Yoga for Adults with Type 2 Diabetes: A Systematic Review of Controlled Trials

A growing body of evidence suggests yogic practices may benefit adults with type 2 diabetes (DM2). In this systematic review, we evaluate available evidence from prospective controlled trials regarding the effects of yoga-based programs on specific health outcomes pertinent to DM2 management. To identify qualifying studies, we searched nine databases and scanned bibliographies of relevant review papers and all identified articles. Controlled trials that did not target adults with diabetes, included only adults with type 1 diabetes, were under two-week duration, or did not include quantitative outcome data were excluded. Study quality was evaluated using the PEDro scale. Thirty-three papers reporting findings from 25 controlled trials (13 nonrandomized, 12 randomized) met our inclusion criteria (N = 2170 participants). Collectively, findings suggest that yogic practices may promote significant improvements in several indices of importance in DM2 management, including glycemic control, lipid levels, and body composition. More limited data suggest that yoga may also lower oxidative stress and blood pressure; enhance pulmonary and autonomic function, mood, sleep, and quality of life; and reduce medication use in adults with DM2. However, given the methodological limitations of existing studies, additional high-quality investigations are required to confirm and further elucidate the potential benefits of yoga programs in populations with DM2.


Introduction
Type 2 diabetes (DM2) has become a leading public health issue globally, with estimated 366 million people affected in 2011 [1]. This figure represents a more than twofold rise in the last three decades and parallels the growing pandemic of obesity and the increasingly widespread adoption of Western lifestyles [1,2]. Worldwide prevalence is expected to continue rising in both industrialized and developing countries [2,3], with numbers projected to reach 552 million adults by 2030 [1,4]. DM2 is now a leading cause of death and disability and significantly increases risk for both macrovascular complications, such as atherosclerosis, and microvascular complications, such as retinitis, diabetic neuropathy, and renal disease [5]. DM2 is also strongly associated with elevated risk for other serious chronic conditions, including depression and dementia [3,6]. Cardiovascular disease (CVD) is the primary cause of morbidity and mortality in those with DM2 [7], accounting for at least 65% of deaths in this population [8]. In addition, the social and economic burden of DM2 is substantial and growing rapidly. For example, DM2 accounts for at least 10 percent of all healthcare expenses in the United States, making DM2 the single most costly chronic disease [9].
DM2 is typified by hyperglycemia in the presence of insulin resistance [6]. Other key related hemodynamic and metabolic abnormalities characterizing DM2 include elevated blood pressure, dyslipidemia, and chronic inflammation, as well as hypercoagulation and increased oxidative stress [5,[10][11][12]. Risk for DM2 rises with increasing age and is further elevated in certain racial and ethnic groups, including non-Hispanic blacks, Asians, Native Americans, and Pacific Islanders [13,14]. However, while race, age, genetic predisposition, and other nonmodifiable factors are important in the pathogenesis of DM2, lifestyle factors, particularly, 2 Journal of Diabetes Research physical inactivity, overnutrition, and related obesity, are thought to be primarily responsible for the current global diabetes epidemic [1,15]. Other contributing lifestyle-related factors include chronic stress, impaired sleep, and smoking [2,10,16,17]. In fact, lifestyle factors may account for 90% of incident diabetes cases [2] and are significant predictors of DM2-related complications and mortality [18]. Thus, a central element of DM2 care is lifestyle management, which is considered critical to the prevention of acute complications and the reduction of risk for long term complications [6].
A central goal in DM2 management is the reduction of blood glucose levels, which has been shown to reduce risk of microvascular and possibly certain other complications [3,5,6]. However, while improving glycemic control remains a priority, the critical importance of multifactorial DM2 management has been increasingly emphasized in recent years, reflecting the complex constellation of factors that underlie the development of DM2 and its sequelae [3,6]. In particular, reducing CVD risk factors is crucial to effective DM management [6]. Additional goals include reducing DM-related distress, alleviating depression, and enhancing emotional well-being and quality of life, factors that are important predictors of glycemic control, complication rates, treatment adherence, and other outcomes [6,[19][20][21][22][23][24].
In light of the above, identifying sustainable lifestyle interventions with the potential to improve multiple factors of relevance to the management of this complex illness is of clear importance. Mind-body practices such as yoga, which capitalize on the ability of the mind to enhance physical health (and vice versa), appear particularly suited for addressing multifactorial conditions. Yoga is a traditional mind-body system originating in India over 4000 years ago [25]. In recent decades, the practice of yoga has been rising in both developed and developing countries worldwide [26][27][28][29], and the field of yoga therapy is now growing rapidly [30]. The goals of yoga, a Sanskrit term meaning "yoke or union," do not center primarily on physical fitness, but rather on integration of mind, body, and spirit, cultivation of balance, calm, harmony, and awareness, and, in classic yoga traditions, the attainment of selflessness and spiritual enlightenment [25,26]. Of the several major branches of yoga, the most widely practiced forms include Raja (royal or classical) yoga and the closely related Hatha yoga, sometimes known as the yoga of activity [25,26]. Mantra yoga, emphasizing the use of specific sounds or chants to achieve mental and spiritual transformation, was popularized in the West by Maharishi Mahesh Yogi, the founder of Transcendental Meditation (TM). Hatha and Raja yoga emphasize specific postures (asanas), including both active and relaxation or restorative poses, as well as breath control (pranayama), concentration (dharana), and meditation (dhyana), with some schools focusing primarily on restorative/meditative practices. Hatha yoga, the branch of yoga most widely practiced in the US and other Western countries, also incorporates cleansing exercises, mantras or chants, and specific hand gestures (mudras). Originally developed to prepare the body for meditation, Hatha yoga itself encompasses many different styles, including Iyengar, Ashtanga, Integral, Kundalini, Viniyoga, Vinyasa, Kripalu, and Bikram yoga [25,26].
A growing body of evidence suggests yoga practice may reduce risk for CVD and lead to improvements in physical health and well-being in a range of populations [31][32][33][34][35][36][37], potentially including those with DM2. In this systematic review, we critically evaluate available evidence from controlled trials regarding the effects of yoga-based programs on health-related outcomes in adults with DM2. We also outline major limitations in the current literature, briefly discuss possible mechanisms that may underlie observed benefits, and suggest directions for future research.

