Chi on depressive symptoms based only on RCT data and 2) to compare the effect of Qigong and Tai Chi with the changes seen in control groups that include usual care, other exercises and education.

2

Methods

2.1

Data sources and eligibility criteria

A systematic search of the literature was conducted in PubMed, Cochrane (CENTRAL), Embase, CINAHL, AMED, and PsycINFO for all citations that referred to either Qigong or Tai Chi and depressive symptoms until February, 2014. The specific keywords included were depression, depressions, depressive symptoms, depressive symptom, emotional depression, Qigong, Qi Gong, Ch'i Kung, breathing exercises, respiratory muscle training, Tai Chi, Tai Ji, Tai-ji, Tai Ji Quan, Taiji, Taijiquan, or Tai Chi Chuan. Controlled vocabulary terms (e.g. MeSH terms) were also used for each of databases.

4

Page 4 of 47

The inclusion of studies was restricted to randomized controlled trials reporting prospective use of Qigong (defined as any exercise method that included Qigong training, external Qigong therapy and breathing training) or Tai Chi (defined as any exercise method that included Tai Chi training) interventions. Additionally, there were no restrictions on mean age (mean age greater

ip t

than or equal to 60 years) or co-morbidities (defined as any additional medical condition other than depression). Finally, all studies were restricted to those that a) had baseline assessment on a depression symptom scale (irrespective of whether they had been diagnosed with depression or

cr

not); b) had clearly reported pre-intervention and post-intervention results in an extractable

2.2

Data extraction and quality assessment

us

format; and c) were written in English language.

an

Data was extracted for each study and is summarized in the data extraction table (see Tables 1 and 2). The data abstracted included the characteristics of each study, including pre-intervention

M

and post-intervention depression symptom scores allocated to Qigong, Tai Chi or any of the control groups; associated conditions; intervention style, duration and frequency; and subject demographic data, including age range, gender, and sample size during the study. The control

d

groups were categorized into usual care, other exercise, education, and a ‘miscellaneous’ group.

Ac ce pt e

The latter consisted of weekly telephone support, newspaper reading group, mahjong leisure activity, handicraft leisure control activity, non-exercise control, attention control intervention (including the wellness education and stretching program), basic rehabilitation activities (including self-care training, remedial activities, and educational programs), and control group without details.



Two outcomes were of interest: The standardized mean difference (d) in depression scores at the end of follow-up and the standardized mean gain (SMG) in depression scores from baseline to end of follow-up. The latter was necessary given that all control groups are different across RCTs, and therefore this secondary outcome (pre-post single arm analysis) is required to demonstrate that all control arms have similar null effects. While d represents a two arm 5

Page 5 of 47

comparison of scores at the end of follow-up, the SMG outcome was pooled across single arms of the RCT’s. The quality of each study was assessed using a specifically created methodological quality checklist for each outcome (separate assessment for two group comparisons and pre-post

ip t

comparisons as sources of systematic error are different by outcome) modified from several sources including the GRADE guidelines and Cochrane Collaboration’s tool for assessing risk of

cr

bias.16,17 For each study, the questions assessed several domains such as study design, selection,

us

information, confounding, and analytical biases (see Tables 3a and 3b). The maximum scores were 19 for the standardized mean difference outcome and 13 for the standardized mean gain

an

outcome. It should be emphasized that these are not validated quality assessment instruments because these items are usually based on the consensus among experts, implicitly based on both

M

empirical evidence and personal opinion,18 differ based on outcomes and study topic and therefore will vary across individual meta-analyses.19 Therefore, while items are well recognized,

d

scales necessarily will vary and validation, while possible, is not as important as it would be for

Ac ce pt e

say scales that assess fixed constructs such as depression or quality of life.19

Quality items were weighted equally given the paucity of meta-epidemiological studies informing us otherwise. The only exceptions were randomization and allocation concealment that were rated higher empirically. For all other safe-guards one point was assigned if it was present and zero otherwise. The quality score (Qi) was then calculated by summing the points awarded for each question and dividing this by the maximum within each meta-analysis, thereby rescaling between zero and one.



6

Page 6 of 47

2.3

Outcome measure and statistical analyses

The primary outcome was the standardized mean difference in symptom scores at the end of follow-up pooled separately for Qigong and Tai Chi studies. The secondary outcome was the

ip t

standardized mean gain in symptom score (SMG) between pre-intervention and post-intervention pooled separately for individual arms which included Qigong, Tai Chi and each of the other

XT 2 − XT1 , sp

us

SMG =

cr

control interventions including usual care. The standardized mean gain (SMG) was given by:

Where

an

s p = ( sT21 + sT2 2 ) / 2

and whose standard error was given by:

M

SEsmg =

2(1 − r ) SMG 2 + n 2n

Where Sp, ST1, and ST2 are the standard deviations of groups pooled, the pre-intervention and the

d

post-intervention results, respectively. The standard deviations were calculated from either the

Ac ce pt e

confidence intervals or the standard error and sample size when standard deviations were not given. In instances where the median and interquartile range were presented, the median was used as an approximation of the mean and the standard deviation was estimated by dividing the interquartile range by 1.35.20 Where median and range were available, the approach given by Hozo et al was used.21

Pooled estimates were obtained after bias adjustment. The quality scores were used to redistribute inverse variance weights based on study ranking of safe-guards used against bias using a quality-effects (QE) model.22,23 Therefore, weight redistribution occurred in favor of the studies with higher methodological quality ranking, which contributed to a higher weighting of these studies towards the overall effect size.23 It should be emphasized that these empirical studies confirm that the gain in estimator efficiency (in terms of mean squared error reduction) far outweigh any errors due to subjectivity of the quality assessment.24 Nevertheless, the random effects results were also computed and presented (in the online material) for comparison given 7

Page 7 of 47

that some statisticians still raise concerns over the use of summary quality scores. Heterogeneity was assessed by the Cochran’s Q test, the I2 index, and the Tau-squared statistic (τ2), as well as through graphical evaluation of the forest plot. Heterogeneity was defined as a Cochran Q test p-

ip t

value of less than 0.1, or a τ2 value of greater than zero.25 A sensitivity analysis was conducted to determine the degree to which the main findings vary depending on the selection criteria of studies included in the meta-analysis. These selection

cr

criteria included calendar period of study (the Qigong and tai chi studies were arbitrarily split by time frame of the study – cutoff at 2010-2011 for more recent studies), study sample size (50-60

us

subjects defined as small samples), severity of depressive symptoms at baseline (based on the recommended cutoff for each scale for depression given as a footnote to Table 4), co-morbidity

an

status (presence of a medical condition other than depression), and elderly status (mean age >=60 years). The sensitivity analysis determined if these selection criteria had a substantial impact on

M

the pooled standardized mean gain in symptom score.

Publication bias was examined by visual inspection of the funnel plot, but given that its visual

d

appearance can be problematic, a Doi plot was also used to graphically evaluate publication bias.

Ac ce pt e

The latter is a plot of the outcome of interest against the absolute value of the z-score derived from the studies and provides a better visual assessment of asymmetry than the funnel plot.26 The results of all analyses were considered statistically significant if the 95% confidence intervals did not include zero. The data was analyzed using MetaXL software version 2.0 for Windows (www.epigear.com).

3

Results

3.1

Search results

In total, 811 studies were located after excluding duplications (Figure 1). Abstracts were reviewed for selection of the studies whose full text were to be retrieved and finally eighteen Qigong27-44 and twelve Tai Chi studies45-56 were included for analysis after excluding irrelevant studies and studies with unclear data whose authors did not respond to a request to clarify the problem (Figure 1). Two Qigong studies28,36 and two Tai Chi studies52,53 were excluded from the standardized mean difference analysis because of non-extractable data for this outcome. 8

Page 8 of 47



3.2

Characteristics of included studies

ip t

All included studies were randomized controlled trials (see Tables 1 and 2). The Qigong intervention period across all studies ranged between 10 days to 24 weeks. These were delivered across all studies in between 1 to 21 sessions per week (3 sessions daily) of between 5 to 120

cr

minutes duration per session. The interventions included Guolin Qigong, Guolin New Qigong,

us

eight-section Brocades Qigong, Eight-section Brocades plus Yi Jin Jing Qigong, Chan Mi Qigong, External Qigong therapy, Wu Xing Ping Gong, breathing training and two studies where Qigong type was not specified. Similarly, the Tai Chi intervention period ranged from 8 to 20

an

weeks with 1 to at least 3 sessions per week of 30 to 120 minutes duration per session. Tai Chi

3.3

Participants

M

Interventions were Yang style and Sun style Tai Chi.

In the included studies, there were 1425 participants (535 males and 890 females) given Qigong

d

and 903 participants (288 males and 615 females) given Tai Chi with a total of 2328 participants

Ac ce pt e

(823 males and 1505 females). Participants in the Qigong studies had a wide variety of different co-morbid health conditions which included Parkinson’s disease, hypertension, coronary heart disease, type 2 diabetes, allogenic haematopoietic stem cell transplantation, burnout and exhaustion syndromes, chronic pain, asthma, depression and osteoarthritis. In some cases subjects were healthy middle-aged adults. Participants in the Tai Chi studies also had a wide variety of different co-morbid health conditions including dementia, cerebral vascular disorder, chronic obstructive pulmonary disease, and osteoarthritis.

