Effectiveness and cost-effectiveness of a Yoga-based Cardiac Rehabilitation (Yoga-CaRe) program following acute myocardial infarction: Study rationale and design of a multi-center randomized controlled trial.

BACKGROUND
Cardiac rehabilitation (CR) is a standard treatment for secondary prevention of acute myocardial infarction (AMI) in high income countries (HICs), but it is inaccessible to most patients in India due to high costs and skills required for multidisciplinary CR teams. We developed a low-cost and scalable CR program based on culturally-acceptable practice of yoga (Yoga-CaRe). In this paper, we report the rationale and design for evaluation of its effectiveness and cost-effectiveness.


METHODS
This is a multi-center, single-blind, two-arm parallel-group randomized controlled trial across 22 cardiac care hospitals in India. Four thousand patients aged 18-80 years with AMI will be recruited and randomized 1:1 to receive Yoga-CaRe program (13 sessions supervised by an instructor and encouragement to self-practice daily) or enhanced standard care (3 sessions of health education) delivered over a period of three months. Participants will be followed 3-monthly till the end of the trial. The co-primary outcomes are a) time to occurrence of first cardiovascular event (composite of all-cause mortality, non-fatal myocardial infarction, non-fatal stroke and emergency cardiovascular hospitalization), and b) quality of life (Euro-QoL-5L) at 12 weeks. Secondary outcomes include need for revascularization procedures, return to pre-infarct activities, tobacco cessation, medication adherence, and cost-effectiveness of the intervention.


CONCLUSION
This trial will alone contribute >20% participants to existing meta-analyses of randomized trials of CR worldwide. If Yoga-CaRe is found to be effective, it has the potential to save millions of lives and transform care of AMI patients in India and other low and middle income country settings.


Background
Coronary artery disease (CAD) is the leading cause of death and disability in India [1,2]. The psychosocial and economic aspects of CAD are particularly important, as CAD typically manifests at younger ages (<55 years) in previously healthy adults.
These adults experience considerable difficulty in accepting their illness, resulting in reduced enthusiasm to engage with the family and return to economically productive activity [3]. Hence, there is a compelling need for effective secondary prevention interventions, aimed at improving both the duration and the quality of life of those affected.
Cardiac rehabilitation (CR) is the process of restoring desirable levels of physical, social, and psychological functioning after the onset of CAD or after interventions for CAD. It is typically achieved through a program of exercise and information sessions.
CR programs for patients with CAD have been shown to reduce mortality and hospital admissions, improve quality of life, and reduce the costs associated with care [ [4], [5], [6]]. However, CR programs are unavailable to most patients in India due to high costs and skills required for multidisciplinary CR teams [7]. The WHO Global action plan for the prevention of non-communicable diseases (NCDs) has advocated for harnessing the potential of traditional and complementary therapies because of their potentially lower costs and greater cultural acceptability [8]. Although alternate models of CR based on complementary and alternative therapies have been proposed, their clinical effectiveness and safety is yet to be established [9,10]. Low-cost and sustainable CR programs based on local traditional practices could address the unmet need of CR globally.
Yoga is an ancient Indian system of mind-body discipline, encompassing an array of philosophical precepts, mental attitudes and physical practices [11]. Although it is practiced in many forms, the most common ones incorporate elements of physical poses, breathing practices and meditation, and moderation in lifestyle including diet, stress and sleep and abstinence from smoking and alcohol. As such, it covers the core elements of a CR program: improved physical conditioning, stress reduction and lifestyle moderation. The structure of a yoga training program is also similar to CR (a series of exercise-cum-education sessions), but requires considerably fewer resources (i.e. a basic instructor). Yoga, therefore, could provide a useful framework on which to develop an economical CR program for India.
Strong rationale exists for the use of yoga in the prevention of cardiovascular diseases. Studies suggest yoga down-regulates effects of stress (primarily on blood pressure) mediated by hypothalamic-pituitary-adrenal axis (e.g. cortisol), specifically the renin-angiotensin system, and the sympathetic nervous system (e.g. improvement in baroreceptor sensitivity) [ [12], [13], [14]]. Variable improvements in functional capacity (e.g. ventricular structure/function), inflammatory markers, carotid intima-media thickness and other conventional cardiovascular risk factors (e.g. insulin, lipids) are also reported [15]. Also, yoga has been shown to improve health related quality of life in chronic diseases [16], [17], [18], [19].
However, these studies have been limited by small sample sizes, lack of control groups and unmasked assessment of outcomes [20], [21], [22]. We therefore hypothesized that yoga could be a useful cardiac rehabilitation tool. We propose to evaluate this through a randomized control trial.

