- Email: [email protected]
Are Indian yoga trials more likely to be positive than those from other countries? A systematic review of randomized controlled trials
Contemporary Clinical Trials 41 (2015) 269–272
Contents lists available at ScienceDirect
Contemporary Clinical Trials journal homepage: www.elsevier.com/locate/conclintrial
Short communication
Are Indian yoga trials more likely to be positive than those from other countries? A systematic review of randomized controlled trials Holger Cramer ⁎, Romy Lauche, Jost Langhorst, Gustav Dobos Department of Internal and Integrative Medicine, Kliniken Essen-Mitte, Faculty of Medicine, University of Duisburg-Essen, Essen, Germany
a r t i c l e
i n f o
Article history: Received 20 December 2014 Received in revised form 11 February 2015 Accepted 12 February 2015 Available online 20 February 2015 Keywords: Yoga Complementary therapies Randomized controlled trial India
a b s t r a c t Objective: To determine whether the conclusions of randomized controlled trials (RCTs) of yoga are more likely to be positive when they were conducted in India and/or when they are published in complementary and alternative medicine (CAM) specialty journals. Methods: Medline/PubMed, Scopus, the Cochrane Library, IndMED, and the tables of content of yoga specialty journals not listed in medical databases were screened through February 2014 for RCTs comparing yoga interventions to non-yoga interventions. The RCTs' conclusions were classified as positive (yoga is helpful for a respective condition) or not positive; and compared between RCTs that were a) conducted in India vs. outside India, and b) published in a CAM specialty journal or another type of journal. Results: A total of 306 RCTs were included; 131 from India and 175 from other countries; and 84 from CAM specialty journals and 222 from other types of journals. Positive conclusions were reached in 277 RCTs (91%); with more positive RCTs being conducted in India than elsewhere (odds ratio = 24.8; 95% confidence interval = 3.3, 184.5; p b 0.001) while type of journal was not associated with the direction of the conclusions (odds ratio = 1.2; 95% confidence interval = 0.5, 2.9; p = 0.828). Conclusions: RCTs on yoga that are conducted in India have about 25 times the odds of reaching positive conclusions as those conducted elsewhere. Indian trials should be dealt with carefully when evaluating the helpfulness of yoga for patients in other countries and vice versa. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Yoga is gaining increased popularity as a therapeutic practice; with more than 20 million Americans (9% of the USA's population) reporting they practiced yoga in 2012. Furthermore, more than two thirds of practitioners utilized yoga explicitly to improve their health status [1]. This tendency might at least partly depend on the strong increase in research evidence for the helpfulness of yoga for a variety of conditions in recent years [2]. Although this research is increasingly ⁎ Corresponding author at: Kliniken Essen-Mitte, Klinik für Naturheilkunde und Integrative Medizin, Knappschafts-Krankenhaus, Am Deimelsberg 34a, 45276 Essen, Germany.
http://dx.doi.org/10.1016/j.cct.2015.02.005 1551-7144/© 2015 Elsevier Inc. All rights reserved.
conducted in Europe or North America, almost half of published randomized controlled trials (RCTs) on yoga still originate from India [2]. It has been argued that clinical trials originating in certain countries, especially Asian countries, are more likely to have positive results than are those from other countries, and that this might inadequately inflate the evidence for positive effects [3]. Moreover, it has been specifically argued that clinical trials on complementary and alternative medicine (CAM) interventions like yoga are more likely to be published when reaching positive conclusions, at least in CAM specialty journals [4,5]. The aim of this analysis was to estimate in how far conclusions of RCTs on yoga depend on country of origin and type of journal a respective RCT was published in.
270
H. Cramer et al. / Contemporary Clinical Trials 41 (2015) 269–272
Fig. 1. Flowchart of the results of the literature search.
2. Methods
2.2. Literature search methods
This is a secondary analysis of a previously published bibliometric analysis of yoga RCTs that descriptively summarized characteristics of randomized controlled trials of yoga [2]. This prior analysis did not assess conclusions of yoga trials [2]. Where applicable, reporting is in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and MetaAnalyses) guidelines [6].