3.4

Quality of the studies

When assessing the Cohen’s d outcome, the main quality deficiencies in more than half of the Qigong studies included were 1) the allocation of randomization results was not concealed; 2) the outcome assessors were not blinded to the nature of intervention; 3) the subjects were not blinded to the nature of intervention or control; 4) the groups were not similar at baseline in key confounding variables or steps were not taken to achieve comparability of key confounders; 5) 9

Page 9 of 47

intention-to-treat analyses were not conducted for the outcome of interest. Similarly, the main quality deficiencies in more than half of the Tai Chi studies included were 1) the allocation of randomization results was not concealed; 2) the subjects were not blinded to the nature of intervention or control; 3) the interventions were not clearly defined; 4) the groups were not

ip t

similar at baseline in key confounding variables or steps were not taken to achieve comparability of key confounders.

cr

When assessing the SMG outcome in individual study arms, the main overall quality deficiencies in more than half of the studies included 1) the outcome assessors were not blinded to the nature

us

of intervention; 2) researchers did not record and adequately control for any impact from change to concurrent intervention or exposure that could affect outcome; and 3) intention-to-treat

Quantitative synthesis

M

3.5

an

analyses were not conducted for the outcome of interest.

Primary outcome

Across RCTs, the effect magnitude ranged between -2.23 and 0.45 for Qigong and between -2.44

d

and 0.47 for Tai Chi (Figure 2). The only significant overall effect (Figure 2) was found for the

Ac ce pt e

Qigong interventions (d -0.48, 95% CI -0.83 to -0.12). There was no significant overall effect seen with Tai Chi (d -0.07, 95% CI -0.44 to 0.31).

Secondary outcome

Figure 3 shows that the only overall effect in the reduction of depressive symptoms was found for the Qigong interventions (SMG -0.52, 95% CI -0.79 to -0.26). There was no significant overall effect seen with Tai Chi, usual care, other exercises, education and the ‘miscellaneous’ category (Figure 3). Significant heterogeneity was seen with the Qigong, Tai Chi, usual care, other exercises and education interventions.

10

Page 10 of 47

3.6

Sensitivity analyses

Table 4 shows the results of the sensitivity analysis for the primary outcome. The Tai Chi studies showed a stronger effect when studies were smaller, more recent, shorter duration, had

ip t

more severe symptoms at baseline and did not have co-morbidities. There was no clear pattern discernable for the Qigong studies.

us

cr



There was gross evidence of publication bias favoring the positive studies of both Qigong and Tai Chi studies. The Doi and funnel plots

(Figures 4 and 5) demonstrated asymmetry

an

suggesting that either there was gross heterogeneity such that small studies had systematically larger effects than larger studies or that those with negative outcomes did not get published and

4

Ac ce pt e

d



M

were therefore missing.

Discussion

The novel finding from the present review was that Qigong demonstrated a significant effect on lessening the severity of depressive symptoms, but not Tai Chi. In single arm pre-post evaluations, usual care, other conventional exercises or education were no better than Tai Chi. This suggests that the traditional Qigong mind-body movement therapy may be more effective than the martial arts based Tai Chi exercise and conventional exercises in reducing depressive symptoms. The current review did not support the conclusions of two prior reviews published in 201012 and 201313 suggesting that Tai Chi was an effective intervention for depressive symptom scores as the review found that Tai Chi studies did not show significant improvement in depressive symptom scores in study participants reviewed. However, it should be noted that the majority of Qigong interventions were < 3 months and majority of Tai Chi interventions were > 3 months, which could explain the Qigong-Tai Chi difference if duration is an important variable and effectiveness drops off. However, the Qigong studies with short intervention duration 11

Page 11 of 47

showed a similar magnitude of effect as those with longer intervention duration, suggesting that the differences may not entirely be explained by duration of intervention. In addition, Qigong (as a medical exercise) focuses more on ‘inside’ energy flow than Tai Chi (as a martial arts) and this

ip t

may partially explain the differences in efficacy for reduction of depressive symptom severity. There was a wide variety of participants in terms of demographic characteristics and health conditions. In addition, study methodology such as sample size and style and dose of exercise

cr

interventions (including length, training sessions and time) varied across the studies. Significant heterogeneity was detected in Qigong, Tai Chi, usual care, other exercises and education. In

us

terms of sensitivity analysis, Tai Chi studies conducted at later time periods demonstrated larger magnitudes of effect sizes on depressive symptoms than those in less recent years, suggesting

an

that the research methodology might have been improved with time. The Tai Chi studies with small sample sizes showed higher magnitudes of effect sizes than those with larger sample sizes, suggesting the studies with larger samples may have been more rigorously designed and

M

executed. There were also larger effects for Tai Chi when co-morbidities were absent. Such a differential did not seem to be present with the Qigong studies. In addition, the Doi plots

d

demonstrated gross asymmetry, suggesting that the magnitude of the effect seen for both Tai Chi

Ac ce pt e

and Qigong may partly be a result of publication bias and may not reflect its real potential effect. The strengths of the present review were analyses with a systematic and transparent literature extraction, standardized inclusion and study evaluation criteria and quality assessment as an essential part of the meta-analysis. Indeed, this was the first review on the effects of Qigong and Tai Chi on depressive symptoms that performed a stringent quality assessment that was a major part of this quality adjusted meta-analysis, and stratification by quality score was therefore unnecessary given its limitations.57 The strength of this meta-analysis is therefore the use of tangible information from studies to account for heterogeneity. This addresses the numerous problems associated with using the random effects model in meta-analysis, with particular reference to the interpretation of results and practicality of its assumptions.58-60

Finally,

limitations of this review were that it did not include unpublished studies and could not assess the dose response effect as most of the included studies had no or limited follow-up. In addition,

12

Page 12 of 47

the effects of styles of intervention and environment and finally gender difference across studies could not be assessed. However, this heterogeneity increases the external validity of the results. In conclusion, Qigong appears to be effective in improving depressive symptoms but not Tai Chi.

ip t

Pre-post evaluation of single arm control groups which included usual care, other conventional exercises and education revealed no effect akin to Tai Chi arm. However, these results are not conclusive given the evidence of publication bias and overall poor quality of the reviewed

cr

studies, and therefore should be viewed cautiously. Further studies with large sample size, longer

an

effectiveness and maintenance of benefits with Qigong.

us

follow-up and cost-effectiveness analyses are required to confirm the clinical effectiveness, cost-

Acknowledgement

The Australian National Heart Foundation and the national depression initiative, beyondblue

d

referees for their very helpful comments.

M

grant G 088 4034 supported preparation of this manuscript. We also thank three anonymous

Ac ce pt e

Conflict of Interest Statement: All authors have no financial disclosure.

13

Page 13 of 47

References The World Health Organization: Depression: Fact sheet N369 (October 2012).http://www.who.int/mediacentre/factsheets/fs369/en/. Accessed October 20, 2012. 2. Charlson FJ, Stapelberg NJC, Baxter AJ,Whiteford HA. Should Global Burden of Disease Estimates Include Depression as a Risk Factor for Coronary Heart Disease? BMC Med. 2011; 9 (47); doi:10.1186/1741-7015-9-47. 3. Luppino FS, de Wit LM, Bouvy PF, et al. Overweight, obesity, and depression: a systematic review and meta-analysis of longitudinal studies. Arch Gen Psychiatry. 2010 Mar;67(3):220-9. 4. Clarke D, Currie K. Depression, anxiety and their relationship with chronic diseases: a review of the epidemiology, risk and treatment evidence. MJA 2009;190(7): S54-S60. 5. Vos T, Flaxman AD, Naghavi M, et al. The Global Burden of Non-Fatal Health Outcomes for 1,160 Sequelae of 291 Diseases and Injures 1990-2010: A Systematic Analysis. Lancet. 2012;380:2163-96. 6. Moussavi S, Chatterji S, Verdes E, et al. Depression, chronic diseases, and decrements in health: results from the World Health Surveys. Lancet. 2007;370:851-58. 7. Chisholm D, Diehr P, Knapp M, et al. Depression status, medical comorbidity and resource costs: evidence from an international study of major depression in primary care [LIDO]. Br J Psychiatry. 2003;183:121-31. 8. Gary MC, Dwan K, Greig C, et al. Exercise for depression. Editorial Group: Cochrane Depression, Anxiety and Neurosis Group. Assessed Jul 13, 2012;doi:10.1002/14651858.CD004366.pub6. 9. Wang CW, Chan CL, Ho RT, Tsang HW, Chan CH, Ng SM. The effect of qigong on depressive and anxiety symptoms: a systematic review and meta-analysis of randomized controlled trials. Evid Based Complement Alternat Med. 2013;2013:716094. doi: 10.1155/2013/716094. Epub 2013 May 20. 10. Zeng Y, Luo T, Xie H, Huang M, Cheng AS. Health benefits of qigong or tai chi for cancer patients: a systematic review and meta-analyses. Complement Ther Med. 2014 Feb;22(1):173-86. doi: 10.1016/j.ctim.2013.11.010. Epub 2013 Dec 18. 11. Payne P, Crane-Godreau MA. Meditative movement for depression and anxiety. Front Psychiatry. 2013 Jul 24;4:71. doi: 10.3389/fpsyt.2013.00071. eCollection 2013. 12. Wang C, Bannuru R, Ramel J, Kupelnick B, Scott T, Schmid CH. Tai Chi on psychological well-being: systematic review and meta-analysis. BMC Complement Altern Med. 2010 May 21;10:23. doi: 10.1186/1472-6882-10-23. 13. Chi I, Jordan-Marsh M, Guo M, Xie B, Bai Z. Taichi and reduction of depressive symptoms for older adults: a meta-analysis of randomized trials. GeriatrGerontol Int. 2013 Jan;13(1):312. 14. Xue LG. Chinese Medical Qigong (pp.1-8). Haerbin: Heilongjiang Science and Technology Publisher;1990. 15. People’s Sport Publisher Group. Taijiquan (pp.3-9). Beijing: People’s Sport Publisher;1988. 16. Gordon H. Guyatt, Andrew D. Oxman, Gunn Vist, Regina Kunz, Jan Brozek, Pablo AlonsoCoello, Victor Montori, Elie A. Akl, Ben Djulbegovic, YngveFalck-Ytter,Susan L. Norris, JohnW.Williams Jr., David Atkins, Joerg Meerpohl, Holger J. SchÜnemann. GRADE guidelines: 4. Rating the quality of evidenced study limitations (risk of bias). J Clin Epidemiol. 2011 Apr;64(4):407-15. doi: 10.1016/j.jclinepi.2010.07.017. Epub 2011 Jan 19.

