Introduction

Osteoarthritis is the most common type of arthritis among adults, affecting more than 500 million people, globally [1, 2]. It is a long-term degenerative condition of the joints, characterised by main symptoms including pain, stiffness, and difficulty in movement [1]. It commonly affects the joints of the knees, hips, hands, and spine [1]. Osteoarthritis poses tremendous health (physical and psychological), social, and economic implications for the affected individual and society, severely affecting the individual’s self-efficacy and quality of life [3,4,5,6]. There is no cure for osteoarthritis, however, it can be managed [7]. The main aim of osteoarthritis management is to minimise joint pain and loss of function [7]. The traditional approaches to managing the symptoms rely mainly on pharmacological options (e.g., Non-Steroidal Anti-Inflammatory Drugs (NSAIDs)) along with non-pharmacological approaches (e.g., moderate-to-high intensity exercises) [8]. However, side effects of long-term usage of pharmacological options (e.g., gastrointestinal toxicity), limited adherence to non-pharmacological approaches (due to exercise-related injuries), and costs associated with both approaches (e.g., treatment-related costs, equipment costs for exercise) are some of the reasons which potentially limit their use among individuals with osteoarthritis [1, 9].

Yoga, a non-pharmacological treatment approach, has been recommended for osteoarthritis by an international osteoarthritis clinical guideline in 2019 [10] and some studies have also proposed yoga as a beneficial practice for people with arthritis [11, 12]. The ancient practice of yoga originated in the Indian subcontinent and imparts a sense of well-being of the body and mind [13]. Yoga philosophy and practice were first described by Patanjali in the classic text Yoga Sutras [14]. The multi-factorial approach of yoga includes components such as yogic poses (asana), breathing practices (pranayama), and meditation (dhyana) and relaxation practices, along with a moderated lifestyle [14]. Yoga practice generally begins with slow movement sequences to increase blood flow and warm up muscles, followed by holding certain yogic poses (e.g., extension, rotation) that engage the muscles in contraction [15, 16]. Movement of joints increases flexibility whereas standing poses improve balance and coordination by strengthening major muscle groups (e.g., hamstring muscles and quads), potentially reducing pain and improving function [17,18,19]. The worldwide popularity of yoga is rising, with nearly 300 million people across the world involved in its practice [20]. Generally, yoga is easy to learn with low risk involved, demands a low-to-moderate level of supervision, is inexpensive to maintain because of the minimal equipment requirement, and can be practised indoors and outdoors [21,22,23].

Existing systematic reviews and meta-analyses have reported the beneficial effects of yoga interventions on osteoarthritis symptoms, such as reduced pain and improved function [24,25,26,27,28,29,30]. These reviews have included randomised controlled trials (RCTs) [24,25,26,27,28,29,30], and in one review, also other study designs [29]. A systematic review and meta-analysis of 20 RCTs and 2 case series on knee and hip osteoarthritis showed that yoga significantly reduced pain scores (mean difference (MD) − 1.82, 95% confidence interval (CI) − 2.96 to − 0.67) and improved physical function scores (− 6.07, − 9.75 to − 2.39) compared to no intervention or usual care [29]. No adverse events related to yoga were reported [24,25,26,27,28,29,30]. However, all the above-mentioned systematic reviews have only described or reported but not synthesised the content, structure, and delivery characteristics of yoga interventions to manage osteoarthritis [24,25,26,27,28,29,30]. Therefore, this systematic review and meta-analysis aimed to fill this gap in the existing literature by narratively synthesising the content, structure, and delivery characteristics of effective yoga interventions for managing osteoarthritis symptoms. The intention was to facilitate the identification of the key features of effective yoga interventions for osteoarthritis, which could be combined for use in subsequent trials to test the effects of this yoga intervention on osteoarthritis patients.

Methods

This systematic review followed the JBI methodology for systematic reviews of effectiveness and the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [31, 32]. This review was conducted according to a priori published protocol [33] and registered with PROSPERO (CRD42022298155).

Inclusion criteria

Population

We included studies conducted among adults (aged ≥ 18 years) diagnosed with osteoarthritis of one or more joints. No restrictions were applied regarding the diagnostic criteria of osteoarthritis; diagnoses based on physical examination, radiographic and MRI findings, and/or arthroscopy were included.

Intervention

Studies reporting at least one of the major components of yoga, namely, asana (yogic poses), pranayama (breathing practices), and dhyana (meditation) and relaxation practices were included. There were no restrictions on the type, frequency, duration, and delivery mode of the yoga intervention. Studies were excluded if they did not explicitly label the intervention as yoga.