Four electronic databases, Medline/PubMed, Scopus, IndMED and the Cochrane Library were searched from their inception through February 12, 2014. The literature search was constructed around search terms for “yoga” and a filter for retrieving randomized controlled trials [7]. The complete search strategy for Medline/PubMed has been published elsewhere [2]. The reference lists of identified original articles or reviews and the tables of contents of the Journal of Yoga & Physical Therapy and the International Scientific Yoga Journal SENSE were searched manually for additional eligible studies. Identified abstracts were screened independently by two review authors; and potentially eligible articles were then read in full by two review authors to determine whether they actually met the eligibility criteria.
2.1. Eligibility criteria Types of studies. RCTs, cluster-randomized trials and randomized cross-over studies were eligible. No language restrictions were applied; if necessary, language experts were consulted. Duplicate publications; this is multiple articles reporting identical or different results on already published studies; were excluded. Types of participants. Studies of all types of participants were eligible. Types of interventions. Studies were eligible if they compared yoga interventions to one or more non-yoga interventions or untreated control groups. No restrictions were applied regarding the tradition, specific yoga practices, length, frequency or duration of the studied yoga programs. Head-to-head comparisons of different yoga interventions without a non-yoga control group were excluded. Types of outcomes. Studies with all types of outcomes were eligible.
2.3. Data extraction Bibliometric data (country of origin, journal of publication) were extracted independently by two authors, using a standardized data extraction form. Country of origin, i.e. the country where the trial was conducted, was categorized as India vs. other countries. There were no multi-country studies [2]. Journal of publication was categorized as CAM specialty journal vs. other types of journals. Journals were categorizes as CAM specialty journals if they explicitly focused on complementary, alternative and/or integrative medicine in their aims and scopes. Journals explicitly focusing on yoga were also coded as CAM specialty journals. The conclusions by the authors of the original articles were extracted from the studies'
H. Cramer et al. / Contemporary Clinical Trials 41 (2015) 269–272
abstracts and rated by a reviewer blinded to the study aims as a) positive: the yoga intervention was stated to be helpful for a respective condition or symptom and/or to be superior to at least one non-yoga control group; b) neutral: no clear statement regarding helpfulness or superiority regarding a respective condition or symptom was made; or c) negative: the yoga intervention was stated to be unhelpful for a respective condition or symptom and/or to be inferior or not superior to all non-yoga control groups. Where conclusions were rated to be neutral or where no conclusion was provided in the study's abstract, conclusions from the respective study's discussion section were additionally checked. Conclusions were further categorized as positive vs. not positive. It was originally planned to also extract the direction of results. A study was defined to have positive results if yoga was described to be significantly superior (p ≤ 0.05) to at least one non-yoga control group on the primary outcome measure. However, this approach was not possible since a) only a minority of the included studies a priori defined a primary outcome measure and b) a considerable number of studies did not statistically compare outcomes between groups but based their conclusions on within-group analyses. 2.4. Statistical analysis Data were analyzed descriptively using IBM SPSS® Statistics for Windows (release 22.0. Armonk, NY: IBM Corp). Chi square tests were used to compare conclusions between a) Indian and non-Indian studies, and b) studies published in CAM specialty journals and studies published in other types of journals; and odds ratios (OR) with their respective 95% confidence intervals (CI) were calculated. Independent predictors of positive conclusions were identified using multiple logistic regression analysis. A backward Table 1 Conclusions of the included RCTs by country of origin. Country of origin
Total number of trials
Number of positive trials (%)
Number of neutral trials (%)
Number of negative trials (%)
Australia Belgium Brazil Canada China Cuba Ethiopia Germany India Iran Iraq Ireland Japan Korea Poland Slovenia Spain Sweden Taiwan Thailand Turkey United Kingdom USA
10 1 5 6 3 2 1 8 131 10 1 1 5 3 2 2 1 2 4 2 2 18 86
9 (90%) 1 (100%) 5 (100%) 2 (33%) 3 (100%) 2 (100%) 1 (100%) 6 (75%) 130 (99%) 9 (90%) 1 (100%) 1 (100%) 5 (100%) 3 (100%) 2 (100%) 2 (100%) 1 (100%) 2 (100%) 4 (100%) 2 (100%) 2 (100%) 13 (72%) 71 (83%)
0 (0%) 0 (0%) 0 (0%) 2 (33%) 0 (0%) 0 (0%) 0 (0%) 1 (13%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 2 (11%) 12 (14%)
1 (10%) 0 (0%) 0 (0%) 2 (33%) 0 (0%) 0 (0%) 0 (0%) 1 (13%) 1 (1%) 1 (10%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 3 (17%) 3 (3%)
271
