Mindfulness-Based Interventions and Cardiovascular Risk Factors in US Racial/Ethnic Minority Populations: A Systematic Review of Implementation Data to Address Health Disparities

Cardiovascular disease (CVD) is a persistent public health challenge. Mindfulness-based Interventions (MBI) have been researched for CVD risk factors, though their effectiveness, generalizability, and potential for implementation to racial and ethnic minorities remain unclear. This review examines studies of MBI on CVD risk for characteristics of and variations in implementation (i.e., intervention design, delivery, uptake, and contextual factors) and analyzes potential barriers and challenges to implementation. A systematic review in February 2020 identied 30 studies from 5 databases and hand searches. Included studies were randomized controlled trials testing meditation or mindfulness-based interventions against any control to measure change or improvement in cardiovascular health measures or risk behaviors in adults living in the United States or territories. Analysis of the implementation characteristics and contextual factors of included studies was conducted using the Oxford Implementation Index. Thirty reports from 26 distinct trials were selected for inclusion, examining outcomes related to diet (k [number of studies] = 13), smoking (k = 11), obesity (k = 9), exercise (k = 4), diabetes (k = 3), and blood lipids (k = 2). All studies were published between 2011 and 2020 and correspond to early stages of research. As such, numerous limitations and implementation characteristics with potential consequence for CVD risk disparities were reported. This review outlines several potential targets for future research. Based on reported ndings across all included studies, MBI could be of benet for cardiovascular disease risk. Further research is needed to explore acceptability, feasibility, and effectiveness in minority populations. Freedom from Smoking (FFS), Integrated Body Therapy (IMBT), Mindfulness-Based Treatment Mindful Living Mindfulness-Based Awareness Mindfulness (MB-EAT-D), Mindfulness-Based Stress Reduction (MBSR), Mindful Mindfulness (MYW), (WHAM)

Disparities in cardiovascular disease and risk.
Signi cant disparities in cardiovascular disease for racial and ethnic minorities are well-documented. Factors contributing to these disparities occur at the level of the patient, the provider, and the overall healthcare system, and include health behaviors, genetics, provider bias (intentional or unintentional), and access to healthcare coverage [7]. Black Americans have higher rates of heart attack, heart failure, and stroke [8][9]. Hypertension prevalence is higher in Black Americans than in White Americans, occurs at younger ages, and is often accompanied by higher blood pressure, correlating with increased overall risk of cardiovascular disease [4]. Along with Mexican-Americans, black patients have the highest rates of high blood pressure and exhibit greater di culty controlling blood pressure levels, relative to other Summary and Aims.
Mindfulness-based Interventions (MBI) are increasingly studied for the treatment of CVD and risk factors [21,37]. According to a scienti c statement from the American Heart Association [3], the current literature suggests that MBI are a low-cost, low risk adjunctive treatment for the modi cation of lifestyle related to cardiovascular risk. Despite the disparate rates of CVD morbidity, mortality, and risk in US racial/ethnic minority populations, the effectiveness and implementation potential of these interventions for US racial/ethnic minorities remains largely unexplored. In light of this gap in current scholarship, this systematic review summarizes variations in study design and implementation (i.e., intervention design, delivery and uptake, and contextual factors) and explores targets for future tailoring to underserved populations with higher risk for cardiovascular disease.

Methods
Protocol & registration. The protocol for this review was developed in consultation with a subject librarian as well as experts in mindfulness-based interventions, public health interventions, and systematic reviews. This protocol was registered with the Prospero International Prospective Register of Systematic Reviews (registration number CRD42019141454) on 09/27/2019.

Population
Healthy or unhealthy adults (18+) residing in the United States or territories.
Intervention/Exposure Meditation or mindfulness-based intervention with a predetermined curriculum consisting of at least 1 practice session.
Control Any control group. Active and/or passive control groups.

Outcomes
Change or improvement in cardiovascular health measures or risk behaviors.

