The impact of yoga on components of energy balance in adults with overweight or obesity: A systematic review

Abstract Background Yoga may reduce body weight in individuals with overweight or obesity, but whether this occurs through decreased energy intake (EI) or increased energy expenditure (EE)/physical activity (PA) is unclear. Methods A systematic search of PubMed, Web of Science, Embase, and PsychINFO was conducted from inception until April 26, 2021. Eligible studies included randomized controlled trials or single‐arm pre‐post studies with any type and duration of yoga intervention in adults with overweight or obesity. Studies with measures related to EI , EE, or PA were eligible. The review initially identified 1,373 articles. Results Of the 10 included studies, one used indirect calorimeter measures of resting EE, while nine used self‐reported measures of EI and PA. Of the seven studies measuring parameters related to EI, only one found greater decreases in EI relative to the control group, although three other investigations reported trends toward improved dietary intake. Of the eight studies measuring PA, two reported greater increases in resting EE or PA in the yoga group relative to the control group. Two reported significant within‐group increases in PA from pre‐post intervention, and four studies reported a trend for increased PA with no p‐values reported. Conclusions Limited evidence suggests yoga may reduce EI and increase PA in adults with overweight or obesity. Additional studies that investigate the effects of yoga interventions on energy balance parameters using objective techniques are warranted.


| INTRODUCTION
Improving lifestyle modification strategies for weight loss is an urgent public health priority due to the continued increasing prevalence of overweight and obesity, 1,2 the resultant incidence of comorbidities, [3][4][5] and substantial financial burden on health care systems worldwide. 6 Current obesity treatment guidelines recommend lifestyle modification interventions that include reducing energy intake (EI), increasing physical activity (PA), and enhancing counseling for behavioral modifications by trained interventionists. 7 This lifestyle modification intervention approach typically produces 5%-10% short-term weight loss, which provides significant health benefits. 7 However, long-term success in maintaining weight loss is poor with about 50% of lost weight typically regained within 1 year. [8][9][10][11] For lasting weight loss maintenance, obesity treatment guidelines recommend continued participation in a long-term (≥1 year) comprehensive weight loss maintenance program. 7,12 It is therefore imperative to evaluate practical and cost-effective longterm strategies to improve the effectiveness of lifestyle modification interventions.
Yoga is a form of complementary medicine, that is, quickly growing in popularity in the United States 13 and may be an effective strategy to improve the efficacy and durability of long-term weight loss outcomes through lifestyle modification interventions. A recent systematic review and meta-analysis of randomized controlled trials (RCTs) demonstrated that yoga interventions led to significant reductions in body mass index (BMI) in the subset of five studies that specifically included adults with overweight, obesity, or metabolic syndrome. 14 However, whether lower BMI occurred through reduced EI, increased TDEE, or both has not been systematically investigated.
This represents a considerable knowledge gap because understanding the extent to which yoga interventions impact specific energy balance parameters can inform the design of more effective lifestyle modification interventions.
There are several mechanisms through which yoga may theoretically improve energy balance. 15 For example, yoga may help reduce EI by heightening mindfulness and the mind-body connection, improving mood affect, and reducing stress. 16,17 Among individuals without obesity, regular yoga practice is associated with better dietary quality and improved management of emotional eating. 18 In a qualitative study, individuals who lost weight through yoga described mindset shifts away from weight loss and toward health, increased mindfulness and focus, and improved selfesteem. 19 Improvements in self-esteem and reductions in stress may be particularly relevant for mitigating the negative effects of weight stigma on weight loss, healthy eating, and PA. 20 In addition, specific yoga postures may help increase EE directly, and by reducing some of the barriers to adopting and sustaining PA more broadly, as yoga can lead to reduced back and joint pain 15 and improvements in physical function, isometric strength, cardiorespiratory fitness, and balance. 17,21 These unique psychological and physical changes associated with yoga interventions may therefore provide a form of activity, that is, more reinforcing for some individuals than higher intensity resistance or aerobic activities, 22 as well as support changes in EI and moderate/vigorous intensity PA known to be critical for weight management. These elements may then facilitate longer-term maintenance of dietary and PA changes after the intervention.
While these results are encouraging, most of the evidence supporting the positive effects of yoga on energy balance are from cross-sectional or retrospective studies in populations without obesity. Understanding how yoga affects specific aspects of energy balance in people with overweight or obesity can inform novel and potentially more effective strategies for sustained weight loss.
However, to date there has not been a systematic evaluation of the literature examining the extent to which yoga interventions lead to reduced EI, increased EE, or a combination of both in people with excess body weight. As such, the objective of this systematic review was to evaluate the existing literature examining the effect of yoga interventions on EI and PA among adults with overweight or obesity.