Methods
We searched eight databases via EbscoHost (Academic Search Complete, Alt HealthWatch, CINAHL with Full Text, Health Source: Nursing/Academic Edition, MEDLINE, Psy-cARTICLES, PsycINFO, and SPORTDiscus with Full Text) from their inceptions through March 2015 for original, scholarly articles reporting on controlled trials of yoga in the management of diabetes. The basic search strategy, (yoga OR yogic) AND diabetes, yielded 468 results; the same search conducted in the Cochrane Central Register of Controlled Trials resulted in 37 trials. A secondary search of the Physiotherapy Evidence Database (PEDro) for the terms yog * AND diabetes yielded 34 articles. Titles and abstracts of the citations were scanned to identify potential articles for inclusion in this review. Potentially eligible papers were retrieved for more detailed review. In addition, we manually searched our own files, the citation sections of all identified articles, and the reference sections of recent review articles regarding diabetes and yoga.
Original, controlled studies were included if they evaluated the effects of yoga and yoga-based interventions on measures of glycemic control and insulin resistance, lipid profiles, body weight or composition, blood pressure, or other relevant outcomes, such as oxidative stress, nervous system function, cardiopulmonary function, mood and sleep impairment, or medication use. We excluded uncontrolled trials, cross-sectional studies, case series, and case studies, as well as trials that were published only in dissertation or abstract form. We also excluded articles that did not specifically target adults with diabetes, involve an intervention focused on yoga as the major component, study a yoga program of at least two week's duration, or report quantitative outcome data. Studies targeting populations with type 1 diabetes were also excluded, as the etiology and management for type 1 diabetes differ from those for DM2.
We categorized clinical measures and outcomes evaluated into several domains. For each domain, we summarized findings from relevant studies. Study quality was evaluated using the PEDro scale, which scores papers based on the following 10 criteria: (1) subjects were randomly allocated to groups, (2) allocation was concealed, (3) the groups were similar at baseline regarding the most important prognostic indicators, (4) there was blinding of all subjects, (5) there was blinding of all therapists who administered the therapy, (6) there was blinding of all assessors who measured at least one key outcome, (7) measures of at least one key outcome were obtained from more than 85% of the subjects Journal of Diabetes Research 3 initially allocated to groups, (8) all subjects for whom outcome measures were available received the treatment or control condition as allocated or, where this was not the case, data for at least one key outcome was analyzed by "intention to treat," (9) the results of between group statistical comparisons are reported for at least one key outcome, and (10) the study provides both point measures and measures of variability for at least one key outcome. One point is awarded for each criterion reported in the paper. In addition, PEDro assesses whether study eligibility criteria were specified; however, as this does not address statistical or internal validity, no points are awarded for this criterion. Whenever available, we reported the score given in the PEDro database; when the study was not in PEDro, we assigned a score based on the PEDro instructional materials (http://www.pedro.org.au/wp-content/uploads/PEDro scale .pdf). In our discussion of findings, we also considered recent meta-analyses of RCTs targeting populations with diabetes.
Each eligible study identified was classified into one of two design categories: randomized controlled trials (RCTs) or nonrandomized controlled trials (NRCTs). In RCTs, each participant is randomly assigned to one of at least two groups: an active intervention (i.e., yoga) group and one or more comparator groups. In NRCTs, the experimental design and analytic strategy are similar to that of RCT's, but the allocation to a given treatment is not performed randomly. Articles were selected and data extracted and evaluated by both authors; conflicts were resolved by discussion and consensus.
Of the 25 studies, 24 (96%) specified adults with DM2; in the one remaining investigation, participants were recruited from a diabetes clinic, but authors did not specify type of diabetes (Table 1). Age range was wide in most studies, with some specifying no upper age limit [57,[60][61][62]65]. Of the 23 studies that specified age cut-offs ( While several studies included adults over the age of 65 [61,62,[66][67][68][69][70], only one study specifically targeted elder adults [61,62]. All but two studies [51,52,60] included participants of both genders; likewise, only two studies clearly specified exclusion of patients on anti-DM medications [38,43]. Patients ranged from newly or recently diagnosed DM2 to those diagnosed with DM2 for 10 years or more, often within the same study, with stated exclusion and inclusion criteria varying from study to study. Of the 17 studies with sufficient detail to allow determination, all but 2 [45,65] excluded patients with significant complications (Table 1), although, again, exclusion conditions were not uniform across studies.