Comparator

Studies comparing yoga interventions with no intervention, sham intervention, non-pharmaceutical intervention (e.g., diet, physical activity, and educational intervention), or pharmaceutical intervention (e.g., NSAIDs) were included. Studies with only a head-to-head comparison of two or more yoga interventions (i.e., different in terms of content, structure, or delivery characteristics) were excluded.

Outcome

We included studies that assessed the core outcomes of osteoarthritis, i.e., pain and/or function, as recommended in several guidelines [34,35,36,37,38]. Pain assessed using any scale (e.g., Visual Analogue Scale (VAS) and Numeric Rating Scale (NRS)) and function assessed using any scale (e.g., Arthritis Impact Measurement Scale (AIMS), including any joint-specific scale such as Foot and Ankle Ability Measure (FAAM), Knee Injury and Osteoarthritis Outcome Score (KOOS), and Hip Disability and Osteoarthritis Outcomes Survey (HOOS)) was eligible [34, 35].

Study design

Only RCTs were included in the review, taking into account the feasibility and practical aspects of the research as well as the hierarchy of study designs.

Data sources and search strategies

The following 13 databases were searched to find published studies from their inception dates to 22 September 2023: (1) MEDLINE (Ovid), (2) EMBASE (Ovid), (3) PsycInfo (Ovid), (4) CINAHL (EBSCOHost), (5) Cochrane Central Register of Controlled Trials (CENTRAL), (6) Allied and Complementary Medicine (AMED) (Ovid), (7) SPORTDiscus (EBSCOhost), (8) Web of Science (Clarivate Analytics), (9) Turning Research Into Practice (TRIP), (10) AYUSH Research Portal (http://ayushportal.nic.in/, accessed 22 September 2023), (11) A Bibliography of Indian Medicine (ABIM) (http://indianmedicine.eldoc.ub.rug.nl/, accessed 22 September 2023), (12) CAM-QUEST (https://www.cam-quest.org/en, accessed 22 September 2023), and (13) Physiotherapy Evidence Database (PeDro). Unpublished studies were searched using (1) OpenGrey (from 1997), (2) EthOS (from 1925), (3) ProQuest Dissertations and Theses (from 1980), and (4) DART-Europe-e-theses portal (from 1999). No language restrictions were applied. The search strategies were developed based on the following and in consultation with a Research Librarian at the University of Nottingham (UK): (1) the yoga component was based on a previous relevant systematic review [39], (2) the osteoarthritis component was based on the search strategies reported in the UK’s National Institute for Health and Care Excellence (NICE) guidelines for osteoarthritis management [40] and existing Cochrane systematic reviews on osteoarthritis [41, 42], and (3) the pre-designed search filters for RCTs were used [43,44,45]. All the search strategies are detailed in the supplementary file (Appendix 1). The reference list of all the included studies and relevant previous systematic reviews was screened for additional studies.

Study screening and selection

Following the searches, all the identified citations were collated and uploaded to Endnote X9 (Clarivate Analytics, PA, USA) [46], and duplicate citations were removed. The remaining records were uploaded into Rayyan (Qatar Computing Research Institute [Data Analytics], Doha, Qatar) [47] and titles and abstracts were screened for eligibility by two independent reviewers (IB and GN). Studies identified as potentially eligible or those without an abstract had their full texts retrieved. The full texts of the studies were assessed for eligibility by two independent reviewers. Full-text studies that did not meet the inclusion criteria were excluded and the reasons for exclusion were reported (Appendix 2). Any disagreements that arose between the two reviewers were resolved through discussion. If a consensus was not reached, a third reviewer was consulted (SL/KC).

Assessment of methodological quality

The included studies were critically assessed using the standardised critical appraisal tool developed by JBI for RCTs, by two independent reviewers (IB and GN), assigning a score as met (Y), not met (N), unclear (U) or not applicable (n/a) [31]. The two reviewers independently assessed each criterion and commented on it. Any disagreements that arose between the two reviewers were resolved through discussion. If a consensus was not reached, then a third reviewer was involved (SL/KC). Regardless of methodological quality, all studies underwent data extraction and synthesis.

