Associations between dimensions of the social environment and cardiometabolic risk factors: Systematic review and meta-analysis

Aim The social environment (SE), including social contacts, norms and support, is an understudied element of the living environment which impacts health. We aim to comprehensively summarize the evidence on the association between the SE and risk factors of cardiometabolic disease (CMD). Methods We performed a systematic review and meta-analysis based on studies published in PubMed, Scopus and Web of Science Core Collection from inception to 16 February 2021. Studies that used a risk factor of CMD, e.g., HbA1c or blood pressure, as outcome and social environmental factors such as area-level deprivation or social network size as independent variables were included. Titles and abstracts were screened in duplicate. Study quality was assessed using the Newcastle-Ottawa Scale. Data appraisal and extraction were based on the study protocol published in PROSPERO. Data were synthesized through vote counting and meta-analyses. Results From the 7521 records screened, 168 studies reported 1050 associations were included in this review. Four meta-analyses based on 24 associations suggested that an unfavorable social environment was associated with increased risk of cardiometabolic risk factors, with three of them being statistically significant. For example, individuals that experienced more economic and social disadvantage had a higher “CVD risk scores” (OR = 1.54, 95%CI: 1.35 to 1.84). Of the 458 associations included in the vote counting, 323 (71%) pointed towards unfavorable social environments being associated with higher CMD risk. Conclusion Higher economic and social disadvantage seem to contribute to unfavorable CMD risk factor profiles, while evidence for other dimensions of the social environment is limited.


Introduction
Cardiometabolic diseases (CMDs), including cardiovascular diseases (CVDs) and type 2 diabetes mellitus (T2DM) are the number one cause of death worldwide (World Health Organization (WHO)).Although some CMDs are highly heritable, e.g., familial hypercholesterolemia (Ison et al.), the non-genetic nature of CMDs is reflected by the close association between lifestyle behaviors and CMD risk.In fact, lifestyle risk factors such as an unhealthy diet or physical inactivity account for more than 70% of total cardiovascular events, 80% of coronary heart disease events and 90% of incidence of T2DM (Mozaffarian et al., 2007).
Alterations of lifestyle risk factors can therefore have a major impact on the prevention of CMDs.Several landmark trials indeed showed that adhering to a Mediterranean diet (Grosso et al., 2017) or a combined lifestyle intervention (Diabetes Prevention Program Research Group, 2002;Maruthur et al., 2009) reduces the risk of CMDs substantially with up to 58% risk reduction of type 2 diabetes.These lifestyle risk factors not only influence established risk markers such as elevated blood pressure, blood lipids, and glucose-insulin homeostasis but also other pathways such as endothelial function, oxidative stress, inflammation (e.g., C-reactive protein), thrombosis/coagulation, arrhythmia, and other intermediary pathways (e.g., psychosocial stress) (Appel et al., 2005;Folsom, 2013;Man & Xia, 2020;Netz et al., 2005;Upadhyay, 2015;Vandercappellen et al., 2022;Wang et al., 2017).Indeed, clinical trials have demonstrated the effects of lifestyle interventions on CMD risk factors (Hochsmann et al., 2021;Wister et al., 2007).On the basis of population-wide benefits and minimizing adverse drug effects, changes in lifestyle are crucially important for primary prevention and may also have beneficial effects for secondary prevention.However, many individuals cannot maintain healthy lifestyle behaviors in the long term (Burgess et al., 2017;Middleton et al., 2013).