stepwise procedure with a Wald statistic p-value of ≤ 0.05 and the independent variables “India” (yes/no) and “CAM specialty journal” (yes/no) was used. 3. Results A total of 1531 non-duplicate records were screened out of which 1041 were excluded because they were not randomized or because they did not include yoga as an intervention. Of the remaining 490 full-text records assessed for eligibility, 184 were excluded because they either were published as conference abstracts only, were not randomized, were duplicate publications, and/or did not include yoga interventions or nonyoga control groups. The final analysis was conducted on 306 RCTs (Fig. 1). The yoga trials originated from 23 different countries with India (131 RCTs) and the USA (86 RCTs) contributing by far the most RCTs (Table 1). Regarding types of journals, 84 RCTs were published in CAM specialty journals, and 222 were published in other journals. A total of 277 (91%), 17 (6%), and 12 (4%) RCTs reached positive, neutral, and negative conclusions, respectively. Out of 131 Indian RCTs, 130 (99%) reached a positive conclusion compared to 147 (84%) out of 175 RCTs from other countries (OR = 24.8; 95%CI = 3.3, 184.5; p b 0.001). Conversely, 77 (92%) out of 84 RCTs published in CAM specialty journals reached a positive conclusion compared to 200 (90%) out of 222 RCTs published in other journals (OR = 1.2; 95%CI = 0.5, 2.9; p = 0.828). In multiple logistic regression, only Indian origin of the RCTs independently predicted positive conclusions (p = 0.002). 4. Discussion About 90% of all published RCTs on yoga reached positive conclusions. The proportion of positive and neutral or negative conclusions was independent of the type of journal the RCT was published in. In 1995 and 2000, only 1% and 5% of the trials published in a number of CAM specialty journals, respectively, had reached negative conclusions [4,5]. While the authors reporting this finding concluded that trials published in those journals are most likely biased, they did not compare RCTs that were published in CAM specialty journals with those that were published in other types of journals. It has been shown that about 89% of all published RCTs are positive regardless of intervention or journal type; but strongly depending on country of origin [3]. In line with this, the present findings indicate that conclusions of Indian yoga RCTs have about 25 times the odds of being positive as RCTs conducted elsewhere. Although not the focus of this analysis, it is obvious that RCTs from several other countries, especially Asian countries, also had extremely high frequencies of positive conclusions. There are a number of possible explanations for this finding: First, yoga might be more effective in India than elsewhere. Being an indigenous practice, yoga is part of the Indian spiritual and philosophical tradition [8] and might thus induce larger unspecific effects in India than in other countries. Moreover, Indian yoga interventions often are much more intense than interventions in Europe or North America [9]; and Indian yoga instructors might be more skilled or more dedicated in teaching yoga than instructors from other cultural and spiritual backgrounds. Second, Indian researchers might be more willing to publish a positive trial on yoga than a negative one.
272
H. Cramer et al. / Contemporary Clinical Trials 41 (2015) 269–272
Publication bias; this is the tendency to publish only positive results; is a worldwide problem [7,10]. Publication bias might be more pronounced for indigenous medical treatments, although this has not proven true for acupuncture in China [3]. On the other hand, publication bias could apply to all RCTs conducted in India regardless of the intervention type, as has been shown for China and Russia [3]. To the best of our knowledge, this issue has not yet been investigated for Indian RCTs outside the field of yoga research. Third, the positive conclusions might be based on inadequate statistics used. While it was originally planned to also compare the direction of results between studies, this proved impossible because of the common lack of an a priori defined primary outcome measure and often even the lack of between-group comparisons. Thus any positive change over time in any variable might lead to a positive conclusion. However, this lack of methodological rigor was not limited to Indian RCTs. Regardless of the specific reasons for the differences in conclusions between RCTs conducted in India and elsewhere, Indian trials on yoga should be dealt with carefully when evaluating the helpfulness of yoga for patients in other countries and vice versa; be it in clinical practice or in the process of conducting systematic reviews or medical guidelines. The same might apply to trials from other Asian countries. Further in-depth studies are needed to clarify reasons for the differences in conclusions between RCTs conducted in India and elsewhere. Acknowledgement This review was supported by a grant from the Rut- and Klaus-Bahlsen-Foundation. The funding source had no influence
on the design or conduct of the review; the collection, management, analysis, or interpretation of the data; or in the preparation, review, or approval of the manuscript. The authors would like to thank Jana Hochstein, Essen, Germany, for her assistance in data extraction; and Dr. Petra Klose, Essen, Germany, and Dr. Hoda Azizi, Mashhad, Iran, for their assistance in assessing the Chinese, Japanese, and Farsi manuscripts.