Ac ce pt e

d

M

an

us

cr

ip t

1.

14

Page 14 of 47

Ac ce pt e

d

M

an

us

cr

ip t

17. Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD, Savovic J, Schulz KF, Weeks L, Sterne JA; Cochrane Bias Methods Group; Cochrane Statistical Methods Group. The Cochrane Collaboration's tool for assessing risk of bias in randomised trials.BMJ. 2011 Oct 18;343:d5928. doi: 10.1136/bmj.d5928. 18. Verhagen AP1, de Vet HC, de Bie RA, Kessels AG, Boers M, Bouter LM, Knipschild PG. The Delphi list: a criteria list for quality assessment of randomized clinical trials for conducting systematic reviews developed by Delphi consensus. J Clin Epidemiol. 1998 Dec;51(12):1235-41. 19. Olivo SA1, Macedo LG, Gadotti IC, Fuentes J, Stanton T, Magee DJ. Scales to assess the quality of randomized controlled trials: a systematic review. Phys Ther. 2008 Feb;88(2):15675. 20. Higgins JP, Deeks JJ.Cochrane Handbook for Systematic Reviews of Interventions V5.0.1: Chapter 7: Selecting studies and collecting data. The Cochrane Collaboration; 2008. 21. Hozo SP, Djulbegovic B, Hozo I. Estimating the mean and variance from the median, range, and the size of a sample. BMC Med Res Methodol. 2005 Apr 20;5:13. 22. Doi SA, Thalib, L. A quality-effects model for meta-analysis. Epidemiology.2008;19(1):94100. 23. Doi SA, Barendregt JJ, Mozurkewich EL: Meta-analysis of heterogeneous clinical trials: an empirical example. Contemp Clin Trials.2011;32(2):288-298. 24. Doi SA, Barendregt JJ, Khan S, Thalib L, Williams GM. Simulation comparison of the quality effects and random effects methods of meta-analysis. Epidemiology 2015; In press 25. Takkouche B, Cadarso-Suarez C, Spiegelman D. Evaluation of old and new tests of heterogeneity in epidemiologic meta-analysis. Am J Epidemiol.1999;150(2):206-15. 26. Onitilo AA, Doi SA, Barendregt JJ. Methods of Clinical Epidemiology:Meta-Analysis II. (pp.253-266). Berlin: Springer-Verlag;2013. 27. Burini D, Farabollini B, Iacucci S, Rimatori C, Riccardi G, Capecci M, Provinciali L, Ceravolo MG. A randomized controlled cross-overtrial of aerobic training versus Qigong in advanced Parkinson's disease. EuraMedicophys. 2006 Sep;42(3):231-8. 28. Chan ES, Koh D, Teo YC, Hj Tamin R, Lim A, Fredericks S. Biochemical and psychometric evaluation of Self-Healing Qigong as a stress reduction tool among first year nursing and midwifery students. Complement Ther Clin Pract. 2013 Nov;19(4):179-83. doi: 10.1016/j.ctcp.2013.08.001. Epub 2013 Aug 30. 29. Chan JS, Ho RT, Wang CW, Yuen LP, Sham JS, Chan CL. Effects of qigong exercise on fatigue, anxiety, and depressive symptoms of patients with chronic fatigue syndrome-like illness: a randomized controlled trial. Evid Based Complement Alternat Med. 2013;2013:485341. doi: 10.1155/2013/485341. Epub 2013 Jul 31. 30. Chen KW, Perlman A, Liao JG, Lam A, Staller J, Sigal LH. Effects of external qigong therapy on osteoarthritis of the knee. A randomized controlled trial. Clin Rheumatol. 2008 Dec;27(12):1497-505. doi: 10.1007/s10067-008-0955-4. Epub 2008 Jul 25. 31. Chen Z, Meng Z, Milbury K, Bei W, Zhang Y, Thornton B, Liao Z, Wei Q, Chen J, Guo X, Liu L, McQuade J, Kirschbaum C, Cohen L. Qigong improves quality of life in women undergoing radiotherapy for breast cancer: results of a randomized controlled trial. Cancer. 2013 May 1;119(9):1690-8. doi: 10.1002/cncr.27904. Epub 2013 Jan 25. 32. Cheung BM, Lo JL, Fong DY, Chan MY, Wong SH, Wong VC, Lam KS, Lau CP, Karlberg JP. Randomised controlled trial of qigong in the treatment of mild essential hypertension. J Hum Hypertens. 2005 Sep;19(9):697-704. 15

Page 15 of 47

Ac ce pt e

d

M

an

us

cr

ip t

33. Chow YWY, Dorcas AS, Andrew MH. The effects of qigong on reducing stress and anxiety and enhancing body–mind well-being. Mindfulness. 2012;3:51–9. DOI 10.1007/s12671-0110080-3. 34. Chung LJ, Tsai PS, Liu BY, Chou KR, Lin WH, Shyu YK, Wang MY. Home-based deep breathing for depression in patients with coronary heart disease: a randomised controlled trial. Int J Nurs Stud. 2010 Nov;47(11):1346-53. doi: 10.1016/j.ijnurstu.2010.03.007. Epub 2010 May 11. 35. Wang F, Wang W, Zhang R, Lin Y, Hong L, Zhao Y, Ni Q, Zhang L, Isiiyasutomo, Tutiutitakuya, Kosikawafusako, Kisitaiti, Havukiyutaka, Suzukiakio. Clinical observation on physiological and psychological effects of Eight-Section Brocade on type2 diabetic patients. J Tradit Chin Med. 2008 Jun;28(2):101-5. 36. Kim SD, Kim HS. Effects of a relaxation breathing exercise on anxiety, depression, and leukocyte in hemopoietic stem cell transplantation patients. Cancer Nurs. 2005 JanFeb;28(1):79-83. 37. Smelson D, Chen KW, Ziedonis D, Andes K, Lennox A, Callahan L, Rodrigues S, Eisenberg D. A pilot study of Qigong for reducing cocaine craving early in recovery. J Altern Complement Med. 2013 Feb;19(2):97-101. doi: 10.1089/acm.2012.0052. Epub 2012 Jul 3. 38. Stenlund T, Birgander LS, Lindahl B, Nilsson L, Ahlgren C. Effects of Qigong in patients with burnout: a randomized controlled trial. J Rehabil Med. 2009 Sep;41(9):761-7. doi: 10.2340/16501977-0417. 39. Thomas M, McKinley RK, Mellor S, Watkin G, Holloway E, Scullion J, Shaw DE, Wardlaw A, Price D, Pavord I. Breathing exercises for asthma: a randomised controlled trial. Thorax.2009;64:55–61. doi:10.1136/thx.2008.100867. 40. Tsang HW, Mok CK, Au Yeung YT, Chan SY. The effect of Qigong on general and psychosocial health of elderly with chronic physical illnesses: a randomized clinical trial. Int J Geriatr Psychiatry. 2003 May;18(5):441-9. 41. Tsang HWH, Fung KMT, Chan ASM, Lee G and Chan F. Effect of a qigong exercise program on elderly with depression. Int J Geriatr Psychiatry. 2006; 21: 890–7. 42. Tsang HW, Tsang WW, Jones AY, Fung KM, Chan AH, Chan EP, Au DW. Psycho-physical and neurophysiological effects of qigong on depressed elders with chronic illness.Aging Ment Health. 2013;17(3):336-48. doi: 10.1080/13607863.2012.732035. Epub 2012 Oct 16. 43. Tsang HW, Lee JL, Au DW, Wong KK, Lai KW. Developing and testing the effectiveness of a novel health qigong for frail elders in Hong Kong: a preliminary study. Evid Based Complement Alternat Med. 2013; 2013:827392. doi: 10.1155/2013/827392. Epub 2013 Sep 10. 44. Yang KH, Kim YH, Lee MS. Efficacy of Qi-therapy (external Qigong) for elderly people with chronic pain. Int J Neurosci. 2005 Jul;115(7):949-63. 45. Cheng ST, Chow PK, Yu EC, Chan AC. Leisure activities alleviate depressive symptoms in nursing home residents with very mild or mild dementia. Am J Geriatr Psychiatry. 2012 Oct;20(10):904-8. doi: 10.1097/JGP.0b013e3182423988. 46. Cheon SM, Chae BK, Sung HR, Lee GC, Kim JW. The Efficacy of Exercise Programs for Parkinson's Disease: Tai Chi versus Combined Exercise. J Clin Neurol. 2013 Oct;9(4):237-43. doi: 10.3988/jcn.2013.9.4.237. Epub 2013 Oct 31. 47. Chou KL, Lee PW, Yu EC, Macfarlane D, Cheng YH, Chan SS, Chi I. Effect of Tai Chi on depressive symptoms amongst Chinese older patients with depressive disorders: a randomized clinical trial. Int J Geriatr Psychiatry. 2004 Nov;19(11):1105-7. 16