Data extraction

Data were extracted from the included studies using a pre-developed and pre-tested data extraction form, by two independent reviewers (IB and GN). If consensus was not reached through discussion, a third reviewer (SL/KC) was consulted. For both the symptoms (pain and function), the authors extracted the end-of-intervention data [34, 48]. Where this time point was not reported, data from the time point closest to the end of the intervention were extracted. Intention-to-treat (ITT) data were preferred compared to per-protocol data [49, 50]. Post-intervention data were extracted in preference to change from baseline data (i.e., post-intervention score − baseline score). Percentage change from baseline was not extracted, as it is highly sensitive to change in variance, and it also fails to protect from baseline imbalances, leading to non-normally distributed outcome data [51].

In the included studies, pain and function were reported as continuous data and so mean and standard deviation (SD) were extracted. Where no SDs were available, they were calculated from standard error (SE) or 95% CI using the formula from the Cochrane Handbook [52]. Where mean and SD for more than one intervention or control group were reported, the combined sample size, mean, and SD were calculated using the formulae in the Cochrane Handbook [52]. The corresponding authors of studies were contacted by e-mail (two times per author) to obtain missing or unclear data.

Data synthesis

Considering the errors in how authors analyse and report yoga interventions to be effective in studies (e.g., conducting pre-post analysis of outcomes within study arms but no comparative analysis between study arms), meta-analyses were conducted for yoga vs. any comparator to determine the true effectiveness of each included yoga intervention for both the outcomes—pain and function. The meta-analyses were conducted using Review Manager 5.4.1 (Copenhagen, The Nordic Cochrane Centre, The Cochrane Collaboration) [53]. Random-effects meta-analyses were conducted due to the heterogeneous nature of the yoga interventions. Since the included studies used difference scales to report pain and function measures, standardised mean differences (SMDs) with 95% CIs were calculated using forest plots.

A narrative synthesis of the identified effective yoga interventions from the meta-analyses for pain and function was conducted with the aid of tables and text, focusing on the content, structure, and delivery characteristics of the yoga interventions. Commonalities and differences of the yoga interventions effective for either or both the outcomes were synthesised. The Sanskrit and English names of all the yogic components used in the effective interventions and the number of RCTs using these practices were tabulated.

Results

Study selection

6693 records were identified through the literature search. After removing duplicate records and title and abstract screening, 44 articles were retrieved for full-text screening. 18 articles were included in this systematic review representing 16 studies (RCTs) and 1402 participants [54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71]. 3 articles described the same RCT, providing data on different outcomes and therefore, were included as a single study in this review [55,56,57]. The study selection process is detailed in the PRISMA flowchart as shown in Fig. 1. No additional articles were identified from citation searching. The list of articles ineligible following the full-text review and ongoing RCTs identified from trial registries are presented in the supplementary file (Appendix 2).

Fig. 1
figure 1

PRISMA flow diagram for included studies from searches of databases and registers only

Full size image

Description of the included studies

9 studies were conducted in the USA [54, 58,59,60,61,62, 66, 67, 70], 4 studies in India [55,56,57, 63, 64, 68], 1 study in Canada [65], 1 study in Iran [69], and 1 study in Australia [71]. The sample size of the studies ranged from 18 to 250. Studies recruited participants aged between 18 and 90 years and where reported, the mean age varied from 57 to 79 years. 13 studies included male and female participants and the remaining 3 recruited only female participants [58, 64, 65]. Where exact numbers were reported, female and male participants were 1049 and 333, respectively. 5 studies included participants with osteoarthritis of one or more lower extremity joints, including knee, hip, ankle or feet [59, 61, 62, 66, 67] and 9 studies included specifically knee osteoarthritis [55,56,57,58, 60, 63,64,65, 68, 69, 71], 1 on hand osteoarthritis [54], and the affected joint was unclear in another study [70]. Where reported, the duration of osteoarthritis amongst participants ranged from more than a month [69] to more than 2 years [55,56,57]. Of 16 studies, 7 reported the use of pain medications among the participants [58, 61, 65,66,67, 70, 71]. In 4 of 7 studies, it was unclear if the participants used pain medications before and/or during the trial [58, 61, 66, 67]. Where reported, 25–63% of participants used pain medications [61, 66]. 3 of 7 studies explicitly reported the use of pain medications among participants before the trial [65, 70, 71], and where specified, this ranged from 61 to 73% [65, 71]. In addition, 3 of 16 studies reported the use of non-pharmacological pain management (using physical therapy) before the trial, amongst 40%, 61%, and 78% of the participants, respectively [61, 70, 71]. Tables 1 and 2 report the characteristics of the studies and the details of the yoga interventions, respectively.