Effects of Yoga on Metabolic, Anthropometric, and Hemodynamic Indices.
Twenty-five eligible studies assessed the potential impact of yoga-based programs on one or more anthropometric or physiologic markers of importance to diabetes management and prognosis, including measures of glucose tolerance and insulin resistance, lipid profiles, body weight and composition, and blood pressure. Findings of these studies are summarized in Table 2 and are discussed briefly below.
Of the 12 RCTs evaluating the effects of yogic practices on indices of glycemia and insulin resistance ( = 1040 12.0% * Including two yoga-based interventions tested within the same study [38]. ¥ Practice frequency not specified in 2 NRCTs [39,40]. ¥¥ Information on program structure lacking in two NRCT's [39,41]. * * Numbers add up to more than 25, as 3 studies included more than one comparator. participants), 10 reported significant reductions in at least one measure (Tables 2 and 3). Reported improvements again included significant declines in PPBG [53,55,63,64], FBG [51, 59-61, 63, 64, 66, 70], insulin [51], and HbA1c [61,66,70] in those assigned to a yoga-based program versus standard care [51,55,58,63,64,66,67,70], group education [61], or brisk walking [60]. Two additional trials observed significant beneficial changes in PPBG, FBG, and HbA1c following a comprehensive yoga program [57,66] that were similar to those detected in participants assigned to a relatively intensive exercise intervention [57,66]. In contrast, two RCTs documented no significant changes in these parameters. In a small 12-month Cuban trial of 40 adults with uncomplicated DM2, yoga group participants showed modest declines in FBG (7%) that did not differ significantly from those of the exercise group (1.7%) [69]. Similarly, a British study of 59 participants with DM2 showed no significant improvements either in glucose control [65], findings that may be in part attributable to low rates of participant compliance (50% class attendance and 0% performance of home practice).
Collectively, 92% of the controlled trials reviewed, including 12/12 NRCTs and 10/12 RCTs, reported improvements in glucose control with yoga-based programs that were statistically and clinically significant, suggesting that yogic practices may improve glucose control in adults with DM2. However, between group comparisons were not reported in ten studies (7 NRCTs [38,40,43,44,(46)(47)(48), 3 RCTs [60, 63, 64]), and, as discussed below, additional methodologic or other limitations plagued most trials, hindering interpretation of findings. Magnitude of effects varied substantially among studies, from relatively modest to very large improvements depending on the study design, population, and comparator group (Table 3).
Overall, 94% of the studies reviewed reported improvements in lipid levels following completion of yogic programs of varying intensity and duration. Again, reported changes were clinically as well as statistically significant, with magnitude of effects varying substantially with the study and target population (Table 3). However, again, interpretation of findings is hampered by the lack of between group comparisons in      Waist-hip circumference ratio 5.38-8.99% [44,47] 3.17% [49] 5.32% [58] Blood pressure Systolic blood pressure 11.27% [47] 5.03% [49] Diastolic blood pressure 12.92% [47] 3.46% [49]   several studies, as well as other design, methodologic, and/or reporting limitations characterizing many of the studies, discussed in more detail below.