That is because many lifestyle programs and interventions do not address the upstream risk factors of unhealthy lifestyle (Lakerveld & Mackenbach, 2017), i.e., the causes-of-the-causes of CMD.One important upstream factor that drives lifestyle behaviors is the social environment, i.e., the social relationships and social context in which groups of people live and interact.The social environment encompasses several concepts organized in dimensions; however, there is no consensus in the literature on which dimensions are included in the social environment.In this work, we adopt the dimensions proposed by Kepper et al., 2019, namely: Economic and Social Disadvantage, Discrimination and Segregation, Crime and Safety, Social Cohesion and Social Capital, Disorders and Incivilities, Social Relationships and Norms, and Civic Participation and Engagement.These concepts and dimensions were found to be related to CVD risk, for example, the extent of connectedness and solidarity in a communityoften labeled "social cohesion" -has been associated with a reduced likelihood of CVDs and related risk factors, such as myocardial infarction, stroke and hypertension (Kim et al., 2013(Kim et al., , 2014;;Lagisetty et al., 2016).Social cohesion may protect against CMD through multiple pathways, including better coping abilities, healthier lifestyle behaviors and positive psychological effects.Similarly, a meta-analysis showed that poor social relationships are associated with 29% increased risk of coronary heart disease and 32% increased risk of stroke (Valtorta et al., 2016).Another important factor which can have a detrimental impact on an individual's lifestyle, and in turn on CVD health, is social isolation -the absence of social connections and interactions (Leigh-Hunt et al., 2017).Social isolation is related to feelings of loneliness, which seems to result in a chronic social stress response (Xia & Li, 2018).These feelings of stress are able to activate different mechanisms in the body which may negatively impact CVD risk.In addition, the results from a global study suggest that civic participation, such as voting, being a volunteer and participating in recreational and sporting activities "strengthens existing social networks, increases social cohesion, creates a common sense of goals and purpose, and improves overall health and wellbeing" (Kim et al., 2015).And a US study that aimed to investigate how the benefits of volunteering get "under the skin" found that middle-aged and older volunteers were less likely to have central adiposity, lipid dysregulation and hypertension than their non-volunteering peers (Burr et al., 2016).In summary, it seems that a variety of aspects related to the social environment may influence CVD risk in different ways.There is even tentative evidence that social network interventions may reduce HbA1c levels in T2DM patients (Spencer-Bonilla, Ponce, Rodriguez-Gutierrez, et al., 2017).Yet, to understand the causes-of-the-causes of CMDs, and to develop effective intervention and prevention strategies, it is important to evaluate the existing evidence on the associations between dimensions of the social environment and risk factors of CMD.Therefore, it was our aim to systematically summarize and meta-analyze the available evidence.

Methods
This is a systematic literature review of English-language scientific articles on the association between different dimensions of the social environment and cardiometabolic risk factors in adults.This systematic literature review is the result of a review protocol that was prospectively registered with the International Prospective Register of Systematic Reviews (PROSPERO) database (registration number CRD42021223035).Because the search resulted in exceptionally large numbers of relevant articles, the reporting is split according to outcome measures: this review focuses on cardiometabolic risk factors and a twin review focuses on cardiometabolic disease endpoints (Abreu et al. submitted elsewhere).This review is written according to the latest Preferred Reporting Items for Systematic Reviews and Meta-Analysis PRISMA guidelines for systematic reviews (Page et al., 2021a(Page et al., , 2021b)).

Search strategy and study selection
To identify all relevant publications, we conducted systematic searches in the bibliographic databases PubMed, Scopus and Web of Science Core Collection from inception to 16 February 2021, in collaboration with a medical information specialist (LS).We used free-text terms in all databases.For PubMed, the search terms also included indexed terms from MeSH.The search comprised a block for "cardiometabolic diseases" (e.g., heart disease, hyperlipidemia, HbA1c), for "dimensions of the social environment" (e.g., social capital, area-level deprivation), and for the "contextual level" of the social environment (e.g., community, network).The latter was added to exclude studies focusing on individual-level social factors (e.g., individual SEP).Supplementary File 1 contains the full search strategy for all electronic databases including number of results.Articles in all languages were accepted during the search.Reference lists of the articles included were manually searched for other relevant publications.
The search was performed and duplicates were removed by a medical information specialist (LS).All de-duplicated titles and abstracts retrieved from the search were screened for eligibility by at least two out of three authors independently (TCA, JDM, JWJB), according to the criteria for inclusion and exclusion, using the semi-automation tool Covidence.A pilot test in a random sample of 100 results ensured consistency among screeners.Afterwards, TCA and JDM independently assessed all full texts for inclusion.Differences in judgement were resolved through a consensus procedure.Studies were included if they met the inclusion criteria as stated below.