References [1] Zhang Y. American adult yoga practice: preliminary findings from NHIS 2012 data. J Altern Complement Med 2014;20:A122. [2] Cramer H, Lauche R, Dobos G. Characteristics of randomized controlled trials of yoga: a bibliometric analysis. BMC Complement Altern Med 2014; 14:328. [3] Vickers A, Goyal N, Harland R, Rees R. Do certain countries produce only positive results? A systematic review of controlled trials. Control Clin Trials 1998;19:159–66. [4] Ernst E, Pittler MH. Alternative therapy bias. Nature 1997;385:480. [5] Schmidt K, Pittler MH, Ernst E. Bias in alternative medicine is still rife but is diminishing. BMJ 2001;323:1071. [6] Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ 2009; 339:b2535. [7] Higgins JPT, Green S. Cochrane handbook for systematic reviews of interventions. West Sussex: John Wiley & Sons Ltd; 2008. [8] Feuerstein G. The yoga tradition. Prescott: Hohm Press; 1998. [9] Cramer H, Lauche R, Langhorst J, Dobos G. Effectiveness of yoga for menopausal symptoms: a systematic review and meta-analysis of randomized controlled trials. Evid Based Complement Alternat Med 2012;2012:863905. [10] Dickersin K. The existence of publication bias and risk factors for its occurrence. JAMA 1990;263:1385–9.
Contents lists available at ScienceDirect
Contemporary Clinical Trials journal homepage: www.elsevier.com/locate/conclintrial
Short communication
Are Indian yoga trials more likely to be positive than those from other countries? A systematic review of randomized controlled trials Holger Cramer ⁎, Romy Lauche, Jost Langhorst, Gustav Dobos Department of Internal and Integrative Medicine, Kliniken Essen-Mitte, Faculty of Medicine, University of Duisburg-Essen, Essen, Germany
a r t i c l e
i n f o
Article history: Received 20 December 2014 Received in revised form 11 February 2015 Accepted 12 February 2015 Available online 20 February 2015 Keywords: Yoga Complementary therapies Randomized controlled trial India
a b s t r a c t Objective: To determine whether the conclusions of randomized controlled trials (RCTs) of yoga are more likely to be positive when they were conducted in India and/or when they are published in complementary and alternative medicine (CAM) specialty journals. Methods: Medline/PubMed, Scopus, the Cochrane Library, IndMED, and the tables of content of yoga specialty journals not listed in medical databases were screened through February 2014 for RCTs comparing yoga interventions to non-yoga interventions. The RCTs' conclusions were classified as positive (yoga is helpful for a respective condition) or not positive; and compared between RCTs that were a) conducted in India vs. outside India, and b) published in a CAM specialty journal or another type of journal. Results: A total of 306 RCTs were included; 131 from India and 175 from other countries; and 84 from CAM specialty journals and 222 from other types of journals. Positive conclusions were reached in 277 RCTs (91%); with more positive RCTs being conducted in India than elsewhere (odds ratio = 24.8; 95% confidence interval = 3.3, 184.5; p b 0.001) while type of journal was not associated with the direction of the conclusions (odds ratio = 1.2; 95% confidence interval = 0.5, 2.9; p = 0.828). Conclusions: RCTs on yoga that are conducted in India have about 25 times the odds of reaching positive conclusions as those conducted elsewhere. Indian trials should be dealt with carefully when evaluating the helpfulness of yoga for patients in other countries and vice versa. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Yoga is gaining increased popularity as a therapeutic practice; with more than 20 million Americans (9% of the USA's population) reporting they practiced yoga in 2012. Furthermore, more than two thirds of practitioners utilized yoga explicitly to improve their health status [1]. This tendency might at least partly depend on the strong increase in research evidence for the helpfulness of yoga for a variety of conditions in recent years [2]. Although this research is increasingly ⁎ Corresponding author at: Kliniken Essen-Mitte, Klinik für Naturheilkunde und Integrative Medizin, Knappschafts-Krankenhaus, Am Deimelsberg 34a, 45276 Essen, Germany.