Body Weight and Composition.
Nine controlled trials have evaluated the potential influence of yoga on body weight and composition in adults with DM2, including 6 NRCTs and 3 RCTs; again, all but one study [65] (89%) documented improvements in those assigned to a yoga intervention. All of the yoga programs included active asanas. Trials ranged from 40 days [47] to 6 months [66] in duration, and, in all but the one trial reporting no significant benefits [65], participants practiced at least 3 times per week. Of the 6 NRCTs ( = 613 total participants), all reported significant beneficial changes in at least one measure of body composition, including reductions in weight [38,44,48], body mass index (BMI) [44,45,48,49], and waist-hip ratio [44,47,49] relative to standard care [38,44,45,48,49] or standard care and light exercise [47]. Likewise, 2 of the 3 RCTs ( = 390 total participants), including a 6-month Cuban study in 231 adults [66] and a 3-month Indian trial of 100 adults [58], reported significant decreases in weight [58], BMI [66], and waist-hip ratio [58] in participants assigned to a yoga program versus standard care; yoga group improvements were again comparable to those documented following a moderate intensity physical activity program [66]. In contrast, a UK study of 59 adults showed no significant reduction in weight or BMI relative to standard care [65], although poor participant adherence renders interpretation of findings problematic.

Blood Pressure.
Only 5 of the 25 controlled trials targeting adults with DM2 evaluated the potential effects of yogic practices on blood pressure, including 3 NRCTs and 2 RCTs ( = total of 588 participants). Trials varied in length from 40 days to 3 months, and all included active asanas. Three of the five studies, two NRCTs and one RCT [59], in Indian adults showed significant drops in systolic and diastolic blood pressure relative to standard care [49,59] or standard care combined with walking [47]. Another Indian study of 123 adults [45] showed similar, but not statistically significant, declines in blood pressure in participants assigned to the yoga versus the standard care group. One exploratory RCT of British adults [65] did not present any quantitative data on blood pressure but reported no significant changes following a 12-week versus standard care program.