| Search strategy
This review was planned, conducted, and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations. 23 A systematic literature search was conducted and included articles from inception until 26 April 2021 using PubMed, Web of Science, Embase, and PsychINFO.
The search strategy entailed three independent themes of key words and medical subject heading (MeSH) terminology related to: (1) "yoga," (2) "EI," "PA," or "EE," and (3) "overweight" or "obesity" (Appendix 1 in the supporting information File). Search terms in each theme were linked using "OR" as a Boolean function and each theme was combined using "AND" as a Boolean function. To maximize study inclusion, no search limits on language of publication or record type were applied. Results from each database were managed using EndNote (Version X9, Clarivate Analytics, Philadelphia, PA, USA). The study team registered the study protocol on the International Prospective Register of Systematic Reviews (PROSPERO) database (CRD42020179845).

| Data extraction
Two authors (A.C. and S.A.P) independently extracted relevant data.
Authors were not blinded to information regarding authorship, institutions of origin, or journal of publication.

| Risk of bias
The National Institutes of Health's study quality assessment tools were used to determine risk of bias. 25 The checklists are designed to   help reviewers determine internal validity for before-after studies   with no control group and RCTs through 12 and 14 questions, respectively. Two authors (A.C. and S.A.P) independently evaluated each study using the appropriate checklist. Discrepancies were resolved through discussion between researchers. Results are presented for each question ("yes," "no," "cannot determine," "not applicable," or "not reported") and as an agreed-upon overall risk of bias (low, moderate, or high, corresponding to the tool's terminology of poor, fair, and good study quality, respectively).

| Description of studies
The literature search uncovered 1373 potentially eligible articles. After removing 482 duplicates and screening titles, abstracts, and full texts, 10 articles met inclusion criteria, Figure 1. These included eight RCTs 26-30 and two pre-post studies. 31,32 Of note, the article by Telles et al. 32 included two study groups but was categorized as a pre-post design due to the lack of randomization and comparison between groups. In addition, two RCTs recruited individuals with metabolic syndrome as opposed to overweight or obesity defined by BMI. These were included in this review because (1) high waist circumference is one of three criteria needed for metabolic syndrome and is also an indicator of central obesity and (2) first line of treatment for metabolic syndrome is changes in dietary intake and PA, similar to obesity treatment. Table 1  Yoga interventions varied widely in the styles and specific body-, breath-, and mind-based practices included, as well on the level of detail provided to interpret the style of yoga or practices included in the intervention. All studies included body-based yoga postures, or asanas, with three studies focusing exclusively on yoga postures. 28,33,34 The intensity of the asana practices included in the interventions varied from being low intensity-described as restorative, therapeutic, or a means of stretching 26,33,35 -in three interventions, with six interventions providing descriptions indicating a more physically challenging asana practice. [28][29][30][32][33][34] Two studies specified using pranayama controlled breathing techniques, 30,32 which were only thoroughly described in the study by Telles et al. 32 One study briefly mentioned breathing as part of the warm-up, four CALDWELL ET AL. studies described a focus on breath during postures 31 or relaxation. 27,29,35 Three studies specified including meditation techniques in line with classical Patanjali teachings, 30-32 while one simply described meditation as part of relaxation, 27 and another included meditations focused on Bible verses during relaxation. 35