Page 16 of 47

Ac ce pt e

d

M

an

us

cr

ip t

48. Frye B, Scheinthal S, Kemarskaya T and Pruchno R. Tai Chi and Low Impact Exercise: Effects on the Physical Functioning and Psychological Well-Being of Older People. Journal of Applied Gerontology.2007;26:433. DOI: 10.1177/0733464807306915. 49. Lam LC, Chau RC, Wong BM, Fung AW, Lui VW, Tam CC, Leung GT, Kwok TC, Chiu HF, Ng S, Chan WM. Interim follow-up of a randomized controlled trial comparing Chinese style mind body (TaiChi) and stretching exercises on cognitive function in subjects at risk of progressive cognitive decline. Int J Geriatr Psychiatry. 2011 Jul;26(7):733-40. doi: 10.1002/gps.2602. Epub 2010 Dec 9. 50. Lavretsky H, Alstein LL, Olmstead RE, Ercoli LM, Riparetti-Brown M, Cyr NS, Irwin MR. Complementary use of tai chi chih augments escitalopram treatment of geriatric depression: a randomized controlled trial. Am J Geriatr Psychiatry. 2011 Oct;19(10):839-50. doi: 10.1097/JGP.0b013e31820ee9ef. 51. Taylor-Piliae RE, Hoke TM, Hepworth JT, Latt LD, Najafi B, Coull BM. Effect of Tai Chi on Physical Function, Fall Rates and Quality of Life Among Older Stroke Survivors. Arch Phys Med Rehabil. 2014 Jan 17. pii: S0003-9993(14)00010-0. doi: 10.1016/j.apmr.2014.01.001. 52. Wang C, Schmid CH, Hibberd PL, Kalish R, Roubenoff R, Rones R, McAlindon T. Tai Chi is effective in treating knee osteoarthritis: a randomized controlled trial. Arthritis Rheum. 2009 Nov 15;61(11):1545-53. doi: 10.1002/art.24832. 53. Wang W, Sawada M, Noriyama Y, Arita K, Ota T, Sadamatsu M, Kiyotou R, Hirai M, Kishimoto T. Tai Chi exercise versus rehabilitation for the elderly with cerebral vascular disorder: a single-blinded randomized controlled trial. Psychogeriatrics. 2010 Sep;10(3):160-6. doi: 10.1111/j.1479-8301.2010.00334.x. 54. Yeh GY, Roberts DH, Wayne PM, Davis RB, Quilty MT, Phillips RS. Tai chi exercise for patients with chronic obstructive pulmonary disease: a pilot study. Respir Care. 2010 Nov;55(11):1475-82. 55. Yeh GY, Wood MJ, Wayne PM, Quilty MT, Stevenson LW, Davis RB, Phillips RS, Forman DE. Tai chi in patients with heart failure with preserved ejection fraction. Congest Heart Fail. 2013 Mar-Apr;19(2):77-84. doi: 10.1111/chf.12005. Epub 2012 Oct 12. 56. Yeung A, Lepoutre V, Wayne P, Yeh G, Slipp LE, Fava M, Denninger JW, Benson H, Fricchione GL. Tai chi treatment for depression in Chinese Americans: a pilot study. Am J Phys Med Rehabil. 2012 Oct;91(10):863-70. 57. Al Khalaf MM, Thalib L, Doi SA. Combining heterogenous studies using the random-effects model is a mistake and leads to inconclusive meta-analyses. J Clin Epidemiol. 2011 Feb;64(2):119-23. doi: 10.1016/j.jclinepi.2010.01.009. Epub 2010 Apr 20. 58. Shuster JJ. Empirical vs natural weighting in random effects meta-analysis. Stat Med. 2010 May 30;29(12):1259-65. doi: 10.1002/sim.3607. 59. Brockwell SE, Gordon IR. A comparison of statistical methods for meta-analysis. Stat Med. 2001;20: 825-40. 60. Noma H. Confidence intervals for a random-effects meta-analysis based on Bartlett-type corrections. Stat Med. 2011 Dec 10;30(28):3304-12.

17

Page 17 of 47

cr us an

Table legends

Figure legends

ed

M

Table 1 Characteristics of included Qigong studies Table 2 Characteristics of included Tai Chi studies Table 3a Quality checklist (Primary outcome – analytic two-arm study) Table 3b Quality checklist (Secondary outcome – single arm) Table 4 Sensitivity analysis for the primary outcome results by time frame, study sample size, intervention duration, severity of depressive symptoms, co-morbidities status and mean age

Ac

ce

pt

Figure 1 Flow of literature extraction Figure 2 Effect sizes for standardized mean difference in depressive symptoms at the end of follow-up. a Qigong. b Tai Chi studies Figure 3 Standardized mean reduction in depressive symptoms from pre- to post-follow-up in single arms of a Qigong. b Tai Chi. c Usual care. d Other excises. e Education. f miscellaneous groups Figure 4 Qigong studies: Doi (left) and funnel (right) plots of between-group outcomes demonstrating gross asymmetry of both favoring stronger effects Figure 5 Tai Chi studies: Doi (left) and funnel (right) plots of between-group outcomes demonstrating gross asymmetry of both favoring stronger effects

Supplementary appendix Supplementary appendix 1 Detailed search strategy for PubMed Supplementary appendix 2 Random effects results for the primary outcome. Effect sizes for standardized mean difference in depressive symptoms at the end of follow-up using the random effects model: a Qigong studies; b Tai Chi studies

18 Page 18 of 47

cr us an

Table 1 Characteristics of included Qigong studies Groups

Population (study sample)

Duration

M

Reference

7 weeks

BDI

QG: Self-healing Qigong; 60 minutes per session; two sessions per week CG: Usual care control

Fist year nursing and midwifery students; Age (range): 18-28 years; Gender (m/f): 7/27

10 weeks

DASS 21

QG: Wu Xing Ping Xing Gong; Group training: 120 minutes per group session, two group sessions per week and at least 30 minutes per day for 5 weeks, followed by 12 weeks of home-based training CG: Waiting-list control

With chronic fatigue syndrome-like illness; Age (mean+SD): 42.5 ± 6.6 years; Gender (m/f): 32/105

17 weeks

HADS

ed

Parkinson’s disease; Age (mean+SD): 65.2±6.5 yrs; Gender (m/f): 9/17

QG: 3 sessions / week; 50 minutes / session CG: Aerobic training, 3 sessions / week; 50 minutes / session

ce

Ac

Chan, et al, 2013A28

Chan, et al, 201329

Baseline N (QG/ CG) 13/13

pt

Burini D et al, 200627

Depressi on Scale

Follow-up Mean (SD) Median (IRQ): QG = 15 (0-29); CG = 10 (0-29)

N Mean (QG/ (SD) CG) 11/11 Median (IRQ):QG = 11 (635); CG =11 (127)

18/16

Median (IQR): QG = 10 (2-26) CG = 8 (2-22)

18/16 Median (IQR): QG = 4 (0-14) CG = 2 (0-20)

72/65

QG = 9.1 (2.0) CG = 9.4 (2.2)

72/65 QG = 7.7 (3.2) CG = 9.8 (4.1)

19 Page 19 of 47

us

cr QG: External Qigong therapy by healer 1; 4 to 7 minutes per session; five to six sessions in 3 weeks CG1: External Qigong therapy by healer 2; 5 to 10 minutes per session; five to six sessions in 3 weeks CG2: Shame Qigong therapy; five to six sessions in 3 weeks