Observed Effects of Yoga on Other Pertinent Indices.
There is mounting evidence, reviewed briefly below, that yoga may improve other risk indices pertinent to the DM2 management as well, including oxidative stress, impairments in mood and sleep, nervous system and pulmonary function, and medication usage. Oxidative stress, an imbalance between free radicals and antioxidants, has been strongly implicated in the development of DM2 and diabetes-related complications [74,75]. For example, the oxidative imbalance that often characterizes DM2 can lead to prothrombotic changes, endothelial dysfunction, and chronic vascular inflammation, exacerbate insulin resistance and associated hyperglycemia, and ultimately promote organ damage; these reciprocally related changes are thought to mediate many of the atherosclerotic and thrombotic alterations associated with the metabolic syndrome and to play a central role in the pathogenesis and progression of diabetes and CVD [10,[75][76][77][78]. Thus, interventions that reduce oxidative stress in those with diabetes may aid in improving risk profiles and decreasing diabetes-related complications. To date, at least five controlled trials have examined the effects of yoga on measures of oxidative stress in adults with DM2. Yoga interventions all incorporated active asanas and ranged from 40 days to 6 months in duration; all indicated beneficial changes with yoga ( Table 2). For example, four NRCTs in Indian adults documented improvements in indices of oxidative balance; these included significant declines in MDA [40,45,47] among participants completing a 40-day [47] to 4-month yoga program [40] versus those assigned to standard care [40,45] or standard care with light exercise [47]; in their 3-month study of 123 adults, Hegde et al. also reported significantly greater increases in serum levels of the antioxidants glutathione and Vitamin C in the yoga versus standard care group [45]; likewise, Mahapure et al. noted significantly greater increases in superoxide dismutase (SOD) levels in the yoga group relative to the standard care control in their 6-week study [41]. Additionally, in an RCT of 231 Cuban adults, those assigned to a 24-week yoga program showed increases in activity of the antioxidant, SOD, and reductions in malondialdehyde (MDA), an estimate of lipid oxidative damage, that were comparable to those completing a 24-week conventional exercise program, and greater than participants assigned to standard care [66].

Nervous System
Function. Autonomic nervous system dysfunction, a correlate of obesity and poor cardiorespiratory fitness, has been strongly and bidirectionally related to insulin resistance and hypertension, implicated in the development of diabetes [120,121] and CVD [122], and associated with increased risk for morbidity and mortality in individuals with DM2 [123,124]. While more than 30 controlled studies, including 17 RCTs, have evaluated the effect of yoga on markers of sympathetic/parasympathetic activation and cardiovagal function [125,126], our search identified only three that targeted adults with DM2, all conducted in India. These include two NRCTs [47,127] assessing the effects of 3month program of simple pranayama exercises [127] and a 40-day comprehensive yoga intervention [47] and an RCT of SKY yoga (a cyclical breathing practice) in adults enrolled in a diabetes lifestyle modification program that included daily brisk walking [54,55] (Table 2). Participants who completed a pranayama program showed significant improvements in multiple indices of cardiac autonomic function [47,55] and significant reductions in heart [47] and respiratory rate [127], relative to controls receiving standard care [55,127] or standard care combined with exercise [47]. These findings suggest that yogic practices may promote a reduction in sympathetic activation, enhancement of cardiovagal function, and a shift in autonomic nervous system balance from primarily sympathetic to parasympathetic in adults with DM2. In addition, findings of three recent NRCTs suggest that yoga may also help mitigate the central and peripheral nervous system damage associated with DM2. For example, following completion of a 45-day daily yoga program, participants showed significant declines in the latency of event-related potentials (ERP), a marker of higher brain, including cognitive and memory function, and increases in ERP amplitude compared to those assigned to standard care [46]. Likewise, adults with DM2 who completed a 40day yoga intervention showed significant improvements in certain measures of median nerve conduction velocity and amplitude relative to controls [47,50], although incomplete information and lack of between group comparisons limit interpretation of findings. [128,129] and a significant predictor [130] of DM2 and has been inversely associated with insulin resistance [128] and glycemic exposure [129,131]. While studies in adults with DM2 are limited, available data from two NRCTs support a possible beneficial influence of yogic practices on pulmonary function in this population [47,127]. Reported improvements included significant increases in forced expiratory volume, forced vital capacity, peak expiratory flow rate, and maximum voluntary ventilation following completion of a 3-month program of simple yogic breathing exercises [127] or a 40-day comprehensive program [47].

Medication Use.
Three controlled trials to date (two NRCTs [44,49], one RCT [57]) have shown significant reductions in diabetes medication use in patients completing a three- [44,49] to nine-month [57] yoga program relative to those assigned to standard care [44,49] or a comprehensive exercise program [57]. Some of these declines were quite substantial. For example, in their trial of 154 adults with diabetes, Agrawal et al. reported 26 to 40% reductions in medication use in the yoga group at the 3-month follow-up [49].