Original studies that examined associations between dimensions of the social environment and risk factors of CMD in adults were reviewed.The scope of this review was limited to exposures that assessed properties of the social context, i.e., social factors assessed at the environmental level (e.g., area-level income), but not those that assessed properties of individuals, i.e., individual reported happiness derived from their social network.We did however include social factors that were assessed at the individual-level but reflected a property of the social context in which an individual is inserted (e.g., one's social network size).
Studies were included for this review on risk factors of CMD or the twin review on cardiometabolic outcomes if they: (i) studied an adult population or followed children and adolescents beyond 18 years of age; (ii) studied risk factors of CMD, or incidence or prevalence of CMD outcomes; (iii) covered any measure of the social environmental that potentially influences CMD; (iv) were observational or intervention studies; and (v) were written in English.We excluded studies if they: (i) were limited to children and adolescents; (ii) studied obesity as outcome -given the recent evidence available for this outcome (Daniels et al., 2021;Glonti et al., 2016;Powell et al., 2015); (iii) studied risk factors of CMD with little or no influence of lifestyle behaviors (e.g., congenital heart disease, rheumatic heart disease, and type 1 diabetes) as outcomes; (iv) focused on treatment, medication, or management of disease outcomes; (v) were conducted in samples of patient populations or pregnant women; (vi) were health economic evaluations, simulation studies, or publications that did not report original scientific research; or (viii) studied mortality outcomes alone, or did not differentiate between morbidity and mortality outcomes.

Data extraction
Three authors (TCA, JDM and FH) performed data extraction from eligible studies according to a standardized protocol and a predefined list of variables including study and sample characteristics.Social environment factors were categorized into one of eight dimensions according to a conceptual framework (Supplementary File 2): Social Cohesion and Social Capital; Sense of Place/Belonging; Crime and Safety; Disorder and Incivilities; Discrimination and Segregation; Economic and Social Disadvantage; Social Relationships and Norms; and Civic Participation/Engagement.Outcomes were categorized into one of four categories namely "glucose metabolism-related risk factors" (e.g., HOMA, HbA1c), "metabolic and inflammatory-related risk factors" (e.g., lipid levels, CRP, cortisol), "cardiovascular health-related risk factors" (e.g., atherosclerosis, systolic blood pressure, intima-media thickness) and "CVD risk scores" (e.g., Framingham Risk Score, metabolic syndrome, allostatic load).In case of missing data on effect measures, study investigators were contacted.When available, data on sex-specific effect metrics were extracted.If relevant papers contained both separate (e.g., HbA1c, lipid levels, systolic blood pressure, etc.) and CVD risk scores (e. g., "overall risk score") in their results section, only the effect measures on combined outcomes were reported.Generally, we report on effect sizes from fully adjusted statistical models except when the fully adjusted model was corrected for lifestyle behaviors (e.g., diet, physical activity, alcohol or smoking).In this case, the associations from the model without lifestyle factors were reported because we hypothesized lifestyle behaviors to be intermediary rather than confounding variables.

Quality assessment
Three authors (TCA, JDM and FH) assessed the quality of all studies included.Disagreements were resolved by consensus.The Newcastle-Ottawa Scale (NOS) was used to assess cohort studies (Wells et al.) (Supplementary File 3).An adapted version of the NOS was used for the quality assessment of cross-sectional studies (Herzog et al., 2013) (Supplementary File 4).Cohort studies were able to gain a total of 9 points based on 8 items, whereas cross-sectional studies could gain a total of 10 points based on 7 items.The assessment was divided into three domains namely selection, comparability and outcome.In order to get comparable quality ratings of all studies, we calculated the percentage of the maximum number of points a paper could gain.Ratings reflect the methodological quality of the associations between social environmental factors and risk factors of CMD, even if this was not the primary research question of the study.Low quality studies were defined as those who received less than 50% of all possible points.Prior to the quality assessment of included studies, the process of quality assessment was piloted in a random subsample of 10 studies.

Data synthesis
Data synthesis was conducted for combined exposure categories (eight social environment dimensions) and combined outcome categories (four risk factor of CMD categories).In accordance with the Cochrane Handbook for Systematic Reviews of Interventions (Higgins et al., 2022), extracted data were synthesized where possible with two approaches namely meta-analysis and vote counting for studies that had a medium or high quality score.For both methods, associations from low quality studies and associations including an exposure measure from the dimension "Discrimination and Segregation" (e.g., proportion of migrants in the neighborhood) were excluded."Discrimination and Segregation" was excluded because in general exposure measures used by authors were poorly defined, highly heterogeneous in operationalization, and therefore not comparable across different studies.