Study design and recruitment strategy
This is a multi-center, single-blind, two-arm parallel group, randomized controlled trial (RCT) conducted across 24 cardiac centers in India. The study sites include a mix of public funded, teaching hospitals and private hospitals to allow recruitment of patients from a wide range of socio-economic backgrounds. As is typical of India, most (

Randomization
Eligible and consenting patients are randomized (block randomization, stratified by age-band (< or ≥60 years) and gender) in a 1:1 ratio by a computer program centrally (Research Coordinating Center (RCC), New Delhi) using an interactive web response system (IWRS) and after collecting key identifying data.
We chose individual randomization (within centers) over cluster randomization (with centers as clusters) because we considered potential bias from differences in clinical practice between centers to be a more serious threat to study validity than contamination, particularly because Yoga-CaRe as a CR program would not be available externally, even if yoga classes could be.

Yoga-CaRe program
The Yoga-CaRe program was developed using a structured process. This included a literature review and qualitative and quantitative interviews with yoga experts, CR experts and patients with myocardial infarction to systematically identify and shortlist appropriate yoga exercises and postures, breathing exercises, meditation and relaxation practices, and lifestyle changes, which were incorporated into a conventional CR framework. The details of the CR program are under review for publication elsewhere. Briefly, it involves 13 hospital-based sessions spread over 3 months (Table 1) This also includes didactic lectures on anatomy and physiology of the cardiovascular system, communicating with patients, barriers and facilitators to cardiac rehabilitation program and importantly identifying warning signs or distress symptoms.
In addition, manual for instructors was provided to the instructors in their local language. We are video-recording of 10 sessions in each of the participating centers and around 80 sessions are video-recorded; these are supplemented by direct observations of the sessions by the study team. Data on adherence to homepractices are collected by questionnaires completed during each follow-up session. Adherence to prescribed medication is assessed by Morisky Medication Adherence Scale (MMAS). MMAS has a high sensitivity and has been used in Indian settings in the past [28].
Data on cognitive function is being collected using the Montreal Cognitive Assessment (MoCA), and anxiety and depression using Hospital Anxiety and Depression Scale (HADS), in a subset of participants (n = 400). MoCA and HADS are validated and widely accepted tools for assessing cognitive assessment and anxiety respectively [29,30].
Every participant will be followed for a minimum of six months till end of the trial. The end of study differs for each participant depending on the time they were enrolled into the study. The site coordinators upload the data collected from participants through electronic case report forms to the central database located in the Research Coordinating Center (RCC). The database has a limited access password protected system and constructed with built-in range checks to avoid errors in data. Also, the data are cross-checked against the source documents during regular on-site monitoring visits and regular central statistical monitoring to detect outliers and data discrepancies. At the clinical site, collected data is protected in fire-proof and locked safes. All the participant data will be de-identified in the final dataset and will be made available for bonafide researchers upon request to the investigators ( Table 2).

Sample size estimation
The required sample size was estimated using incidence data from a multinational registry that includes India (cumulative annual incidence rate of ~20-25% for a composite outcome of all-cause mortality, nonfatal MI and stroke, following AMI) [31] and effectiveness data from systematic reviews of CR trials (a 20-25% reduction in adverse vascular outcomes). The total sample required was initially estimated as 3102, assuming 20% annual incidence in the control arm and a 20% effect size at a median follow-up of 12 months, a 10% loss to follow-up, at 80% power and 5% significance level. Estimation was based on time-to-first-event data (censoring the patient subsequently). Midway through the study, the sample size was re-estimated to account for a lower observed incidence of outcomes, which is consistent with a rapid improvement in cardiac care in India over the last decade. We therefore proposed to recruit 5000 patients; however, the Data Monitoring and Ethics committee recommended stopping recruitment at 4000 and extending the study follow-up after reporting the results until this point (including the minimum 6 months follow-up).
A considerably smaller sample is required to detect differences in quality of life; a total sample of 818 (409 in each group) would be adequate to detect a 5% difference in the mean EQ-5D scores, with 95% confidence. analysis for the quality of life will be assessed by estimating the difference between the treatment arms in the change score from baseline to 12 weeks in health thermometer by fitting conventional regression models. Also, EQ-5D-5L index as a whole and its five dimensions will be analyzed as scaled outcomes and differences between treatment arms will be examined by fitting regression models.
All primary analyses will be on an intention to treat principle and will be unadjusted, and subsequently adjusted for covariates with baseline imbalances including center.
We will also report the per-protocol or on treatment analyses. For missing data, sensitivity analyses will be conducted for primary end points, including a complete case analysis and multiple imputations to evaluate the potential effect of missingness.