Discussion
Overall, findings of these studies suggest that yoga-based practices may have significant beneficial effects on multiple factors important in DM2 management and prevention, including glycemic control, insulin resistance, lipid profiles, body composition, and blood pressure. These findings are further supported by recently published meta-analyses regarding the effects of yoga on specific CVD risk factors of relevance to DM2, detailed in Table 4 [34,[71][72][73]. For example, in subanalyses restricted to RCTs in adults with DM2, Cramer et al. reported greater average declines of 26 mg/dL in FBG and 0.5% in HbA1c, greater mean reductions of 13 mg/dL in TC, 10 mg/dL in LDL, 5 mg/dL in VLDL, and 24 mg/dL in Tg levels, and a higher average increase of 6 mg/dL in HDL in participants assigned to a yoga versus standard care group [34]. Notably, the authors also found significantly greater reductions in LDL (9 mg/dL) and higher increases in HDL levels (4 mg/dL) in the yoga versus conventional exercise group. Likewise, Cramer et al. also reported greater mean reductions in waist-hip ratio in participants with DM2 assigned to a yoga versus a standard care group [34]. Similarly, in a subanalysis of RCTs in adults with diabetes or metabolic syndrome, Chu et al. reported a mean BMI reduction of over 1.6 kg/m 2 in those in the yoga versus control group [71]. While studies regarding the effects of yoga on blood pressure in adults with DM2 remain few, recent metaanalyses regarding the effects of yoga on other populations at risk for CVD suggest that yoga may be beneficial for regulating blood pressure in patients with diabetes as well. For example, in subanalyses of RCTs limited to 8 trials of adults without frank diabetes but at high risk of DM2 and CVD (e.g., those with obesity, impaired glucose tolerance, or metabolic syndrome), the authors noted significant mean declines of 10 mmHg in systolic blood pressure and 7.5 mmHg in diastolic blood pressure in participants receiving a yoga intervention versus standard care [34]. Meta-analyses of controlled trials in adults with hypertension have yielded similar findings [72,73], indicating significant declines of 8-10 mmHg in systolic blood pressure and 6-7 mmHg in diastolic blood pressure in the yoga versus standard care groups (Table 4).
More limited data suggest that yoga may also lower oxidative stress, decrease sympathetic activation and improve nervous system function, enhance pulmonary performance, mood, sleep, and quality of life, and reduce medication use in those with DM2. Relatively few studies have examined the effects of yogic practices on psychological status and sleep,  which are frequently compromised in DM2 and significant contributors to diabetes progression. Likewise, studies evaluating the effects of yoga on clinical outcomes and certain emerging risk factors in DM2, including proinflammatory markers, remain sparse, and studies to determine optimal yoga dosing and program structure for different DM2 populations are lacking. As indicated in Table 1, the vast majority of controlled studies published to date regarding the effects of yoga for DM2 have been conducted in India, underscoring the need for rigorous trials in other racial/ethnic groups and in other developed and developing countries. Finally, as discussed in more detail below and reflected in the study quality scores (Table 2), the methodological limitations characterizing existing studies highlight the need for additional high-quality RCTs that address these concerns.