Criteria for associations to be included in meta-analyses were: i) three or more associations available per combination of social environment dimension and CMD risk factor category; ii) effect estimates reported as ratios (e.g., odds ratios, relative risk, and hazard ratios); iii) having variance measures; and iv) exposures being operationalized as dichotomous or categorical variables.In case of categorical exposures, we compared the two extreme categories (e.g., highest vs. lowest deprivation).We did not meta-analyze associations on continuous outcomes or continuous or incremental exposures due to methodological challenges in converting continuous data into categorical data, standardizing scales/scores, and in pooling together different effect estimates.We performed random-effects meta-analysis, which accounted for the multilevel structure of the data as many studies reported on more than one association of interest.All models where based on a t-distribution as recommended by the guidelines of R package used (Viechtbauer, 2010).The reference category across all exposures was harmonized and defined as the group with the most favorable social environment.Subgroup analyzes were performed for sex-specific effect measures when sufficient data was available.We aimed to perform sensitivity analyzes according to the studies' country income level (World Bank, 2021) but were unable to given the absence of sufficient studies from low-and middle-income countries.We also aimed to run sensitivity analyzes with odds ratio effect estimates only as opposed to all ratios combined, but there were not sufficient associations.Forest plots were generated for each meta-analysis performed and heterogeneity was assessed with I 2 statistics accounting for the dependency among associations originating from the same study.Results are expressed as odds ratio and 95% confidence interval (OR, 95% CI).Analyzes were performed in R version 4.2.1 (Team, 2020), using the functions rma.mv and forest of the Metafor package (Viechtbauer, 2010).
In addition to the meta-analysis, we used vote counting based on the direction of effects as an alternative method to synthesize the available evidence (McKenzie et al.).This method categorizes and compares the number of associations showing that unfavorable social environments are associated with higher risk of cardiometabolic risk factors and the number of associations showing that unfavorable social environments are associated with lower risk of cardiometabolic risk factors.Criteria for associations to be included in vote counting were: i) being overall, rather than sex-specific associations; and ii) we could establish a direction of effect such that "unfavorable social environment" was associated with either a higher, equal or lower risk of cardiometabolic risk factors.In accordance with Cochrane's recommendations (McKenzie et al.), statistical significance nor effect size were considered in the categorization.Results of vote counting are presented in a direction-of-effect plot.

Results
After screening 7521 titles and abstracts, 555 articles proceeded to full-text screening (Supplementary File 5).Of the 333 included articles, 208 were included in a review on hard outcomes reported elsewhere (Abreu et al. submitted elsewhere) and 168 are included in this review (some articles included both hard outcomes and risk factors of CMD) (see Fig. 1).
The large majority of studies were conducted in high income countries (95%), were cross-sectional in nature (74%) and were published in the last 10 years (64%) (Table 1).No qualitative nor experimental studies were included.The 168 included articles together described 1050 associations that were relevant for this review, of which 576 (49%) were overall associations, 317 (27%) were female-specific and 271 (23%) were male-specific (Supplementary Table 1).
The average percentage of quality assessment points scored was 57% for cross-sectional and 79% for longitudinal studies.Many crosssectional studies scored low on the "selection" domains "sample size" and "non-respondents" (see Fig. 2a).Cohort studies generally scored low on the "selection" domains "representativeness" and "correction for presence of the outcome at baseline" (see Fig. 2b).Of all studies included, 20 (12%) were considered low quality, which were subsequently excluded from vote counting and meta-analyses.