| Energy intake outcomes
Seven studies measured some dimension of dietary intake before and after a yoga intervention. 26,[28][29][30][31][32][33] Of these, five studies combined yoga with dietary guidelines or advice, 28,[30][31][32][33] while two studies did not provide any instructions on diet and explicitly asked participants to not change their diet during the intervention. 26,29 All studies assessed dietary intake or nutrition via self-reported measures, including a modified nutrition-behavior questionnaire, 31 the Eating Behavior Inventory, 33 food frequency questionnaire, 26 24-hour recall, 30,32 or 3days food records. 28,29 No studies used more rigorous measures of free-living appetite or dietary intake (e.g., measures of hunger and satiety, doubly labeled water, etc.).
Of the five studies in which participants were given dietary advice, four reported changes in dietary intake and one found no change. In a large 12-weeks RCT reported by Yadav et al. 30 all participants (n = 260) received a personalized diet plan by a registered dietitian. In this study 130 individuals were randomized to also receive a yoga intervention 5 days/week in-person and athome. Both groups had decreased total EI and % of EI from fat, increased % EI from protein and carbohydrates, and increased fiber (g/day). The yoga group had greater reductions in total EI. In a recent RCT by Jakicic et al. 33 50 adults with overweight or obesity were randomized to either a Vinyasa or restorative yoga for 24 weeks. Both groups received instructions on EI from 1200 to 1800 kcal/day. Both groups decreased EI and % of EI from fat and increased % of EI from carbohydrates; the Eating Behavior Questionnaire score also increased. In a small RCT (n = 26) among women with overweight or obesity by Ruby et al. 28 participants in all intervention groups were given macronutrientbalanced meal plans (50% carbohydrate, 15% protein, and 25% fat) designed to meet 100% of their estimated energy requirements. Following the 12-weeks yoga intervention (3 days/ week in-person at-home), there were no changes in dietary parameters attained from 3-days food records. 28 In a pre-post pilot study by Telles et al. 32 women with abdominal obesity (n = 29) received a 12-weeks yoga intervention (3 days/week, in-person) and a lacto-vegetarian diet plan consisting of 1900 to 2000 kcal/ day. EI appeared to decrease with concomitant decreases in protein and increases in carbohydrate and fat intake (in g/day) in the yoga group. 32 However, as this was a pilot study, no p-values were presented. In a single-arm, 5-days residential yoga study by Braun et al. 31 participants attended yoga and lifestyle/nutrition classes centered around mindful eating that included cooking demos and meal planning (n = 39). A general "nutrition score" improved directly following the brief intervention and after 12 weeks; however, the 12-weeks change was not significant after Bonferroni correction. 31 The RCTs by Siu et al. 29 and Cohen et al. 26 explicitly asked participants to not change dietary intake. Siu et al. 29 randomized 146 people to a 1-year yoga intervention (3 days/week in-person) or a no intervention control (n = 137). There were no significant differences self-reported EI, macronutrients, sugar, or cholesterol attained by 3-days food records between groups. Notably, selfreported EI assessed by the Block Food Frequency Questionnaire (FFQ) appeared to increase in both yoga and wait-list control groups in the pilot study by Cohen et al. 26 However, no withingroup p-values were presented and the differences in EI change between groups was not significant.

| Physical activity and energy expenditure
Seven studies measured PA and one study measured resting metabolic rate before and after a yoga intervention. In six of these studies, 27 36 Given that weight is included in this predictive equation In the study by Jakicic et al. 33  However, this increase included yoga completed in a residential setting and was not maintained at the 3-months follow-up. 31 One investigation noted higher resting metabolic rate after the yoga intervention that was not observed in the control condition. 34 Two RCTs explicitly instructed participants not to change habitual levels of PA. Despite instructions not to raise PA, Siu et al. 29 observed increases in MET-minutes/week from baseline to postintervention in both the yoga and control groups. However, these changes were not significantly different between groups, and pvalues were not reported for within-group changes. In the pilot study by Cohen et al. 26 no differences between groups in PA change were observed. However, results suggest that both changes in PA favored the yoga group in both PA hours/week and METs/week, though no pvalues were reported for within-group changes. 26

| Bias
Most studies were likely to have a low (n = 3, 30%) or moderate risk of bias (n = 6, 60%); 1 (10%) had a high risk of bias (Table 2). In studies with pre-post designs, the primary sources of bias were from not reporting or not measuring the following: whether all eligible participants were included, details of loss to follow-up, statistical analyses, and measuring outcomes more than once during and after study periods. For RCTs, the main concerns were not reporting if the study team were blinded to the participants' group assignments, adherence, and avoidance of other similar interventions. Drop-out rates >20% and lack of intention-to-treat analyses were also common contributors to high risk of bias in RCTs.