Study Design
Randomized Controlled Trials (RCTs) Search and selection of articles. The search strategy for this systematic review was guided by specification of the research question parameter, based on PI(E)CO (Table 1). This strategy included crossing terms specific to a) our interventions of interest, and b) our defined comparison (RCTs and semi-randomized trials), and did not cross these terms with specific health conditions of interest or populations in order to widen our search. Searches were limited to all studies in: a) the English language and, b) to adult human studies published in the United States and territories. Databasespecific search strategies were created to accommodate truncation and MeSH or Emtree terms. The search was conducted on February 9, 2020. The following five electronic databases were searched: PubMed, EMBASE, PsycINFO, Cochrane Library (CENTRAL) and CINAHL using search strategies outlined in the study protocol (See Appendix 1). Additionally, the electronic database OpenGrey was searched and no eligible studies were discovered. We also conducted forward and backward citation tracking for included studies, key papers, and relevant systematic reviews. Search results and other relevant articles were to a web-based systematic review tool designed to facilitate the process of screening, data extraction, and analysis (http://www.covidence.org).
The primary author performed initial screening of all identified titles and abstracts generated by our search strategy and removed studies which clearly did not meet inclusion criteria (see Table 1). A second author independently screened 10% of these studies and met at least 80% agreement between authors. Full-text articles were obtained of all selected abstracts and screened for inclusion. Two authors performed an independent review of articles to determine final eligibility with differences resolved through a senior author. Finally, a subject expert was consulted prior to the conclusion of this research to verify that included studies were representative of the current literature and appropriate for the study's aims and objectives. The results of this screening are summarized by the PRISMA diagram in Figure 1.
Inclusion and exclusion criteria. This review considered randomized controlled trials of meditation or mindfulness-based interventions (as defined by Crane et al., 2016 [38]) conducted in the United States and territories using adult subjects, available in English. Studies were limited to the United States to maintain focus on the unique healthcare considerations and other US-specific racial/ethnic disparities present there. "Mindfulness-informed" therapies (as defined by Crane et al. 2016 [38]) where mindfulness or meditation practice is optional or may not be considered to be the intervention's core component-such as Dialectical Behavioral Therapy (DBT) and Acceptance and Commitment Therapy (ACT)-were excluded. Similarly, studies using psychotherapies such as Cognitive Behavioral Therapy (CBT) were also excluded. Interventions including multivariate components to tailor interventions to health conditions or populations were considered eligible for inclusion.
This review was narrowed to include studies of MBI for the improvement of the following six modifiable risk factors for CVD: (1) diabetes mellitus (type-2), (2) smoking, (3) obesity, (4) sedentary lifestyle, (5) unhealthy diet, (6) blood lipids (high cholesterol). This set of risk factors was chosen in part due to previous work by Loucks and colleagues [21,32], who have proposed a theoretical framework and potential mechanisms for mindfulness on cardiovascular risk. To qualify for inclusion, studies needed to measure at least one physical or behavioral outcome relating to these risk factors (ie: weight, glycemic levels, emotional eating, cigarettes smoked). Primary correlational findings (ie: predictive studies on trait mindfulness) were not considered. Potential secondary outcomes of interest were also considered. These included (1) onset of CVD or other disease condition (e.g. diabetes), mortality from CVD, and incidence of hospitalization or surgical remediation due to CVD, (2) changes to or improvements in diagnosed CVD conditions, such as hypertension, coronary artery disease, peripheral artery disease, or atherosclerosis, (3) changes in psychological outcomes such as depression measures, stress levels.
Data extraction and analysis. The Oxford Implementation Index [39] was used to guide the extraction and analysis of implementation data relevant for this review (see Appendix 2). Five key domains from this Index were selected based on their relevance to the implementation of MBI: 1) dosage; 2) delivery method; 3) staff characteristics; 4) settings, locations, and dates/times; and 5) participant characteristics. Dosage in high amounts or for long periods of time may lead to worse outcomes or attrition due to schedule conflicts or childcare needs. Delivery methodparticularly in-person classes or sessions-may pose a barrier if the classes are nonrepresentative/held in noncommunity locations. In-person classes may also have schedules which are more rigid as opposed to mobile interventions which can be tailored to a subject's own needs. Similar to dosage and delivery methods, settings, locations, and dates/times of interventions may pose a barrier to engagement in certain populations due to scheduling constraints, travel constraints, or discomfort with the setting (laboratory settings on college campuses). Finally, non-representative staff and/or participant characteristics have the potential to act as a barrier to engagement, as often, minority subjects report [40]. We also made note of potential sources of contamination where applicable, in which aspects of study design could influence either the subjects recruited or the study's outcomes.
The Oxford Implementation Index acknowledges the potential for variation across intervention studies in design and implementation aspects, thus allowing reviewers to appraise this data for potential sources of bias and its generalizability across populations. Risk of bias analysis was not deemed necessary for the purposes of this study, as efficacy reviews have been conducted elsewhere and study quality was not the primary focus of this review. Examination of implementation factors provided a critical framework for this review to examine generalizability to and potential barriers to intervention engagement by minority populations.