Of the 1050 unique associations, 167 were from low quality studies, 377 were sex-specific associations and for 43 the direction of effect could not be determined.Of the 458 associations included in the vote counting, 323 (71%) pointed towards unfavorable social environments being associated with higher risk of cardiometabolic risk factors (see Fig. 3).Of the four cardiometabolic risk factors categories considered, most associations used "CVD risk scores" such as Cumulative Biological Risk or a cardiovascular health risk score as outcome (n = 175, 38%), followed by "cardiovascular health-related risk factors" such as systolic and diastolic blood pressure (n = 164, 36%), "glucose metabolism-related risk factors" such as HbA1c or HOMA (n = 90, 20%), with least associations for "metabolic and inflammatory risk factors" such as CRP or IL-6 (n = 29, 6%).Two-hundred-sixty-two of the 458 associations (57%) were on the social environment dimension "Social and Economic Disadvantage", which consistently showed that being exposed to more social and economic disadvantage is associated with less favorable cardiometabolic risk factors profiles.Similar consistent associations were found for the domain "Crime and Safety": 34 out of 41 associations (83%) suggested that higher levels of crime and less safety was associated with higher risk of cardiometabolic risk factors.Associations in the dimension "Social Relationships and Norms" were considerably less consistent, with only 60 out of 121 associations pointing towards the direction of       disadvantageous social relationships and norms being associated with less favorable cardiometabolic risk factors profiles.There was little evidence for associations of "Social Cohesion and Social Capital" (16 associations) and "Civic Participation and Engagement" (18 associations).There were no eligible associations from the dimensions "Sense of Place/ Belonging" and "Disorder and Incivilities".Meta-analysis was possible for only 24 associations across four exposure-outcome combinations that had a dichotomous outcome and categorized exposure measure.As such, only one out of eight social environmental domains was represented in the overall meta-analyses (see Fig. 4, Supplementary Fig. 1 and Table 2).In all four metaanalyses "Economic and Social Disadvantage" was associated with increased cardiometabolic risk, but one was not statistically significant.Individuals that experienced more economic and social disadvantage had higher "CVD risk scores" (OR = 1.54, 95%CI: 1.35; 1.84), higher "glucose metabolism-related risk factors" (OR = 1.91, 95%CI: 1.56;   2.32) and somewhat higher "cardiovascular health-related risk factors" (OR = 1.06, 95%CI: 1.00; 1.12).Only one of four meta-analyses had high heterogeneity, namely for the outcome "metabolic and inflammatory-related risk factors" with six associations included (I 2 = 97%).It was possible to perform four sex-specific meta-analyses for the same combination of exposure-outcome for both men and women (Figs.5a, b, 6a and 6b and Supplementary Table 2).Two of those, namely "Economic and Social Disadvantage" with "cardiovascular health-related risk factors" and with "CVD risk scores", were also performed with overall estimates and therefore could be compared (Fig. 4, Supplementary Fig. 1 and Table 2).The association between "Economic  and Social Disadvantage" and "cardiovascular health-related risk factors" was similar for the sex-specific meta-analysis as for the overall meta-analysis.However, whereas exposure to more economic and social disadvantage was overall associated with higher "CVD risk scores" (OR = 1.54, 95%CI: 1.35; 1.84, Table 2), this was not the case for men (OR = 0.59, 95%CI: 0.05; 6.58) and women (OR = 0.98, 95%CI: 0.34; 2.78) separately.It is important to note that the extremely wide confidence intervals and high heterogeneity for the OR for "Economic and Social Disadvantage" and "CVD risk scores" among men suggests there is serious imprecision in this pooled effect estimate.Two of the sex-specific meta-analyses covering "Social Relationships and Norms" could not be compared with overall estimates.None of these sex-specific meta-analyses were statistically significant and direction of associations differed both between outcomes and men and women (see Fig. 6a and b and Supplementary Table 2).For example, more favorable social relationships and norms were associated with lower "CVD risk scores" for men (OR = 0.91, 95%CI: 0.51; 1.60) and with higher "CVD risk scores" for women (OR = 1.26, 95%CI: 0.88; 1.80).

Discussion
This systematic review and meta-analysis including a total of 168 studies reporting on 1050 associations indicates that unfavorable social environments are associated with higher cardiometabolic risk factors.Most associations focused on aspects of "Social and Economic Disadvantage" such as area-level deprivation, for which the evidence consistently pointed towards an adverse effect on cardiometabolic risk factors.Sex-specific associations showed inconsistent results, with both similarities and differences between associations for men and women, as well as similarities and differences in comparison with overall effect estimates.For other aspects of the social environment such as "Civic     Participation and Engagement" limited evidence was found.Through this work we provide the most comprehensive overview to date of the literature on the social environmental determinants of cardiometabolic risk factors.