| DISCUSSION
Theoretically, yoga may influence several key determinants of health behavior change that support weight loss in people with obesity. [15][16][17][18][19] Whether yoga-induced weight loss occurs through decreased EI, increased PA, or both is unclear. As such, this systematic review examined the effect of yoga on EI and PA in adults with overweight and obesity. Our results suggest that the addition of yoga to a weight loss program may help reduce EI and improve several aspects of dietary intake, 30,32 but there is insufficient evidence to suggest that stand-alone yoga interventions independently alter EI. Limited evidence suggests that self-reported PA increases in response to a yoga intervention, 30 though there is currently not consistent evidence that yoga is associated with increased non-yoga PA. Importantly, in the only large RCT included in the systematic review where participants were not instructed keep habitual diet and PA patterns the same, the yoga + diet intervention led to greater decreases in EI and increases in PA than the diet intervention alone. 30 Thus, while the overall body of literature is insufficient to definitively describe the effects of yoga on EI and PA among individuals with overweight and obesity, there is promising preliminary evidence that yoga can lead to improvements in both diet and PA, particularly when added to a lifestyle CALDWELL ET AL.  Risk of bias  1  2  3  4  5  6  7  8  9  10  11  12  13  14 Pre-post design Notes: Questions are from the National Institutes of Health (NIH) Quality Assessment Tool for Before-After (Pre-Post) Studies with No Control Group and the NIH Quality Assessment of Controlled Intervention Studies (listed below). Each cell represents an assessment of each study, based on review and consensus from two reviewers (A.C. and S.P) as follows: Y, yes; N, no; CD, cannot determine; NA, not applicable; NR, not reported.
intervention. More large-scale, rigorously designed studies are needed that are adequately powered to test for between group differences in changes in diet and PA.
There was also wide variability across studies included in this review in the different styles and specific yoga practices that  30 Notably, all dietary data was ascertained using participant self-reports, which are often inaccurate in people with obesity 40,41 and may have obscured legitimate alterations in dietary intake. The effects of a yoga intervention on objective EI and the determinants of such in adults with obesity therefore remain largely unknown.
There are several unique aspects of yoga that suggest yoga positively influences factors upstream could enhance EI-related sustained behavior changes and support maintenance of weight loss.
Improvements in subjective ratings of appetite occur following 10 minutes of slowed breathing (6 bpm vs. 9 bpm). 42 Consistent slowed breathing exercises performed as part of a traditional yoga practice could subdue hunger, making reductions in EI easier to sustain over the long-term. Yoga may also enhance self-efficacy in regulating diet and PA 17,43 and reduce the frequency of binge eating, 44 which would support lasting weight loss. In a qualitative study of individuals who lost weight through yoga, all respondents with overweight prior to weight loss described the process as involving a mindset shift to healthy eating, as well as improved mood, and emotional stability. 19 Nearly all (90%) described it as a "different weight loss experience," and many described that yoga led to more mindful eating, changes in food choices, and less emotional and/or stress eating. 19 55 In a cross-sectional epidemiological study of 15,550 adults age 50 to 76 years old, individuals with >4 years of yoga practice reported more than two-fold higher PA than individuals who did not engage in yoga. 56 In a non-obese sample, yoga practice increased positive affect, physical function, and selfesteem during a 20-weeks behavioral weight loss intervention with mind-body techniques (i.e., meditation, imagery). 17 Finally, among yoga practitioners in a nationally representative survey, nearly two-thirds said that doing yoga motivated them to eat healthier (63%) and exercise more regularly (63%). 57 Thus, there is strong rationale for future studies to test the extent to which yoga leads to changes in total and non-yoga PA levels in adults with obesity using objective measures to address remaining gaps in the literature.
This review is the first to describe changes in EI and PA in response to yoga in people with obesity and uncovered several limitations that should be considered when interpreting the literature.
The widespread reliance on self-reported measures of dietary intake and PA is the most important limitation of the current evidence.
These measures have poor accuracy in identifying actual EI and PA, 40,41,58 and therefore impede current understanding of how engaging in yoga practice may influence other aspects of health behavior. Error is especially pronounced with abbreviated methods of dietary intake and PA such as 24-hour diet recalls and PA questionnaires, as they do not capture variability in behavior. Furthermore, the use of various self-report measures and different metrics from the same measure precludes the ability to compare findings across studies. Future research using more rigorous techniques for assessment of EI, appetite, PA, and total EE (e.g., doubly labeled water, accelerometers, meal studies, and food photography) is warranted and essential. Current interpretation is also limited by the guidelines on health behaviors individuals received in several interventions. Instructions to avoid changes in EI and PA obscures any independent changes and likely diminishes the true effect of yoga on EI and PA. There was also a wide variability in the type of yoga in each intervention, with some providing more details about the intervention than others. The array of different methodologies regarding participant instructions and yoga interventions also hampers a rigorous comparison of studies. A limitation of this particular review is that a meta-analysis was not possible given the differences in study design and outcomes measures.
In conclusion, there is currently only limited evidence that yoga improves EI and PA parameters in adults with overweight or obesity. However, this is at least partially due to the small number of studies that have measured these outcomes in this population in response to a yoga intervention (most with small sample sizes), the wide variability in study designs, and lack of rigorous, objective measurements of these variables. The rationale for examining the effects of yoga on EI and PA during weight loss is compelling and suggests that yoga holds promise as a strategy to support weight loss and weight loss maintenance in people with obesity. Whether

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.