Results
Overview of included studies. Electronic database searches produced 10,630 articles with 2,075 remaining after the removal of duplicates. After screening the title and abstract of each, the full text of 91 articles was retrieved for further review. A total of 30 articles reporting findings from k [number of studies] =26 distinct intervention trials were selected for final inclusion, all of which were handsearched for additional studies. A full breakdown of this search process including reasons for exclusion can be seen in Figure 1. Numerous studies (k=11) examined more than one risk factor. Of the included studies, the majority examined outcomes related to diet (k=13), obesity (k=9), and smoking (k=11). The remaining studies examined outcomes related to exercise (k=4), diabetes (k=3), and blood lipids (k=2). Number of participants ranged from 18-412 with k=20 studies having n<100 participants. All articles included were published between 2011-2020. Six articles that met inclusion criteria were part of the same two parent studies [65][66][67][68][69][70] and present the same basic implementation characteristics.    Table 2 and summarized here briefly. Mindfulness-based interventions produced positive results in smoking outcomes. Two studies reported a statistically significant reduction in smoking over controls [58][59]. Another two studies found smoking abstinence outcomes superior in the treatment group [48,54,59,64] while others found comparable between-group outcomes for smoking reduction [49] and abstinence [42,[49][50]. Reductions in craving were also observed [42, 49-50, 54, 58] and support for the effect of MBI on craving was demonstrated by a reduction in brain activity related to smoking cues, correlated to an overall reduction in cigarette consumption in the intervention condition [53]. Additionally, brain areas relating to self-control (the anterior cingulate and prefrontal cortex) showed increased activity with meditation, and demonstrated a 60% reduction in smoking, with no reduction shown in the control group [62].
Positive dietary changes and/or changes in eating behaviors associated with MBI were supported in numerous studies [41, 44-46, 52, 55, 61, 69] and decreases in emotional eating behavior were reported in 3 studies [44][45]68]. Reductions in dining out were reported in one study [69]. Treatment was also found to produce nonsignificant reductions in the consumption of sweets [67]. Reductions in cholesterol levels were observed in 2 studies [52,65].
Mindfulness was found to be as effective as a standard diabetes self-care treatment in one study [56] and produced significant decreases in fasting glucose levels compared with controls [57]. Significant increases in HbA1C concentrations-a test which reveals average blood sugar over 2-3 months-were found in the mindfulness group in one study [45] and nonsignificant between-group outcomes were reported in another [55]. Finally, significant improvements in fasting blood glucose were found in participants with high compulsive eating scores post-intervention [68].
Weight loss or maintenance was reported for MBI to varying degrees, with some studies producing greater results than control interventions [46, 55, 68-69] and others found nonsignificant between-group improvements [44-45, 56-57, 60]. Mindfulness was found to improve positive affective response to exercise but did not increase minutes exerted relative to controls [51]. Weight loss was also found to be predicted post-intervention by reductions in reward-based eating [66].
The included pilot feasibility studies found MBI feasible and acceptable for weight loss [44], emotional eating [45], dietary changes in prostate cancer patients [43], and smoking abstinence [48]. While both the benefits and the feasibility and acceptability of mindfulnessbased interventions in American minorities is understudied, the research presented in this review points to potential benefits of MBI as a treatment option for CVD risk and provides a basis for future research in this area.
Oxford Implementation Index: Implementation characteristics. Table 3 presents information extracted from primary studies regarding implementation characteristics of the primary studies. Dosage varied widely across studies, with the majority closely modeled after the multi-week, multi-hour model set forth by MBSR and its contemporaries. Program length ranged between 1 day and 6 months, with k=24 studies consisting of ≥6 sessions. Session durations ranged between 5 minutes and 7 hours. Dates and times of program delivery remained largely undescribed. Thirteen studies reported home practice of between 15-45 minutes for between 5-7 days per week. Delivery method was largely in-person in a group setting (k=19), with 2 delivered in person individually, and 5 delivered remotely or digitally. Staff characteristics-which were at least partially detailed in 16 studies-were universally described in terms of vocation and training, with no descriptive characteristics of the staff/providers described in the primary studies (age, racial/ethnic background). No proscribed/incompatible activities or details of meetings/communication between staff were reported. Numerous studies reported differences between trial arms which did not directly relate to the research question of this review and are therefore not outlined herein.