Our twin review on hard CMD outcomes (Abreu et al. submitted elsewhere) similarly showed that being exposed to a worse social environment is consistently associated with increased risk of CMD, with most evidence available for the dimension "Economic and Social Disadvantage".The dominance of area-level disadvantage has also been observed in other reviews that considered a range of social environmental dimensions (Kepper et al., 2019;Suglia et al., 2016).The second-most studied social environmental dimension in this review was "Social Relationships and Norms", with much less evidence for other social environmental dimensions.
It is challenging to compare the direction and strength of associations across the four categories of CMD risk factors we used in our data synthesis.Meta-analyses suggest that adverse social and economic circumstances are more strongly associated with "glucose metabolismrelated risk factors" than with "CVD risk scores" or "cardiovascular health-related risk factors", but this was based on a small number of associations only.Vote counting also demonstrate that the large majority (88%) of studies using "glucose metabolism-related risk factors" as outcome found associations in the expected direction, which was a stronger indication for the direction of the evidence than for "cardiovascular health-related risk factors" (67%), "CVD risk scores" (68%) and "metabolic and inflammatory-related risk factors" (52%).
Our results are partly in line with findings from previous reviews and meta-analyses that studied specific dimensions of the social environment or specific cardiometabolic risk factors.The results from another systematic review (Lovasi et al., 2009) suggest that neighborhood safety might be an important factor in decreasing obesity in more disadvantaged populations.We were unable to meta-analyze the association between the social environment dimension "Crime and Safety" in relation to "cardiovascular health-related risk factors".However, the vote counting suggests that increased neighborhood crime is negatively associated with "cardiovascular health-related risk factors" and "CVD risk scores" outcomes, which is in line with earlier studies (Lovasi et al., 2009).
A meta-analysis including 21 studies found that living in neighborhood of low socioeconomic status was associated with a 31% higher odds of overweight and 45% higher odds of obesity compared to living in a neighborhood of high socioeconomic status (Mohammed et al., 2019).This study did unfortunately not stratify for sex while we found differences between the overall and sex-specific analyses.Sex differences have also been observed for associations between neighborhood environments and health (Macintyre et al., 2005), whereby worse economic indicators, like unemployment, on neighborhood level were associated with significantly better self-rated health in women but not in men.The authors suggest that this may be due to either women spending more time in their local neighborhoods than men, or a greater vulnerability to adverse neighborhood features among women.These explanations might help understand the large difference in the overall association on "Social and Economic Disadvantage" and "CVD risk scores" and the sex specific associations.While we were unable to include studies pertaining to "Segregation and Discrimination" at a contextual (i.e., community) level in our vote count and meta-analyses, a review on individual-level racial discrimination found a small but positive association with hypertensive status but not with resting blood pressure or diastolic blood pressure (Dolezsar et al., 2014).It needs to be noted that for some aspects of the social environment it is difficult to distinguish between individual-level and contextual dimensions of the social environment, and this is especially true for aspects such as discrimination.At the same time, it is challenging to operationalize social environment measures as contextual variables generally, which has been discussed in detail elsewhere (Oberndorfer et al., 2022).
Our vote counting results are also in line with a recent meta-analysis by Uchino et al. (Uchino et al., 2022) who found that perceived social support as individual-level construct was not significantly associated with ambulatory blood pressure.In another meta-analysis of Uchino et al. (Uchino et al., 2018), better social support and social integration were associated with lower levels of inflammatory cytokines.We were unable to reproduce a similar finding in the vote counting and stratified analyzes and even observed elevated, albeit non-significant, inflammatory values for men.However, it should be noted that Uchino et al. (Uchino et al., 2018) included 47 associations in their meta-analyses whereas we only included three due to the decision to only include associations with dichotomized exposures and outcomes.It is noteworthy that our vote counting results also points towards an unfavorable association of "Social Relationships and Norms" with "metabolic and inflammatory-related risk factors".This may be attributed to the fact that there may be a ceiling effect in the beneficial effects of the number of social relations, i.e., after a certain number of social relations is reached, one extra social relation does not result in reduced risk.