Table 1 Characteristics of included studies
Full size table
Table 2 Intervention details (key features of content, structure and delivery characteristics) of the included RCTs
Full size table

Methodological quality of included studies

Table 3 reports the methodological quality of the included studies. Overall, the methodology was not adequately reported in the included studies, resulting in low methodological quality scores (total “yes” percentage ranging from 15 to 54%). In real practice, yoga providers delivering the yoga intervention cannot be blinded. So, the response to question 5 of the checklist was marked as N/A (not applicable) in our methodological assessment. Some of the major issues in these studies included: (1) inadequate reporting of the randomisation process used to assign participants to study arms and thus, it was unclear if true randomisation was used or not; (2) inadequate reporting of the allocation concealment process and thus, it was unclear if the allocation to study arms was concealed or not; (3) imbalance between the treatment groups at baseline; (4) inadequate reporting of blinding of participants and outcome assessors (could have been achieved through sham therapies); (5) inadequate reporting of whether the study arms were treated identically other than the intervention of interest; (6) insufficient analysis of the differences between study groups about loss to follow up and reasons for loss to follow up in case of incomplete follow up for the entire trial duration; (7) inadequate reporting of ITT analysis and its details (i.e., whether participants were analysed in the groups to which they were initially randomised); (8) inadequate reporting of the measurement process of outcomes (including adverse events) and thus, it was unclear if the outcomes were measured in the same way for study arms or not; (9) no or inadequate description of the number of raters who assessed outcomes or their training, hence making it unclear if outcomes were assessed in a reliable manner; and (10) issues in the statistical power analysis, unclear minimum clinically important difference for the sample size calculation, no information on assumptions of statistical tests used, and errors in statistical analysis and reporting (e.g., pre-post analysis and not between groups).

Table 3 Methodological quality assessment of the included studies
Full size table

Meta-analysis to determine effective studies

Of 16 studies included in this review, 2 could not be included in the meta-analysis because of insufficient data to calculate the mean and SDs for the yoga and control groups [61, 70]. Therefore, a total of 14 studies (16 articles) were included in the meta-analysis to identify the individual effective interventions for each symptom—pain (13 studies) and function (13 studies). These studies compared yoga interventions with pharmacological and non-pharmacological interventions [54,55,56,57,58,59,60, 62,63,64,65,66,67,68,69, 71].

Yoga vs comparator for pain

Overall, yoga interventions reduced pain compared to pharmacological and non-pharmacological interventions (SMD − 0.70; 95% CI − 1.08 to − 0.32) (Fig. 2).

Fig. 2
figure 2

Forest plot for yoga vs comparator (pain). N, sample size; SD, standard deviation; CI, confidence interval

Full size image

Yoga vs comparator for function

Overall, yoga interventions were effective in improving function compared to pharmacological and non-pharmacological interventions (SMD − 0.40; − 0.75 to − 0.04) (Fig. 3).

Fig. 3
figure 3

Forest plot for yoga vs comparator (function). N, sample size; SD, standard deviation; CI, confidence interval

Full size image

Yoga interventions in 9 (out of 13) studies were effective in reducing pain (6 in knees [55,56,57,58, 60, 64, 68, 71], 2 in lower extremities [59, 62], and one in hand [54]. Yoga interventions in 6 (out of 13) studies were effective in improving function (5 in knees [55,56,57, 60, 63, 68, 71]) and one in the lower extremity [62]. Overall, 10 (out of 14) interventions were effective in reducing pain and/or improving function [54,55,56,57,58, 60, 62,63,64, 68, 71].

Content, structure, and delivery characteristics of effective yoga interventions for pain and/or function

Notably, 6 of 10 effective interventions had centre-based (supervised, group) sessions with additional home-based (unsupervised, individual) sessions [55,56,57,58,59,60, 62, 64], 2 interventions were completely centre-based [54, 63] and one home-based [71]. Of the 2 completely centre-based interventions, one was supervised [54] and supervision detail was unclear in the other [63]. One study did not provide any details on intervention delivery [68].

The content of effective yoga interventions was heterogeneous and included 34 different yogic poses (asana; 12 sitting, 10 standing, 8 supine, and 4 prone), 8 breathing practices (pranayama), and 3 meditation (dhyana) and relaxation practices. 8 of 10 effective interventions reported the major components of yoga used; all included asana, and 7 also incorporated pranayama and/or dhyana and relaxation practices. Three of the 10 interventions included all of these 3 major components of yoga [55,56,57, 59, 63], 3 consisted of asana and dhyana and relaxation practices [58, 60, 71], one included asana and pranayama [54], and one included only asana [68]. The majority of these studies were not specific about the exact yoga style, except for 3 which mentioned Hatha yoga [58, 60, 62].