http://dx.doi.org/10.1016/j.cct.2015.02.005 1551-7144/© 2015 Elsevier Inc. All rights reserved.
conducted in Europe or North America, almost half of published randomized controlled trials (RCTs) on yoga still originate from India [2]. It has been argued that clinical trials originating in certain countries, especially Asian countries, are more likely to have positive results than are those from other countries, and that this might inadequately inflate the evidence for positive effects [3]. Moreover, it has been specifically argued that clinical trials on complementary and alternative medicine (CAM) interventions like yoga are more likely to be published when reaching positive conclusions, at least in CAM specialty journals [4,5]. The aim of this analysis was to estimate in how far conclusions of RCTs on yoga depend on country of origin and type of journal a respective RCT was published in.
270
H. Cramer et al. / Contemporary Clinical Trials 41 (2015) 269–272
Fig. 1. Flowchart of the results of the literature search.
2. Methods
2.2. Literature search methods
This is a secondary analysis of a previously published bibliometric analysis of yoga RCTs that descriptively summarized characteristics of randomized controlled trials of yoga [2]. This prior analysis did not assess conclusions of yoga trials [2]. Where applicable, reporting is in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and MetaAnalyses) guidelines [6].
Four electronic databases, Medline/PubMed, Scopus, IndMED and the Cochrane Library were searched from their inception through February 12, 2014. The literature search was constructed around search terms for “yoga” and a filter for retrieving randomized controlled trials [7]. The complete search strategy for Medline/PubMed has been published elsewhere [2]. The reference lists of identified original articles or reviews and the tables of contents of the Journal of Yoga & Physical Therapy and the International Scientific Yoga Journal SENSE were searched manually for additional eligible studies. Identified abstracts were screened independently by two review authors; and potentially eligible articles were then read in full by two review authors to determine whether they actually met the eligibility criteria.
2.1. Eligibility criteria Types of studies. RCTs, cluster-randomized trials and randomized cross-over studies were eligible. No language restrictions were applied; if necessary, language experts were consulted. Duplicate publications; this is multiple articles reporting identical or different results on already published studies; were excluded. Types of participants. Studies of all types of participants were eligible. Types of interventions. Studies were eligible if they compared yoga interventions to one or more non-yoga interventions or untreated control groups. No restrictions were applied regarding the tradition, specific yoga practices, length, frequency or duration of the studied yoga programs. Head-to-head comparisons of different yoga interventions without a non-yoga control group were excluded. Types of outcomes. Studies with all types of outcomes were eligible.