Duration

an

Population (study sample)

Depressi on Scale

With knee osteoarthritis; Age (mean+SD): 61.9 ± 8.63 years; Gender (m/f): 30/76

3 weeks

CES-D10

Women undergoing radiotherapy for breast cancer; Age (mean+SD): 45.0 ± 8.0 years; Gender (f): 96

5-6 weeks

CES-D20

Baseline

Follow-up

N (QG/ CG) QG/C G1/C G2: 45/12/ 49

Mean (SD)

49/47

QG = 13.1 (8.9) CG = 12.2 (9.2)

QG = 35.6 (4.3) CQ1 = 39.4 (6.6) CQ2 = 32.8 (9.0)

N (QG/ CG) QG/ CG1/ CG2: 45/12 /49

Mean (SD) QG = 34.2 (3.1) CQ1 = 36.7 (5.6) CQ2 = 34.4 (4.7)

ce

pt

ed

Chen, et al, 200830

Groups

M

Reference

QG: Guo Lin New Qigong; Group intervention: 40 minutes per group session, 5 group sessions per week; DVD with a recording of the qigong program also provided and encouraged to do home-based training when participants did attend group session. CG: Waiting-list standard care control

Ac

Chen, et al, 201331

49/47 QG = 12.3 (8.2 ) CG = 11.6 (8.1 )

20 Page 20 of 47

us

cr QG: Guolin Qigong, 2h / session / week during first 4 weeks, and 2h session / month during 12 weeks follow up CG: Conventional exercise (stretching, walking and stepping), 2h / session / week during first 4 weeks, and 2h session / month during 12 weeks follow up.

Duration

an

Population (study sample)

Depressi on Scale

Baseline

Hypertension; Age (mean+SD): 54.2±8.5 yrs; Gender (m/f): 37/51

16 weeks

BDI

N (QG/ CG) 47/44

Healthy middle-aged adults; Age (mean+SD): 44.2 ± 11.03 years; Gender (m/f): 23/45

12 weeks

DASS 21

34/34

Follow-up Mean (SD) NR

N Mean (QG/ (SD) CG) 47/41 Mean change (se): QG = - 1.9 (±0.8) CG = (se) = - 2.2 (±1.0)

QG: Chan mi Qigong; 90 minutes per session/week in the first 8 weeks followed by 4 weeks home-based practice CG: Waislist control (usual care)

Ac

Chow, et al, 201233

ce

pt

ed

Cheung, et al, 200532

Groups

M

Reference

QG = 6.29 (5.33) CG = 6.97 (5.93)

34/31 QG = 2.47 (3.38) CG = 6.65 (6.40)

21 Page 21 of 47

Population (study sample)

us

cr Chung, et al, 201034

Groups

Duration

an

Reference

Depressi on Scale

Baseline

Follow-up

N (QG/ CG) 33/39

Mean (SD)

With coronary heart disease; Age (mean+SD): 71.59±11.1 yrs; Gender (m/f): 43/19

4 weeks

BDI

QG: Eight-Section Brocade Qigong: Homebased; 1 hour/time/day; 7 days/week. CG: details, unknown.

With type 2 diabetics; Age (mean+SD): 58.8±8.11yrs; Gender (m/f): 18/27

2 months

Symptom Checklist 90

25/20

NR NR

25/20 QG = 0.4 (0.47) CG: 0.7 (0.53)

QG: Relaxation Breathing Exercise, 30 minutes daily/session, 7 days / week CG: Usual care control

With allogenic hemopoietic stem cell transplantation ; Age (mean+SD); 33.6±7.4yrs; Gender (m/f): 17/18

6 weeks

BDI

18/17

QG = 19.2 (1.9) CG = 18.4 (3.5)

18/17 QG = 8.4 (4.4) CG: 21.6 (7.2)

ce

pt

ed

M

QG: Home based breathing training, 10 minutes / time, 3 times daily CG: Weekly telephone support control group, 5 minutes per session / week

QG = 16.61 (7.35) CG = 19.44 (8.14)

N Mean (QG/ (SD) CG) 28/34 QG = 9.21 ( 6.04) CG = 21.03 (10.81)

Ac

Fang, et al, 200835

Kim, et al, 200536

22 Page 22 of 47

cr Population (study sample)

QG: External Qigong therapy by healer; 15 minutes per session; two to three sessions per week CG: Sham Qigong therapy by healer; 15 minutes per session; two to three sessions per week

Stenlund, et al, 200938

QG: One hour/session, biweekly. CG: Usual care control

Duration

Depressi on Scale

Baseline

Recently abstinent cocainedependent individuals; Age (mean+SD): 38.2 ± 10.8 years; Gender (m/f): 97/4

2 weeks

BDI

N (QG/ CG) 45/41

With diagnosis of burnout and exhaustion syndrome; Age (mean+SD); 40.4±9.1yrs; Gender (m/f): 14/68

12 weeks

HADS

41/41

Follow-up Mean (SD) QG = 13.6 (7.0) CG = 11.9 (8.3)

N Mean (QG/ (SD) CG) 45/41 QG = 3.7 (4.9) CG = 5.1 (5.6)

Median (IQR): QG = 11 (8–12) CG = 10 (7–12)

33/35 Median (IQR): QG = 7 (5–12) CG = 8 (5–11)

Ac

ce

pt

ed

M

Smelson, et al, 201337

us

Groups

an

Reference

23 Page 23 of 47

us

cr QG: Breathing training; consisted of three sessions, an initial 60 min small group session followed by two individual sessions of 30–45 min; and encouraged to practice these exercises for at least 10 min each day. Explanation of normal breathing and possible effects of abnormal ‘‘dysfunctional breathing’’ such as over-breathing, mouth breathing and upper chest breathing was provided. CG: Asthma education control; consisted of three sessions, an initial 60 min small group session followed by two individual sessions of 30–45 min.

Duration

an

Population (study sample)

With asthma; Median (interquartile range): 46.0 (33.0–57.3 (intervention group); 46.0 (35.0–57.0) (control group); Gender (m/f): 71/112

24 weeks

Depressi on Scale

HADS

Baseline N (QG/ CG) 94/89

Follow-up Mean (SD) NR

N Mean (QG/ (SD) CG) 63/66 Mean change (95% confidenc e interval): QG = 0.31 (0.72 to 0.10) CG = 0.44 (-0.08 to 0.94)

Ac

ce

pt

ed

Thomas, et al, 201239

Groups

M

Reference

24 Page 24 of 47

cr Duration

Depressi on Scale

Baseline

Elderly with chronic physical illnesses; Age (mean+SD): 74.6±8.97 yrs; Gender (m/f): 26/24

12 weeks

GDS

N (QG/ CG) 26/24

With depression or obvious features of depression; aged 65 or above; Gender (m/f): 16/56

16 weeks

GDS

NR

M

QG: Eight-Section Brocades Qigong, at least 30 minutes daily home training plus 2 group sessions / week CG: Basic rehabilitation activities including selfcare training, remedial activities, and educational programs, etc

us

Population (study sample)

Follow-up Mean (SD)

QG = 5.17 (2.75) CG = 6.50 (1.42)

QG = 7.39 (3.91) CG = 6.05 (3.40)

N Mean (QG/ (SD) CG) 26/24 QG = 6.13 (4.14) CG = 5.16 (4.14)

QG: Eight-Section Brocades Qigong; 30 to 45 min per session, 3 sessions per week plus 15 minutes daily home-based practice CG: Newspaper reading group; 30 to 45 min per session, 3 sessions per week

Ac

Tsang, et al, 200641

ce

pt

ed

Tsang, et al, 200340

Groups

an

Reference

34/48 QG = 3.19 (2.12) CG = 6.15 (1.46)

25 Page 25 of 47

cr Duration

Depressi on Scale

Baseline

Diagnostic Depression; Age (mean+SD): 80.2 ± 5.1 years; Gender (m/f): 12/26

12 weeks

HRSD

N (QG/ CG) 21/17

Frail elders; Age (mean+SD): 84.1 ± 6.0 years; Gender (m/f): 29/87

12 weeks

GDS

61/55

M

QG: Eight-section brocades; Group training: 45 minutes per group session, 3 group sessions per week CG: Newspaper reading and discussion comparison group; Group intervention: 45 minutes per group session, 3 group sessions per week

us

Population (study sample)

Follow-up Mean (SD) QG = 7.8 (6.6) CG = 8.9 (6.3)

N Mean (QG/ (SD) CG) 21/17 QG = 3.8 (3.6) CG = 6.8 (3.5 )

QG = 4.1 (3.1) CG = 4.4 (3.3)

61/55 QG = 4.3 (3.3) CG = 4.8 (3.5)

QG: Yan Chai Yi Jin TenSection Brocades (Eightsection brocades + Yi Jin Jing); 60 minutes per session; two sessions per week CG: Newspaper reading program; 60 minutes per session; two sessions per week

Ac

Tsang, et al, 201343

ce

pt

ed

Tsang, et al, 201242

Groups

an

Reference

26 Page 26 of 47

us

cr QG: Qi-therapy (external Qigong); 20 minutes/session, biweekly CG: Usual care control