Yoga and DM2: Possible Underlying Mechanisms
Although the mechanisms underlying the apparent beneficial effects of yoga therapy on diabetes risk profiles are not yet well understood, mechanistic pathways are likely complex and interacting. The observed changes may occur through at least four pathways, as depicted in Figure 1 [10,76,113,117,125,132]. First, yoga may lessen the negative impact of stress and promote multiple positive downstream effects on metabolic function, neuroendocrine status, and related inflammatory responses and, ultimately, reduce risk for CVD and other vascular complications, by enhancing well-being and reducing reactivity and activation of the HPA axis and the sympathoadrenal system [10,76,113,132]. For instance, recent controlled trials suggest even short term yoga training programs can reduce perceived stress, improve mood, and lower catecholamine and cortisol levels, cardiovascular response to stress, blood pressure, and other indices of sympathetic activation in both healthy and clinical populations, including adults with diabetes [10,35,80,113,[132][133][134]. Sympathetic activation, HPA dysregulation, and psychological distress have also been linked to the development and exacerbation of several risk factors for DM2 [10,76,80,83,88,132,133,135,136], including insulin resistance, impaired glucose tolerance, hypertension, dyslipidemia, central obesity, and proinflammatory and prothrombotic states. These conditions have been shown to have adverse effects on metabolic control and neuroendocrine function and have also been implicated in elevated risk for DM2, CVD, and other vascular disorders [10,76,80,83,88,132,133,135].
Second, yogic practices may shift the autonomic nervous system balance from primarily sympathetic to parasympathetic, by directly enhancing parasympathetic output, possibly via vagal stimulation [76,122,137], resulting in positive changes in cardiovagal function and associated neuroendocrine, hemodynamic, and inflammatory profiles, in sleep and affect, and in related downstream metabolic parameters ( Figure 1, pathway 2) [10,80,113,133]. For example, recent controlled studies in adults with DM2, CVD, hypertension, and other chronic conditions, as well as in healthy populations, have shown yogic exercises to reduce resting heart rate, enhance baroreflex sensitivity, and increase heart rate variability, both immediately and following short term (6-12 weeks) yoga programs [10,113,133,138,139]. Heart rate variability, resting heart rate, and baroreflex sensitivity are widely used markers of cardiovagal autonomic function, parasympathetic activation, and autonomic balance [76,120,140,141] and are thought to in part reflect stimulation of the vagus nerve [76,122,137]. These factors have been strongly linked to increased risk for DM2, as well as CVD [10,76,80,142].
Third, yoga may also promote favorable changes in autonomic balance, memory and mood, neurological structure and function, and related metabolic and inflammatory responses by selectively activating specific brain structures and neurochemical systems related to attention and positive affect (Figure 1, pathway 3), as suggested by recent neurophysiological and neuroimaging research findings [143][144][145][146]. And finally, yogic practices may improve both metabolic and psychological risk profiles, support increased physical activity, enhance neuroendocrine function, improve body composition, and promote weight loss by increasing strength, overall fitness, and physical function (Figure 1, pathway 4). Yoga may also reduce CVD risk in other ways. For instance, by reducing stress and leading to improved sleep and mood, yoga may indirectly improve CVD risk profiles by leading to healthier lifestyle choices and enhanced self-care [147]. Yoga may also increase resilience to stress, a factor that has been linked to improved outcomes in DM2 [148,149], although study findings to date have been inconsistent [150][151][152].
In addition, yoga may benefit those with DM2 indirectly by encouraging improvements in health-related attitudes and lifestyle choices [147,153,154] and by providing a source of social support, a factor linked to improved diabetes self-care and clinical outcomes [155,156]. As suggested by recent studies of patients with heart failure, chronic obstructive pulmonary disease, and DM1 [157][158][159], yogic breathing practices may also, by increasing arterial and tissue oxygenation, alleviate underlying hypoxia and thereby enhance autonomic cardiac and respiratory function and related endpoints in adults with diabetes [157][158][159][160]. Finally, several recent genomic investigations in dementia caregivers [161,162] and healthy adults [163,164] suggest that yogic meditative practices can slow cellular aging and induce beneficial epigenetic changes in pathways regulating inflammation, oxidative stress, energy metabolism, insulin secretion, mitochondrial function, and other related factors; these changes may, in turn, help buffer the deleterious effects of stress, improve glucose control, enhance mood, sleep, and autonomic function, reduce blood pressure, and promote improvements in other related risk factors of relevance to DM2 management [76,165,166].