2.3. Data extraction Bibliometric data (country of origin, journal of publication) were extracted independently by two authors, using a standardized data extraction form. Country of origin, i.e. the country where the trial was conducted, was categorized as India vs. other countries. There were no multi-country studies [2]. Journal of publication was categorized as CAM specialty journal vs. other types of journals. Journals were categorizes as CAM specialty journals if they explicitly focused on complementary, alternative and/or integrative medicine in their aims and scopes. Journals explicitly focusing on yoga were also coded as CAM specialty journals. The conclusions by the authors of the original articles were extracted from the studies'
H. Cramer et al. / Contemporary Clinical Trials 41 (2015) 269–272
abstracts and rated by a reviewer blinded to the study aims as a) positive: the yoga intervention was stated to be helpful for a respective condition or symptom and/or to be superior to at least one non-yoga control group; b) neutral: no clear statement regarding helpfulness or superiority regarding a respective condition or symptom was made; or c) negative: the yoga intervention was stated to be unhelpful for a respective condition or symptom and/or to be inferior or not superior to all non-yoga control groups. Where conclusions were rated to be neutral or where no conclusion was provided in the study's abstract, conclusions from the respective study's discussion section were additionally checked. Conclusions were further categorized as positive vs. not positive. It was originally planned to also extract the direction of results. A study was defined to have positive results if yoga was described to be significantly superior (p ≤ 0.05) to at least one non-yoga control group on the primary outcome measure. However, this approach was not possible since a) only a minority of the included studies a priori defined a primary outcome measure and b) a considerable number of studies did not statistically compare outcomes between groups but based their conclusions on within-group analyses. 2.4. Statistical analysis Data were analyzed descriptively using IBM SPSS® Statistics for Windows (release 22.0. Armonk, NY: IBM Corp). Chi square tests were used to compare conclusions between a) Indian and non-Indian studies, and b) studies published in CAM specialty journals and studies published in other types of journals; and odds ratios (OR) with their respective 95% confidence intervals (CI) were calculated. Independent predictors of positive conclusions were identified using multiple logistic regression analysis. A backward Table 1 Conclusions of the included RCTs by country of origin. Country of origin
Total number of trials
Number of positive trials (%)
Number of neutral trials (%)
Number of negative trials (%)
Australia Belgium Brazil Canada China Cuba Ethiopia Germany India Iran Iraq Ireland Japan Korea Poland Slovenia Spain Sweden Taiwan Thailand Turkey United Kingdom USA
10 1 5 6 3 2 1 8 131 10 1 1 5 3 2 2 1 2 4 2 2 18 86
9 (90%) 1 (100%) 5 (100%) 2 (33%) 3 (100%) 2 (100%) 1 (100%) 6 (75%) 130 (99%) 9 (90%) 1 (100%) 1 (100%) 5 (100%) 3 (100%) 2 (100%) 2 (100%) 1 (100%) 2 (100%) 4 (100%) 2 (100%) 2 (100%) 13 (72%) 71 (83%)
0 (0%) 0 (0%) 0 (0%) 2 (33%) 0 (0%) 0 (0%) 0 (0%) 1 (13%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 2 (11%) 12 (14%)
1 (10%) 0 (0%) 0 (0%) 2 (33%) 0 (0%) 0 (0%) 0 (0%) 1 (13%) 1 (1%) 1 (10%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) 3 (17%) 3 (3%)
271
stepwise procedure with a Wald statistic p-value of ≤ 0.05 and the independent variables “India” (yes/no) and “CAM specialty journal” (yes/no) was used. 3. Results A total of 1531 non-duplicate records were screened out of which 1041 were excluded because they were not randomized or because they did not include yoga as an intervention. Of the remaining 490 full-text records assessed for eligibility, 184 were excluded because they either were published as conference abstracts only, were not randomized, were duplicate publications, and/or did not include yoga interventions or nonyoga control groups. The final analysis was conducted on 306 RCTs (Fig. 1). The yoga trials originated from 23 different countries with India (131 RCTs) and the USA (86 RCTs) contributing by far the most RCTs (Table 1). Regarding types of journals, 84 RCTs were published in CAM specialty journals, and 222 were published in other journals. A total of 277 (91%), 17 (6%), and 12 (4%) RCTs reached positive, neutral, and negative conclusions, respectively. Out of 131 Indian RCTs, 130 (99%) reached a positive conclusion compared to 147 (84%) out of 175 RCTs from other countries (OR = 24.8; 95%CI = 3.3, 184.5; p b 0.001). Conversely, 77 (92%) out of 84 RCTs published in CAM specialty journals reached a positive conclusion compared to 200 (90%) out of 222 RCTs published in other journals (OR = 1.2; 95%CI = 0.5, 2.9; p = 0.828). In multiple logistic regression, only Indian origin of the RCTs independently predicted positive conclusions (p = 0.002). 4. Discussion About 90% of all published RCTs on yoga reached positive conclusions. The proportion of positive and neutral or negative conclusions was independent of the type of journal the RCT was published in. In 1995 and 2000, only 1% and 5% of the trials published in a number of CAM specialty journals, respectively, had reached negative conclusions [4,5]. While the authors reporting this finding concluded that trials published in those journals are most likely biased, they did not compare RCTs that were published in CAM specialty journals with those that were published in other types of journals. It has been shown that about 89% of all published RCTs are positive regardless of intervention or journal type; but strongly depending on country of origin [3]. In line with this, the present findings indicate that conclusions of Indian yoga RCTs have about 25 times the odds of being positive as RCTs conducted elsewhere. Although not the focus of this analysis, it is obvious that RCTs from several other countries, especially Asian countries, also had extremely high frequencies of positive conclusions. There are a number of possible explanations for this finding: First, yoga might be more effective in India than elsewhere. Being an indigenous practice, yoga is part of the Indian spiritual and philosophical tradition [8] and might thus induce larger unspecific effects in India than in other countries. Moreover, Indian yoga interventions often are much more intense than interventions in Europe or North America [9]; and Indian yoga instructors might be more skilled or more dedicated in teaching yoga than instructors from other cultural and spiritual backgrounds. Second, Indian researchers might be more willing to publish a positive trial on yoga than a negative one.