Several sources of possible contamination were found. In nearly all studies, previous interest in, knowledge of, or experience with mindfulness was not controlled for or was not described. One study excluded participants' current or recent (previous 6 months) mindfulness or meditation practice [45], another explicitly excluded participants with prior participation in a mindfulness program [46] and two [69-70] excluded participants with prior MBSR experience or a current meditation or yoga practice. Additionally, it was unclear in most studies if recruitment or screening materials mentioned mindfulness or meditation, which may contribute to self-selection. Two studies [48, 60] avoided mention of mindfulness in recruitment materials to avoid potential self-selection or the contamination of the control group. One study [58] lists the mention of mindfulness as a limitation. Two additional studies listed limitations for not excluding based on prior mindfulness experience [58] or not screening for prior exposure [62]. Finally, one study [64] did include subjects with previous experience which they report did not impact findings. Additional potential sources of contamination existed in several studies.
Another potential source of contamination was observed in most studies, such that it was unclear if the experimenter also served as a facilitator for either treatment or control group, which could create demand effects. In one study [61] the facilitator was described as being a co-authoring doctoral student. An additional source of contamination was found in two studies which focused on low-SES smokers. These studies explicitly stated that participants could not be blinded, as many participants "worked odd hours" and it was necessary to allow them to decline if they could not attend all sessions [48][49]. This lack of blinding may bias study results to favor participants with the ability to attend sessions, and points to a potential structural barrier to engaging with in-person, group-based mindfulness interventions in low-SES individuals.
Oxford Implementation Index: Contextual factors. Table 3 presents contextual factors extracted from primary report that were deemed relevant to this review. Setting, geographic location, and dates/times were at least partially described in 10 studies. Three reported campus or laboratory settings [41,43,54] and two reported home or naturalistic settings [51,58]. The other sixteen studies provided no description of these factors. Participant characteristics were at least partially described in 22 studies (see Table 3). A total of 17 studies reported majority white participants, with 5 reporting mixed race or majority black, and 4 not reporting participant demographics (see Table 2).
Other salient characteristics were partially described (see Table 3). Median age ranged from 20-69 across all studies. Women were disproportionately represented in 15 studies (>50% female), 5 studies reported over 50% males, and 4 studies had a combination of the two. Gender identity was not reported in any study, and LGBTQ+ demographics were considered in one study [52], which consisted of lesbian participants. Educational attainment was reported in 15 studies and revealed that the majority of recruited participants had at least some college education (11 studies). Income was only described in seven studies, five of which were comprised of participants making >50k annually. Three studies specifically focused on low-SES participants [48-49, 64], though two of these studies did not describe income-based inclusion criteria or annual income for participants, instead focusing on "low-income neighborhoods" during recruitment [48][49].
In addition to these factors, it should be noted that the majority of studies were described as being in the early stages of research, either as preliminary research or pilot trials. This may account for many commonly reported limitations including small sample sizes, convenience sampling, high attrition rates, and lack of generalizability. Future research will need to address these limitations, and reasons for attrition should be explored vis-à-vis racial and ethnic background.