Interventions that included all the 3 major components of yoga reported the time allocated to each component] [55,56,57, 59, 63]. The median time allocated to asana was around 19 min (IQR 10–25 min), pranayama was around 10 min (IQR 3–32 min), and dhyana and relaxation practice was around 12 min (IQR 10–82 min). Some of the common practices were Tadasana (palm tree pose) [55,56,57, 59, 60, 68], Virabhadrasana 1&2 (warrior pose) [59, 60, 68], Nadishuddhi pranayama (alternate nostril breathing) [55,56,57, 63], and Nadanusandhana (A-U-M Kara chanting) [55,56,57, 63]. Table 4 details the various yoga practices (along with their Sanskrit and English names) effective for pain and/or function [72, 73].

Table 4 Yoga practices used in interventions that were effective for pain and/or function
Full size table

The median duration of centre-based sessions was 8 weeks (interquartile range (IQR) 8–12 weeks) and each session was around 53 min (IQR 45–60 min), and these sessions were mostly delivered once a week (reported in 4 of 8 interventions) [54,55,56,57,58, 60]. Where reported (in 4 of 7 interventions) [55,56,57,58, 60, 71], the median duration of home-based sessions was 10 weeks (IQR 8–12 weeks) and each session was around 30 min (IQR 29–30 min), and these sessions were instructed to practice for 4 times a week [58, 60]. One intervention which provided no details on whether it was centre- or home-based, reported session duration and frequency i.e., 30 min thrice a week for 4 weeks [68].

3 of 8 centre-based sessions specifically reported that lectures and counselling on yoga for osteoarthritis, educational materials with written instructions, peer support through group discussion and question-and-answer sessions were used to deliver the intervention [54,55,56,57, 63]. To address participants’ needs (e.g., physical limitations), certain yogic poses were adapted in 4 of 10 interventions e.g., by using blocks, straps, blankets, and chairs [58,59,60, 62]. Where reported (6 of 7 interventions), home-based sessions were delivered by providing written documents with pictures [58,59,60, 62, 64] and yoga videos (including demonstrations of modified yoga postures to meet individual needs) [71] to the participants.

Strategies were used to monitor and improve adherence to yoga practice at home in 6 of 7 interventions e.g., sharing self-recorded yoga diaries/log sheets, self-recorded yoga videos (documenting home practice details) with the trial team, and reminding and motivating participants e.g., through telephone calls and emails [55,56,57,58,59,60, 62, 71]. Participants were encouraged to practice yoga in the long term in 2 of 10 interventions [62, 71].

Discussion

Our review found beneficial effects of yoga on two major symptoms of osteoarthritis i.e., pain and function, as also reported in previous reviews and meta-analyses [28, 29]. However, this review is novel as it aims to identify and synthesise the key features (content, structure, and delivery characteristics) of effective yoga interventions. Although the components of effective yoga interventions were heterogeneous, some commonalities were identified. A majority of them involved participants attending yoga at a centre for supervised group sessions, once a week for a sustained period. Most also involved individual yoga practice at home, which was unsupervised, several times a week between centre-based sessions. These interventions generally incorporated yoga postures (asana), but a majority also included other major components including breathing practices (pranayama) and/or meditation (dhyana) and relaxation practices. Keeping in mind the participants’ needs (e.g., knee/hip pain), various modifications were provided to certain yogic poses using props (e.g., a chair as a support for standing poses). The review also indicated ways in which these interventions ensured maximum adherence to yoga among participants, especially to yoga practice at home, through regular contact with the participant or journaling of home practice.

Yoga has been recommended by an international osteoarthritis evidence-based clinical guideline developed by the American College of Rheumatology (ACR) and the Arthritis Foundation (AF) [10]. This guideline used the GRADE methodology to rate the quality of the available evidence and to develop the recommendations [10]. The guideline was developed by an expert panel representing the ACR, including rheumatologists, an internist, physical and occupational therapists, and osteoarthritis patients. The guideline “conditionally” recommends the use of yoga for managing symptoms of knee osteoarthritis. The conditional recommendation inferred that yoga could be used by the patients to manage knee osteoarthritis only after shared decision-making, including a detailed explanation of the benefits and harms of yoga to the patients, in a language and context they understand. However, due to a lack of evidence, no recommendations could be made for managing hand and hip osteoarthritis through yoga [10]. It is important to note that this guideline was based on a generic definition of yoga laid out by the National Center for Complementary and Integrative Health [NCCIH] as “a mind–body practice consisting of physical postures, breathing techniques, and meditation or relaxation”. Yoga includes a diverse range of components, so it is important to consider the aspects of yoga that are effective.