272
H. Cramer et al. / Contemporary Clinical Trials 41 (2015) 269–272
Publication bias; this is the tendency to publish only positive results; is a worldwide problem [7,10]. Publication bias might be more pronounced for indigenous medical treatments, although this has not proven true for acupuncture in China [3]. On the other hand, publication bias could apply to all RCTs conducted in India regardless of the intervention type, as has been shown for China and Russia [3]. To the best of our knowledge, this issue has not yet been investigated for Indian RCTs outside the field of yoga research. Third, the positive conclusions might be based on inadequate statistics used. While it was originally planned to also compare the direction of results between studies, this proved impossible because of the common lack of an a priori defined primary outcome measure and often even the lack of between-group comparisons. Thus any positive change over time in any variable might lead to a positive conclusion. However, this lack of methodological rigor was not limited to Indian RCTs. Regardless of the specific reasons for the differences in conclusions between RCTs conducted in India and elsewhere, Indian trials on yoga should be dealt with carefully when evaluating the helpfulness of yoga for patients in other countries and vice versa; be it in clinical practice or in the process of conducting systematic reviews or medical guidelines. The same might apply to trials from other Asian countries. Further in-depth studies are needed to clarify reasons for the differences in conclusions between RCTs conducted in India and elsewhere. Acknowledgement This review was supported by a grant from the Rut- and Klaus-Bahlsen-Foundation. The funding source had no influence
on the design or conduct of the review; the collection, management, analysis, or interpretation of the data; or in the preparation, review, or approval of the manuscript. The authors would like to thank Jana Hochstein, Essen, Germany, for her assistance in data extraction; and Dr. Petra Klose, Essen, Germany, and Dr. Hoda Azizi, Mashhad, Iran, for their assistance in assessing the Chinese, Japanese, and Farsi manuscripts.
References [1] Zhang Y. American adult yoga practice: preliminary findings from NHIS 2012 data. J Altern Complement Med 2014;20:A122. [2] Cramer H, Lauche R, Dobos G. Characteristics of randomized controlled trials of yoga: a bibliometric analysis. BMC Complement Altern Med 2014; 14:328. [3] Vickers A, Goyal N, Harland R, Rees R. Do certain countries produce only positive results? A systematic review of controlled trials. Control Clin Trials 1998;19:159–66. [4] Ernst E, Pittler MH. Alternative therapy bias. Nature 1997;385:480. [5] Schmidt K, Pittler MH, Ernst E. Bias in alternative medicine is still rife but is diminishing. BMJ 2001;323:1071. [6] Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ 2009; 339:b2535. [7] Higgins JPT, Green S. Cochrane handbook for systematic reviews of interventions. West Sussex: John Wiley & Sons Ltd; 2008. [8] Feuerstein G. The yoga tradition. Prescott: Hohm Press; 1998. [9] Cramer H, Lauche R, Langhorst J, Dobos G. Effectiveness of yoga for menopausal symptoms: a systematic review and meta-analysis of randomized controlled trials. Evid Based Complement Alternat Med 2012;2012:863905. [10] Dickersin K. The existence of publication bias and risk factors for its occurrence. JAMA 1990;263:1385–9.