Duration

an

Population (study sample)

With chronic pain; Age (mean+SD): 72.59±6.45yrs ; Gender (m/f): 8/32

ed

Yang, et al, 200544

Groups

M

Reference

4 weeks

Depressi on Scale

Korean version of the Profile of Mood States

Baseline N (QG/ CG) 20/23

Follow-up Mean (SD) QG = 11.26 ± 6.07 CG = 10.10 ± 5.62

N Mean (QG/ (SD) CG) 19/21 QG = 1.16 ± 1.74 CG = 7.86 ± 5.08

Ac

ce

pt

QG = Qigong group; CG = control group; NR = no report; IQR = interquartile range BDI = Beck Depression Inventory; GDS = Geriatric Depression Score; HADS = Hospital Anxiety and Depression Scale; DASS 21 = Depression Anxiety and Stress Scale 21; CES-D10 = Centre for Epidemiological Study Depression Scale 10; HRSD = Hamilton Rating Scale of Depression

27 Page 27 of 47

us

cr Groups

Population (study sample)

Duration

Depression Scale

M

Reference

an

Table 2 Characteristics of included Tai Chi studies

Cheng, et al, 201245

Mean (SD)

Mean (SD) TG = 7.75 (2.83) CG1 = 5.17 (4.57) CG2 = 9.17 (2.76)

GDS

TG: Sun style Tai Chi-for-arthritis program; 50-65 minutes per session, three sessions per week CG1: Combined exercise program including folk dancing, stepping and elastic-bank exercises CG2: Usual care control group (no intervention)

With mild-tomoderate Parkinson’s Disease; Age (mean+SD): 64.3 ± 7.2 years; Gender (f): 23

8 weeks

BDI

TG/CG 1/CG2: 9/7/7

TG = 23.4 (10.2) CG1 = 27.6 (12.3) CG2 = 23.2 (18.1)

TG/C G1/C G2: 9/7/7

TG = 20.3 (6.3) CG1 = 26.1 (11.2) CG2 = 29.3 (12.7)

TG: Yang style Tai Chi; 45 minutes/session, 3 sessions per week CG: Waiting list usual care control

With major depression or dysthymia; Age (mean+SD): 72.6±4.2yrs; Gender (m/f): 7/7

12 weeks

CES-D 20

7/7

TG = 32.0 (9.9) CG = 32.7 (8.7)

7/7

TG = 15.3 (9.8) CG = 39.1 (9.7)

28 Page 28 of 47

ed

pt

TG = 9.25 (2.14) CG1 = 8.42 (2.50) CG2 = 9.08 (2.11)

N (TG/ CG) TG/C G1/C G2:1 2/12/ 12

12 weeks

ce

Chou, et al, 200447

N (TG/C G) TG/CG 1/CG2: 12/12/1 2

Post-treatment

With at least moderate depressive symptoms; Age (mean+SD): 81.8±6.3 yrs; Gender (m/f): 12/24

TG: A seated 12-form Yang style Tai Chi; 60 minutes/session; 3 sessions / week CG1: Mahjong leisure activity; 60 minutes/session; 3 sessions / week CG2: Handicraft leisure control activity; 60 minutes/session; 3 sessions / week

Ac

Cheon, et al, 201346

Pre-treatment

cr us

ed

M

TG: Yang style Tai Chi; 60 minutes /session, 3 sessions / week CG1: Low impact exercise intervention (traditional Western exercise); 60 minutes /session, 3 sessions / week CG2: Non-exercise control TG: 24 forms simplified Tai Chi; at least 30 minutes daily; at least 3 sessions / week CG: Stretching and toning exercise; at least 30 minutes daily; at least 3 sessions / week

Ac

ce

Lam, et al, 201049

Population (study sample)

pt

Frye, et al, 2007 48

Groups

an

Reference

Lavretsky,, et al, 201150

TG: Tai Chi Chih; 2 hours / week CG: Health education control; 2 hours / week

Duration

Depression Scale

Pre-treatment N (TG/C G) TG/CG 1/CG2: 31/30/2 3

Mean (SD)

N (TG/ CG) TG/C G1/C G2: 23/28 /21

Mean (SD) TG = 9.6 (8.87) CG = 8.7 (8.80) CG = 10.6 (8.82)

Sedentary older people; Age (mean+SD): 69.2±9.26 yrs; Gender (m/f): 30/54

12 weeks

CES-D 20

With either a Clinical Dementia Rating (CDR 0.5) or amnestic-MCI; Age (mean+SD): 77.8±6.5 yrs; Gender (m/f): 92/297 With major depression; Age (mean+SD): 70.6±7.2 yrs; Gender (m/f): 28/45

20 weeks

CDS

171/21 8

TG = 0.9 (1.8) CG = 0.8 (1.8)

135/1 94

TG = 0.7 (0.9) CG = 0.6 (0.9)

10 weeks

HDRS

36/37

TG = 17.8 (2.0) CG = 17.0 (1.6)

33/35

TG = 5.1 (3.5) CG = 6.7 (4.4)

29 Page 29 of 47

TG = 11.1 (7.98) CG1 = 10.1 (5.70) CG2 = 9.1 (6.83)

Post-treatment

cr us

Population (study sample)

M

TG: Yang style Tai Chi 24-posture short-form; 60-minute sessions, three sessions per week CG1: Sliver Sneakers (a fitness group program with different types of exercise, e.g. aerobics, strength and range of movement, water aerobics, yoga); 60-minute sessions, three sessions per week CG2: Control group: received written materials and resources for participating in community-based physical activity, plus weekly phone call to inquire of their health status to provide individual attention TG: Yang style Tai Chi; 60-minute sessions twice weekly; plus at least 20-minute home based training daily with DVD provided CG: The attention control intervention (the wellness education and stretching program): two 60minute class sessions per week

Duration

Depression Scale

Pre-treatment

Old stroke survivors; Age (mean+SD): 69.9 ± 10.0 years; Gender (m/f): 77/68

12 weeks

CES-D 20

N (TG/C G) TG/CG 1/CG2: 53/44/4 8

Symptomatic tibiofemoral knee osteoarthritis; Age (mean+SD): 65.5 ± 7.6 years; Gender (m/f): 10/30

12 weeks

CES-D 20

20/20

Mean (SD) TG = 14.3 (9.8) CG1 = 11.1 (7.4) CG2 = 15.7 (11.9)

Post-treatment N (TG/ CG) TG/C G1/C G2: 53/44 /48

Mean (SD) TG = 14.0 (9.6) CG1 = 11.4 (9.6) CG2 = 13.6 (10.2)

Ac

Wang, et al, 200952

ce

pt

ed

TaylorPiliae, et al, 201451

Groups

an

Reference

30 Page 30 of 47

TG = 13.6 (11.7) CG = 9.3 (9.2)

20/20

Mean change (CI%): TG = -7.40 (10.88, 3.92) CG = 0.70 (4.18, 2.78)

cr us

Groups

Population (study sample)

an

Reference

Duration

Pre-treatment

Depression Scale

N (TG/C G) 17/17

Mean (SD)

N (TG/ CG) 16/13

Mean (SD)

TG: Yang style Tai Chi; 50 minutes / session, 1 session / week CG: Rehabilitation (non-resistance training + resistance training) control; 80 minutes / session, 1 session / week

With Cerebral vascular disorder; Age (mean+SD): 77.1±11.0 yrs; Gender (m/f): 9/25

12 weeks

GHQ-60 severe depression score

Yeh, et al, 201054

TG: Yang style Tai Chi; 60 minutes per session, 2 sessions/week; plus 3 times home based training CG: Usual care control

Moderate to severe Chronic Obstructive Pulmonary Disease; Age (mean+SD): 66±6yrs; Gender (m/f): 6/4

12 weeks

CES-D 20

5/5

Median (IQR): TG = 14 (11– 46) CG = 12 (2–17)

5/5

Median (IQR): TG = 5 (1–27) CG = 8 (0–17)

PMSD

8/8

TG = 4.0 (2.0) CG = 1.3 (2.0)

8/8

TG = 2.3 (2.0) CG = 3.0 (3.0)

Ac

ce

pt

ed

M

Wang, et al, 201053

Yeh, et al, 201355

TG: Cheng Man-Ch’ing’s Yang style; 60 minutes per session, two sessions per week; provided with a 35 minutes videotape that reviewed the exercises presented in class and was encouraged to practice at home at least 3 times per week CG: Aerobic exercise comparison group; 60 minutes per session, two sessions per week; provided with a 35 minutes videotape that reviewed the exercises presented in class and was encouraged to practice at home at least 3 times per week

12 weeks With Heart Failure with Preserved Ejection Fraction; Age (mean+SD): 66.0 ± 12.0 years; Gender (m/f): 8/8

31 Page 31 of 47

TG = 2.1 (2.2) CG = 1.3 (1.7)

Post-treatment

TG = 0.7 (0.8) CG = 1.2 (1.5)

cr us

Population (study sample)