Economic analyses
We will calculate health expenditures and within-trial cost-effectiveness of the intervention compared to enhanced standard care using a health care system and patient perspective as our base-case analysis, and a societal perspective in sensitivity analyses.
Health expenditures will include direct medical costs and implementation/delivery costs of the intervention (i.e. costs attributable to research activities will be excluded) from our clinic and study expenditures. We will estimate the incremental costs of intervention (outpatient visits, medications, in-patient hospitalizations, and indirect medical costs) compared to enhanced standard care group. Given that public and private clinics are participating in our trial, we will also report expenditures for interventions relative to enhanced standard care by type of clinic and cumulatively.
The primary outcome for cost-effectiveness will be major adverse cardiovascular events (MACE) averted and quality adjusted life years (QALYs) gained. Both MACE and QALY data are being collected over the trial duration. QALYs can be compared across interventions and countries (using a ceiling ratio [set cost per QALY value as a reference point]). Uncertainty around incremental costs and incremental outcomes (MACE averted and QALYs gained) will be estimated using nonparametric bootstrap methods.
A probabilistic decision analytic health state-transition model (Markov model) will be constructed to extrapolate the trial effects over patients' lifetime horizon. A hypothetical cohort of acute myocardial infarction patients with a given set of characteristics (e.g. age, comorbidities) will be followed until death over successive time cycles of set length (1 year). Where available, appropriate prognostic models will be used to predict the time-dependent risk of major adverse cardiovascular events. The incurred costs and health outcomes over the model time horizon will be summed and averaged for intervention and enhanced standard care groups and the long-term incremental cost-effectiveness of intervention will be calculated. An annual discount rate of 3% will be applied to both costs and outcomes, which will be varied in sensitivity analyses. A range of scenario analyses will be conducted to investigate the influence of alternative assumptions on the base case results.

Discussion
CR programs are an integral part of cardiovascular management globally [32,33].
They effectively reduce mortality (up to half) and morbidity, improve quality of life, lower medical and social costs, and increase economic productivity, comparing favorably to effective pharmacological treatments for secondary prevention of AMI (e.g. statins and beta blockers) with minimal side effects. Yet, their uptake remains low. In low and middle income countries (LMICs) such as India, the cost and skills required for large multidisciplinary teams used to deliver established CR programs is prohibitive, making them un-scalable. Even in HICs where CR programs are universally available, uptake is often low, particularly among disadvantaged groups such as women, the elderly and minority ethnic groups, who find the gymnasium style exercise programs unappealing due to unfamiliarity and cultural and attitudinal reasons. Hence there is an urgent need to develop other forms of CR programs which are affordable and culturally acceptable (in LMICs) and offer a choice (in HICs) [7,34]. Yoga-CaRe will address this unmet need if it is found to improve outcomes and is cost-effective, and may also instigate development of CR programs based on other cultural practices, such as the Chinese Tai Chi.

Strengths and limitations
Previous trials of CR in post MI patients have been relatively small randomizing up to a few hundred participants with exception of the WHO (1983) and West (2012) that randomized 3184 participants and 1813 participants respectively [6]. Also, most data are from the developed countries. Our trial will be considerably larger and will contribute to >20% of participants in meta-analyses of randomized trials of CR. Also, it is independently powered for policy relevant outcomes of clinical effectiveness. It is embedded within the typical healthcare system (mix of public and private) making findings generalizable, and has inbuilt economic data collection to undertake robust economic analyses to guide actions by policy makers and private healthcare providers. The intervention was designed to be extremely low cost and scalable, ensuring immediate impact on policy and clinical practice. Conversely, the pragmatic design of the intervention (only a few supervised sessions) also precludes estimation of true efficacy of yoga for CR, increasing the likelihood of a false negative result.
We lacked the resources to explore the role of specific mechanistic pathways that have been proposed for cardiovascular health effects of yoga. However, we have recently completed a separate parallel study on that question in the UK, which will be reporting its results soon.

Trial status
The

Conflict of interest
The authors report no relationships that could be construed as a conflict of interest.