To the best of our knowledge, this is the first systematic review conducted by following a robust methodology to synthesise the content, structure, and delivery characteristics of effective yoga interventions for managing major osteoarthritis symptoms. Determination of the effectiveness of interventions was standardised across studies by using meta-analysis before finally the detailed characteristics of potentially effective yoga interventions were synthesised. Although we used a comprehensive search, only 16 studies met our inclusion criteria, 2 of which could not be included in the meta-analysis.

Our review highlighted some limitations in the included studies. Some studies had incomplete or unclear information on the time allocated to each component of the yoga intervention. Future RCTs should improve the reporting of this detail which would help in establishing the aspects of yoga practice that are most beneficial for improving osteoarthritis symptoms and therefore where the emphasis for future yoga-based interventions should lie. Most studies did not describe the qualifications of the yoga providers, instead only described the yoga providers as certified, experienced, or trained. Where yoga provider qualifications were reported in detail, there was no description of the yoga providers' training to deliver the specific intervention protocol. The expertise of yoga providers and their training in the intervention protocol are likely to be essential to the effectiveness of the intervention and in ensuring the fidelity and safety of the intervention [74, 75]. Thus, future RCTs should ensure adequate reporting of the training details of the yoga providers. In terms of safety, a majority of the studies did not explicitly report adverse events related to yoga. This issue should be addressed and reporting should be improved in future RCTs, to indicate their safety. Most included studies were based in the US, predominantly included females, and reported on knee osteoarthritis. These factors potentially limit the generalisability of our findings. Existing evidence has shown a higher uptake of yoga among women as compared to men [11]. Although osteoarthritis is more common in women, it also affects men. So, it is important to explore the possible barriers to yoga practice among men (e.g., gender-based perceptions and their preference for other forms of physical activity) and take some initiatives (e.g., men-only yoga classes) to promote yoga among men for osteoarthritis, in future RCTs [76]. Lastly, most included studies involved relatively short-term follow-up, and only 2 of our included studies reported on encouraging participants to continue yoga practice. For a chronic condition like osteoarthritis, it will be important to explore whether yoga practice and its effects are sustained in the longer term.

Despite the low quality and heterogeneity of included studies, our findings suggest yoga interventions might be effective in managing osteoarthritis symptoms and highlight the key characteristics of effective yoga interventions for osteoarthritis. Supervised yoga practice in groups at a centre seemed to be one of the most consistent characteristics in the studies included in this review. This might reflect the importance of creating a community for yoga practice, given that most people with osteoarthritis are older, and may be socially isolated [77]. Older participants generally tend to be interested and adherent to yoga classes (at a centre) [78], which may also be why centre-based yoga sessions are an important feature of yoga practice. Adherence of participants to the yoga interventions is likely to play an important role in its effectiveness on osteoarthritis symptoms [79]. Most studies included in this review mentioned strategies such as self-reporting of home practice and ensuring regular contact with the participants to encourage them to adhere to yoga practice at home, though the level of adherence was generally not reported. To improve adherence to mind–body interventions such as yoga, some strategies including providing reminders, following up with the individual to ensure regular yoga practice, and catering to individual needs (e.g., modifying yogic poses), have been identified [80]. Further, it will be important to explore how yoga interventions might be implemented and integrated within the orthopaedic healthcare system [81]. Hence, future research should aim to develop a yoga program for managing osteoarthritis by using the synthesised findings from this review. This is likely to require the input of a variety of stakeholders, including osteoarthritis patients, yoga providers, and orthopaedic doctors to address the potential challenges and to consider the evidence base for its effective and safe incorporation, by using Delphi or similar methods.

Conclusion

Considering the methodological limitations of previous studies, including low quality of studies and heterogeneity between studies, a high-quality long-term RCT should be conducted to determine the effectiveness of yoga in managing osteoarthritis symptoms by using the synthesised key characteristics of previous effective yoga interventions, as identified by this review.