Study Limitations and Directions for Future Research
This systematic review was limited in that it did not include unpublished studies, dissertations, or abstracts. In addition, some controlled trials of yogic practices may have been missed if the article did not make reference to the word yoga or yogic, although every effort was made to locate all relevant studies meeting our inclusion criteria. Many of the studies reviewed here suffer from methodological problems, poor reporting, and other limitations that render interpretation of findings challenging and the formation of definitive conclusions difficult. PEDro scores ranged from a low of 1 [38,40,43,64] to a high of 6 [57,59,61,62,66,67] points, out of 10 (Table 2). Importantly, participants were not randomly assigned to treatment in more than half of the studies, increasing risk for selection bias, and concealment of treatment allocation was reported in only 3 [57,61,65] of the 25 trials. Among the 12 RCTs, only 2 [57,65] specified how randomization was performed, and only 7 [52-56, 58, 59, 61-63, 66-68, 70] demonstrated baseline comparability between the groups regarding the most important prognostic indicators. Interestingly, 8 [39,[44][45][46][47][48][49][50] of the 13 NRCTs also exhibited groups that were similar at baseline. Given the nature of the intervention, blinding of the therapists and participants was likely not feasible; understandably, none of the studies satisfied these two PEDro criteria. This deficiency in blinding potential places even greater importance on ensuring that outcome assessors are blinded to group assignment. Unfortunately, blinded outcome assessment was specifically addressed in only two trials [57,66,67], raising the possibility of information bias. Retention was not always reported and varied widely (from 60% to 100%), with less than half of the studies ( = 11 [39, 42, 45, 50, 56, 58-62, 66, 67, 69, 70]) reporting outcome data from 85% or more of those initially assigned to a group. Reasons for dropout were rarely specified. Only five studies included specific data on participant adherence [45,56,65,67,70], and even in these studies, adherence was highly variable, ranging from excellent [45,67] to very poor [65]; moreover, for one study [53], adherence data were reported for only a subset of participants [56].
In some articles, details regarding analytic methods were sparse, and/or analyses were inadequate. For example, while all studies in this review included a comparator group, between group comparisons were not presented in 10 of the 25 studies, including 2 [60,64] RCTs. In those studies documenting attrition, only eight [39,42,50,57,59,60,65,69] indicated all participants completed the study as designed or that intention to treat analyses had been conducted. Additionally, although all but two [64,65] studies provided point measures and measures of variability, other data were often incompletely or confusingly presented. Furthermore, in many papers, the study population, participant recruitment, yoga-based intervention, and/or comparator condition were poorly described, rendering replication of these studies challenging and limiting conclusions. Most of the trials were relatively small, with 56% of them including 60 or fewer participants and 24% including 40 or less (Table 1), although, overall, sample size has been increasing in the last decade, especially the last 5 years.
While the investigations published to date have yielded overall positive results, the considerable heterogeneity in study design, intervention, comparator condition, and duration characterizing the 25 trials render direct comparisons across studies challenging. Participant characteristics also varied substantially, both within and between studies. Nonetheless, while this heterogeneity precludes specific recommendations for particular yoga programs or subpopulations, the consistently positive findings reported in multiple and varied samples of adults with DM2 suggest that yoga may be helpful for a broad range of patient groups. Of the 25 eligible controlled studies identified in this review, only five were conducted in countries other than India, including two in the UK, two in Cuba, and one in Iran.

Journal of Diabetes Research
Rigorous trials in Western and many other countries, including many developing nations, where DM2 is now reaching epidemic proportions, remain few. It thus remains unclear if the positive findings reported in existing studies can be generalized to other populations, cultures, and/or regions. As indicated earlier, the mechanisms underlying the observed benefits of yoga are still poorly understood. Finally, rigorous dosing, cost-effectiveness, and long-term follow-up studies are lacking, as are trials assessing the effects of yoga on clinical endpoints such as diabetes-related morbidity and mortality. All of these areas warrant future research.

Conclusion
In conclusion, the findings of controlled trials published to date suggest that yogic practices may promote significant improvements in several indices of major importance in the management of DM2, including glycemic control, lipid levels, and body composition. More limited data suggest that yoga may also lower oxidative stress and blood pressure, enhance pulmonary and nervous system function, improve mood, sleep, and quality of life, and reduce medication use in adults with DM2. However, given the methodological limitations and heterogeneity of existing studies, findings must be interpreted with caution. Additional high-quality investigations are required to confirm and further elucidate the potential therapeutic benefits of standardized yoga programs in populations with DM2 [10,132].

Disclaimer
The contents are solely the responsibility of the authors and do not represent the official views of West Virginia University or the National Institutes of Health.