M

TG: Yang style Tai Chi; 60 minutes per session, biweekly; plus home based practice at least 3 sessions per week CG: Waitlist control (usual care)

ed

Yeung, et al, 201256

Groups

an

Reference

With major depressive disorder; mean+SD): 55±10yrs; Gender (m/f): 9/30

Duration

12 weeks

Depression Scale

HRSD

Pre-treatment N (TG/C G) 26/13

Mean (SD) TG = 18 (3) CG = 17 (3)

Post-treatment N (TG/ CG) 25/13

Mean (SD) Mean change (SD): TG = 5.2 (5.1) CG = 4.5 (2.4)

Ac

ce

pt

TG = Tai Chi group; CG = control group; IQR = interquartile range GDS = Geriatric Depression Score; CDS = Cornell Depression Score; HRSD = Hamilton Rating Scale of Depression; CES-D20 = Centre for Epidemiological Study Depression Scale 20; BDI = Beck Depression Inventory; PMSD = Profile of Mood States Depression

32 Page 32 of 47

Ac ce pt e

d

M

an

us

cr

ip t

Table 3a Quality checklist (Primary outcome – analytic two-arm study)

33

Page 33 of 47

Item

Score 1=Yes/Not applicable, 0=No/Unclear

Design bias What was the type of design? a) randomized and allocation concealed – 3 points b) randomized only – 2 points c) prospective cohort – 1 point d) retrospective cohort or case control – 0 point [note of b)：1. Was the study described as randomized (this includes words such as randomly, random, and randomization)? Yes=1, No=0 2. Was the method used to generate the sequence of randomization described and appropriate (table of random numbers, computergenerated, etc)? Yes=1, No=0] Was the duration of active treatment appropriate for the demonstration of study outcome (e.g. >= 3 months for depressive symptoms)? Selection bias Did the inclusion/exclusion criteria remain consistent across the comparison groups of the study? Was the strategy for recruitment into the study the same across comparison groups (e.g. not from same populations or both groups were not recruited over the same time period)? Was the interval between the start of intervention and outcome the same across comparison groups, or if different, were appropriate analyses used to equalize this (e.g. time-to-event analyses)? Was attrition < 20%, or if not, was follow-up done for these subjects to ensure their loss was not related to outcome? Information bias Were the outcomes of interest in the study pre-specified? Were reproducible measures (clear name of predefined scale or clear details of non-predefined scale were presented) of study outcomes implemented in the same way across comparison groups? Were the outcome assessors blinded to the nature of intervention or control (e.g. Qigong, acupuncture or usual medical care)? Were the subjects blinded to the nature of intervention or control (e.g. Qigong, acupuncture or usual medical care)? Apart from blinding, were any other safeguards described and used for assuring the reliability of study outcomes (e.g. any of validated instruments, duplicated measurement, independent assessment)? Were data assessed and recorded in the same way for both comparison groups and across time points? Were interventions/exposures clearly defined (all essential components were described) and implemented in the same way across both study groups? Confounding bias Were the groups similar at baseline in key confounding variables or if 34 not were steps taken to achieve comparability of key confounders (e.g. through matching, stratification, interaction terms, multivariate analysis, or other statistical adjustment such as instrumental variables)?

6

7 8

9 10 11

12 13

14

us

an

5

M

4

d

3

Ac ce pt e

2

cr

ip t

1

Questions

Page 34 of 47

35

Page 35 of 47

d

Ac ce pt e us

an

M

cr

ip t

Item

Questions

Score 1=Yes, 0=No/Unclear

Ac ce pt e

d

M

an

us

cr

ip t

Design bias 1 Was the sample community-based? 2 Was the duration of active treatment appropriate for the demonstration of study outcome (>=3 months for depressive symptom)? Selection bias 3 Was attrition < 20%, or if not, was follow-up done for these subjects to ensure their loss was not related to outcome? Information bias 4 Were the outcomes of interest in the review pre-specified? 5 Were study outcomes clearly defined (clear name of predefined scale or clear details of non-predefined scale were presented)? 6 Were the outcome assessors blinded to the nature of intervention (e.g. Qigong, acupuncture or usual medical care)? 7 Apart from blinding, were any other safeguards described and used for assuring the reliability of study outcomes (e.g. any of validated instruments, duplicated measurement, independent assessment)? 8 Were data assessed and recorded in the same way at two time points? 9 Were interventions clearly defined (all essential components were described)? Confounding bias 10 Did researchers record and adequately control for any impact from change to concurrent intervention or exposure that could affect outcome (e.g. medication for outcome of interest, etc)? Analytical bias 11 Were effect sizes based on the data available at post assessment or pre-defined subgroups rather than a post hoc portion of the data? 12 Was intend-to-treat analyses conducted for the outcome of interest? 13 Were all Mean Gain and SE/CI or Mean and SD available (ie they did not need to be estimated from median/range or median/IQR)? Table 3b Quality checklist (Secondary outcome – single arm)

36

Page 36 of 47

Table 4 Sensitivity analysis for the primary outcome results by time frame, study sample size, intervention duration, severity of depressive symptoms, co-morbidities status and mean age

Ac ce pt e

d

Time frame of study <2011 >= 2011 Study sample size < 50 >= 50 Duration <3 months >= 3 months Severe depressive symptoms** Yes No Co-morbidities Yes No Mean age >= 60***

Tai Chi d = -0.07 (-0.44 to 0.31), 58%

ip t

Heterogeneity & datasets in the analysis I2, %, (N)

90 (8) 71 (8)

cr

-0.54 (-1.27 to 0.19) -0.44 (-0.76 to -0.12) -0.71 (-1.40 to -0.03) -0.41 (-0.80 to -0.02)

85 (7) 86 (9)

-0.39 (-0.97 to 0.19) -0.55 (-1.01 to -0.09)

87 (9) 83 (7)

-0.33 (-0.75 to 0.09) -0.66 (-1.28 to -0.05)

80 (9) 88 (7)

-0.40 (-0.91 to 0.11) -0.59 (-1.11 to -0.08)

86 (10) 82 (6)

us

Time frame of study <2010 > =2010 Study sample size < 60 >= 60 Duration < 3 months >= 3 months Severe depressive symptoms** Yes No Co-morbidities Yes No Mean age >= 60 Yes No

Estimate for altered criteria (Cohen’s d, 95% CI)

M

Qigong d = -0.48 (-0.83 to 0.12), 84%

Altered inclusion/exclusion criteria

an

Outcome & estimate, I2

-0.52 (-1.17 to 0.13) -0.44 (-0.83 to -0.05)

90 (8) 76 (8)

0.07 (-0.74 to 0.87) -0.22 (-0.52 to 0.08)

71 (5) 12 (5)

-0.32 (-1.06 to 0.42) 0.00 (-0.21 to 0.21)

67 (6) 21 (4)

-0.48 (-0.92 to -0.04) 0.00 (-0.40 to 0.40)

0 (2) 56 (8)

-0.62 (-1.67 to 0.43) 0.06 (-0.11 to 0.23)

72 (3) 0 (7)

0.07 (-0.14 to 0.28) -0.50 (-1.17 to 0.17)

8 (6) 66(4)

**Severe depressive symptoms were defined based on the cutoff in each symptom scale assessed at baseline (Yes=study sample with severe depressive symptoms, No=study sample with mild depressive symptoms). These cutoffs were as follows for the purposes of this review: BDI <10, BDI-II <14, GDS < 6, HADS <8, PMS < 35, DASS-21 <5, CES-D-10 <10, CES-D-20 <16, CSDD <11, GHQ-60 < 12, HAM-D < 8 and depression status inventory <50. ***All datasets had subjects with mean age >= 60 years.

37

Page 37 of 47

Databases: 811 studies Excluded: 544 studies • Non-English • Non-Qigong or Tai Chi related • Single group • Single session • Cross-sectional • Quasi-experiment • Non-randomized • Review

ip t

Excluded: 236 studies • Duplications

cr

575 studies

an

us

Excluded: 3 studies • Data presentation bias • Unknown scale

12 Tai Chi studies (12 Tai Chi single arms and 10 Tai Chi comparison arms)

Ac ce pt e

d

18 Qigong studies (19 Qigong single arms and 16 Qigong comparison arms) )

M

30 studies included in analysis

Figure 1 Flow of literature extraction

38

Page 38 of 47

d (95% CI) % Weight 0.45 ( -0.40, 1.29) 2.23 -2.23 ( -3.08, -1.37) 2.81 -1.73 ( -2.46, -1.00) 3.23 -0.59 ( -1.19, 0.01) 3.36 0.23 ( -0.32, 0.79) 3.81 -1.32 ( -1.87, -0.76) 5.78 -1.68 ( -2.19, -1.17) 5.72 -0.83 ( -1.34, -0.32) 5.77 -0.21 ( -0.68, 0.27) 6.92 0.20 ( -0.20, 0.61) 10.52 -0.27 ( -0.69, 0.16) 8.47 -0.13 ( -0.81, 0.54) 2.53 -0.15 ( -0.51, 0.22) 10.36 -0.57 ( -0.92, -0.23) 13.45 -0.84 ( -1.51, -0.18) 4.27 0.09 ( -0.31, 0.49) 10.77

us

cr

Study Burini 2006 Kim 2005 Yang 2005 Fang 2008 Tsang 2003 Chung 2010 Tsang 2006 Chow 2012 Stenlund 2009 Chen 2008A Smelson 2013 Chan 2013A Tsang 2013 Chan 2013B Tsang 2012 Chen 2013 Overall Q=96.42, p=0.00, I2=84%