Discussion
While research points to potential bene ts of MBI for cardiovascular risk, this review nds that the current literature might not generalize to the US racial/ethnic minorities who are most disproportionately affected, due to limitations in the current literature with regard to inclusion/engagement of diverse populations and consideration of cultural or social factors that might affect program relevance and effectiveness with racial/ethnic minorities. Furthermore, data for many of the implementation factors needed to tailor these interventions to underserved populations are underreported in the literature. Men, racial and ethnic minorities, and individuals of low socioeconomic status were underrepresented in the included literature. This systematic review found that research into the primary risk factors for cardiovascular disease is still in the early stages, with the majority of studies showing promising effects on outcomes for smoking reduction and abstinence, weight loss, and changes in eating behaviors.
Considering that the included studies report bene cial ndings for cardiovascular risk-which occurs at disparate rates in minorities-the lack of representation presented in these ndings presents a clear gap in the literature which should be addressed in future studies. Details on implementation factors (implementation site/provider/delivery factors, contextual factors, participant factors) were unevenly reported in the primary papers. Notably, participant characteristics, staff characteristics, and study settings, locations, times and dates were underreported. No intervention effectiveness or studies otherwise meant to tailor MBI to minorities at risk for CVD have been conducted at this time.
Prior knowledge, beliefs, and social norms about mindfulness/meditation may drive participant engagement and outcomes, as supported by behavior change theories [71][72]. The need for considering prior knowledge, beliefs, and social norms about mindfulness in minority groups is also consistent with literature on tailoring interventions to minorities [35,[73][74]. While some studies described explicit exclusion for either mindfulness training or practice (see Results), the potential effects of prior knowledge of, interest in, or experience with mindfulness or meditation remains unclear across all studies. Several studies included in this review highlight these gaps in understanding. The 2016 study by Daubenmier and colleagues [65] on obesity and weight loss found that treatment participants who were missing 18-month data reported less bene ts from the study, compared with controls. These participants also demonstrated less engagement and less improvement. The researchers suggest that this may indicate a lack of interest in mindfulness and cite motivation and engagement as strong factors. Additionally, in the 2017 study by Ingraham and colleagues [18], some participants appear to have had an established prior practice, and felt the intervention was too basic for their needs. These ndings support the need to understand the perceived cultural relevance of these interventions in underserved populations.
Lack of reporting on staff characteristics-particularly facilitator characteristics-was one of the major ndings of this review. Minority groups are less likely to nd representation in mindfulness practice [23] and are more likely to respond to a MBI program facilitator with whom they can identify or whom they feel represents their community [35,[73][74]. This lack of understanding was highlighted directly by one of the studies included in this review. In Davis et al., 2014a [48], preference for one or the other study arm was cited as a secondary reason for attrition, with expressed preference for the mindfulness group among participants who had higher levels of education. Daubenmier and colleagues, 2016 [65] also reported that participants in the intervention group who rated their instructor as more helpful had a signi cant difference in weight loss compared to controls. This nding suggests that results may be instructor-dependent and is especially relevant in underrepresented minorities who may be less likely to identify with their facilitator. This nding also underscores the importance of reporting and evaluating staff characteristics in future research.
In our literature review, we found that MBI may be effective for cardiovascular disease risk outcomes.
However, our ndings regarding the current limitations in reported studies highlight signi cant gaps in the literature, leaving the potential e cacy of MBI in minority communities in question. In light of this, we propose several suggestions for future research. As is the case with most intervention effectiveness studies, future research on this topic should seek to elucidate attitudes, social norms, and perceived barriers to engaging in MBI as well as exploring their e cacy in real-world circumstances. These insights can help to better tailor interventions to the speci c needs of the population. Limitations should be further investigated to understand the potential role of race, self-selection, and motivation toward or identi cation with the intervention. Likewise, future studies to establish effectiveness in minority populations should seek to expand generalizability and may seek to recruit from community centers and/or churches, as supported by scholarship on intervention tailoring [75].
Our ndings highlight several potential targets for tailoring mindfulness-based interventions to cardiovascular risk disparities in racial/ethnic minorities in the US. Many studies currently center around in-person group sessions. Using this format, tailoring studies should seek facilitators who are representative of the group and who speak the language of the group being tailored to. Accessible and acceptable intervention locations should be considered. Likewise, practice groups should represent greater diversity. Acceptability to the population should be explored along with the potential role of prior knowledge, experience, perceptions, expectations, and social norms around mindfulness and meditation in self-selection and engagement with interventions. Alternatives to lengthy, intensive, in-person sessions with xed times and dates should be explored, such as in the case of app-based interventions and brief/ultra-brief formats. Limitations.
This study presents several limitations. As with all systematic reviews, this study may have missed identi cation of studies meeting eligibility criteria. This study was also limited to articles available in the English language. The inclusion of a meta-analysis was not possible given the lack of homogeneity of included studies and this study is therefore limited to the scope of a narrative review. Considering that this study does not assess intervention e cacy, the homogeneity of across variables of included studies itself does not present a limitation, though it should be stated that the intervention effects may not be equivalent across all studies. The risk factors presented here were researched independent of their role in cardiovascular risk or disease, and this paper does not explore the potential interactions between these (e.g. diet and diabetes, or diabetes and CVD).

Conclusions
This study outlines several potential targets for future research. As all included studies indicate promise for these interventions, there is cause to believe that MBI could be of bene t to minority groups at risk for cardiovascular disease, perhaps as a complement to standard forms of treatment. Further research is needed to explore acceptability, feasibility, and effectiveness in minority populations. Bibliography