-0.48 ( -0.83, -0.12) 0

1

b)

d (95% CI) 0.33 ( -0.92, -2.44 ( -3.87, -0.51 ( -1.32, 0.47 ( -0.27, -0.11 ( -0.71, -0.40 ( -0.88, 0.11 ( -0.11, 0.04 ( -0.35, -0.94 ( -1.99, -0.27 ( -1.26,

Ac ce pt e

d

M

Study Yeh 2010 Chou 2004 Cheng 2012 Wang 2010 Frye 2007 Lavretsky 2011 Lam 2010 Taylor-Piliae 2014 Cheon 2013 Yeh 2013

Overall Q=21.46, p=0.01, I2=58%

% Weight 1.58) 2.57 -1.01) 2.24 0.31) 3.48 1.21) 5.09 0.48) 6.27 0.08) 11.23 0.33) 45.55 0.43) 17.43 0.11) 2.48 0.71) 3.65

-0.07 ( -0.44, 0.31)

-2

100.00

an

-3 -2 -1 d

-4

ip t

a)

100.00

0

d

Figure 2 Effect sizes for standardized mean difference in depressive symptoms at the end of follow-up. a Qigong. b Tai Chi studies

39

Page 39 of 47

Qigong

b

Ac ce pt e

d

M

an

us

cr

ip t

a

Tai Chi

40

Page 40 of 47

Usual care

d

Ac ce pt e

Other exercise

d

M

an

us

cr

ip t

c

41

Page 41 of 47

Education

Miscellaneous

Ac ce pt e

d

M

an

us

f

cr

ip t

e

Figure 3 Standardized mean reduction in depressive symptoms from pre- to post-follow-up in single arms of a Qigong. b Tai Chi. c Usual care. d Other excises. e Education. f miscellaneous groups

42

Page 42 of 47

ip t cr

Ac ce pt e

d

M

an

us

Figure 4 Qigong studies: Doi (left) and funnel (right) plots of between-group outcomes demonstrating gross asymmetry of both favoring stronger effects

43

Page 43 of 47

ip t cr us

Ac ce pt e

d

M

an

Figure 5 Tai Chi studies: Doi (left) and funnel (right) plots of between-group outcomes demonstrating gross asymmetry of both favoring stronger effects

44

Page 44 of 47

Supplementary appendix 1. Detailed search strategy for PubMed

Ac ce pt e

d

M

an

us

cr

ip t

(("depressive disorder"[MeSH Terms] OR ("depressive"[All Fields] AND "disorder"[All Fields]) OR "depressive disorder"[All Fields] OR "depression"[All Fields] OR "depression"[MeSH Terms]) AND ("depression"[MeSH Terms] OR "depression"[All Fields] OR "depressions"[All Fields]) AND ("depression"[MeSH Terms] OR "depression"[All Fields] OR ("depressive"[All Fields] AND "symptoms"[All Fields]) OR "depressive symptoms"[All Fields]) AND ("depression"[MeSH Terms] OR "depression"[All Fields] OR ("depressive"[All Fields] AND "symptom"[All Fields]) OR "depressive symptom"[All Fields]) AND ("depression"[MeSH Terms] OR "depression"[All Fields] OR ("emotional"[All Fields] AND "depression"[All Fields]) OR "emotional depression"[All Fields]) AND ("qigong"[MeSH Terms] OR "qigong"[All Fields]) AND ("qigong"[MeSH Terms] OR "qigong"[All Fields] OR ("qi"[All Fields] AND "gong"[All Fields]) OR "qi gong"[All Fields]) AND ("qigong"[MeSH Terms] OR "qigong"[All Fields] OR ("ch'i"[All Fields] AND "kung"[All Fields]) OR "ch'i kung"[All Fields]) AND ("breathing exercises"[MeSH Terms] OR ("breathing"[All Fields] AND "exercises"[All Fields]) OR "breathing exercises"[All Fields]) AND ("breathing exercises"[MeSH Terms] OR ("breathing"[All Fields] AND "exercises"[All Fields]) OR "breathing exercises"[All Fields] OR ("respiratory"[All Fields] AND "muscle"[All Fields] AND "training"[All Fields]) OR "respiratory muscle training"[All Fields]) AND ("tai ji"[MeSH Terms] OR ("tai"[All Fields] AND "ji"[All Fields]) OR "tai ji"[All Fields] OR ("tai"[All Fields] AND "chi"[All Fields]) OR "tai chi"[All Fields]) AND ("tai ji"[MeSH Terms] OR ("tai"[All Fields] AND "ji"[All Fields]) OR "tai ji"[All Fields]) AND ("tai ji"[MeSH Terms] OR ("tai"[All Fields] AND "ji"[All Fields]) OR "tai ji"[All Fields]) AND ("tai ji"[MeSH Terms] OR ("tai"[All Fields] AND "ji"[All Fields]) OR "tai ji"[All Fields] OR ("tai"[All Fields] AND "ji"[All Fields] AND "quan"[All Fields]) OR "tai ji quan"[All Fields]) AND ("tai ji"[MeSH Terms] OR ("tai"[All Fields] AND "ji"[All Fields]) OR "tai ji"[All Fields] OR "taiji"[All Fields]) AND ("tai ji"[MeSH Terms] OR ("tai"[All Fields] AND "ji"[All Fields]) OR "tai ji"[All Fields] OR "taijiquan"[All Fields])) OR ("tai ji"[MeSH Terms] OR ("tai"[All Fields] AND "ji"[All Fields]) OR "tai ji"[All Fields] OR ("tai"[All Fields] AND "chi"[All Fields] AND "chuan"[All Fields]) OR "tai chi chuan"[All Fields])

45

Page 45 of 47

Overall Q=96.42, p=0.00, I2=84%

-0.57 ( -0.90, -0.25) -2

0

b)

d (95% CI) % Weight 0.33 ( -0.92, 1.58) 4.67 -2.44 ( -3.87, -1.01) 3.76 -0.51 ( -1.32, 0.31) 8.50 0.47 ( -0.27, 1.21) 9.45 -0.11 ( -0.71, 0.48) 11.86 -0.40 ( -0.88, 0.08) 13.98 0.11 ( -0.11, 0.33) 19.22 0.04 ( -0.35, 0.43) 15.83 -0.94 ( -1.99, 0.11) 6.09 -0.27 ( -1.26, 0.71) 6.64

Ac ce pt e

d

Study Yeh 2010 Chou 2004 Cheng 2012 Wang 2010 Frye 2007 Lavretsky 2011 Lam 2010 Taylor-Piliae 2014 Cheon 2013 Yeh 2013

100.00

M

d

cr

% Weight 1.29) 5.05 -1.37) 5.03 -1.00) 5.53 0.01) 6.09 0.79) 6.27 -0.76) 6.29 -1.17) 6.46 -0.32) 6.47 0.27) 6.60 0.61) 6.86 0.16) 6.79 0.54) 5.78 0.22) 7.01 -0.23) 7.08 -0.18) 5.80 0.49) 6.88

us

d (95% CI) 0.45 ( -0.40, -2.23 ( -3.08, -1.73 ( -2.46, -0.59 ( -1.19, 0.23 ( -0.32, -1.32 ( -1.87, -1.68 ( -2.19, -0.83 ( -1.34, -0.21 ( -0.68, 0.20 ( -0.20, -0.27 ( -0.69, -0.13 ( -0.81, -0.15 ( -0.51, -0.57 ( -0.92, -0.84 ( -1.51, 0.09 ( -0.31,

an

Study Burini 2006 Kim 2005 Yang 2005 Fang 2008 Tsang 2003 Chung 2010 Tsang 2006 Chow 2012 Stenlund 2009 Chen 2008A Smelson 2013 Chan 2013A Tsang 2013 Chan 2013B Tsang 2012 Chen 2013

ip t

a)

-0.19 ( -0.50, 0.11)

Overall Q=21.46, p=0.01, I2=58%

-4

-2

100.00

0

d

Supplementary appendix 2. Random effects results for the primary outcome. Effect sizes for standardized mean difference in depressive symptoms at the end of follow-up using the random effects model: a Qigong studies; b Tai Chi studies

46

Page 46 of 47

Highlights: •

Qigong has been shown to be beneficial for reducing depressive symptom severity

•

Tai Chi appears to be ineffective in reducing depressive symptom severity

•

All other forms of care included in this study (including usual care, other exercises and

ip t

education) do not result in a change in depressive symptom severity over time suggesting that heterogeneity was not because of the range of control interventions combined in the

Ac ce pt e

d

M

an

us

cr

primary analysis.

47

Page 47 of 47