Strengths & limitations
The findings of this review and meta-analysis should be seen in the light of its limitations.First, due to the heterogeneity in social environment factors and the complexity of harmonizing continuous data, the meta-analysis was only limited to dichotomous outcomes.This resulted in a small number of associations that could be included in the metaanalyses which limits the generalizability of our results.An approach to pool different measures would be the standardized mean difference (SMD), which converts the results of the associations in a standardized measure before they can be combined in a unitless measure of pooled results, of which the disadvantage is its interpretability.
To reduce the impact of this limitation in our results, we performed complementary vote counting, which enabled us to give a visual and inclusive summary of the data.However, vote counting provides no information on the magnitude of the associations, does not account for differences in relative sizes of studies and is less powerful than metaanalysis (Borenstein et al., 2009).Secondly, as also observed in other reviews (Kepper et al., 2019;Mohammed et al., 2019), the high heterogeneity across studies in terms of measurement of exposures and outcomes, and adjustment for covariates hampered the comparison of retrieved data.Therefore, we advise readers to reflect on the results of the meta-analysis and its limitations in combination with the vote counting findings.Third, much of the data was from cross-sectional studies, limiting causal inference.This is particularly relevant for studies assessing contextual exposures, where observational studies may suffer from amongst others selection bias (i.e., an individual's choice to reside in a certain area is related to the study outcome).Finally, almost all studies were conducted in high-income countries.This limits the ability to generalize the results to low-income countries.We were also unable to stratify study results by age group.It is thinkable that older people spend more time in their neighborhood than younger, employed people and older people more often experience feelings of social isolation/loneliness than young and middle-aged adults (Luhmann & Hawkley, 2016).
The main strength of this review is its broad scope.We employed a thorough and broad search across three large databases, assisted by an information specialist.In this way, we were able to capture all available relevant evidence and provide a comprehensive overview of the existing literature on this topic.In addition, we performed a backward reference check to complement this thorough search.Moreover, following the call for more consistency in the social environment literature (Kepper et al., 2019), we combined all social environment factors that were categorized into the same dimension for the purposes of synthesis of findings.

Implications for practice, policy and research
This review focusing on the general adult population suggests that unfavorable social environments are associated with higher cardiometabolic risk.Other reviews have demonstrated the importance of favorable social environments for the treatment and management of CMD.For instance, Schram et al. showed that smaller network size and less social support was associated with increased risk of diabetes complications (Schram et al., 2021), and Spencer-Bonilla et al. found promising tentative effects of social network interventions on glycemic control and quality of life in T2DM patients (Spencer-Bonilla, Ponce, Rodrigeuz-Gutierrez, et al., 2017).It is therefore no surprise that there is increasing attention for the Social Determinants of Health (World Health Organization) -including social environmental aspects -among healthcare providers.In response to this development, White-Williams et al. suggest a number of conceptual models and screening tools that healthcare providers could use to consider the role of social determinants of health in the treatment of patients with heart failure (White- Williams et al., 2020).One of the best practices considered is having a list of availability community resources available in the healthcare clinic (Davidson & McGinn, 2019).
Given the evidence for links between the social environment and other health outcomes such as frailty (Duppen et al., 2019), mental health (Bjorlyhaug, Karsson, Kim Hesook, & Kleppe, 2021;Breedvelt et al., 2022), and general health (Ehsan et al., 2019;Pérez et al., 2020), the importance of the social environment goes beyond its effects on cardiometabolic health.Other studies have shown the impact of social isolation and loneliness on CVD risk through chronic stress (Leigh-Hunt et al., 2017;Rentscher et al., 2022;Xia & Li, 2018).The adverse effects of social isolation and societal polarization on health should therefore not be underestimated and taken into account in deliberations around policies such as Covid-19 restrictions (Lippke et al., 2021).
Yet, how exactly social relations may protect against CMD remains to be explored as this review showed inconsistent results between the dimension "Social Relationships and Norms" and cardiometabolic risk factors.This requires a better conceptualization and measurement of the aspects of social relationships that may be beneficial or harmful to health.The inconsistent terminology and conceptualization of the dimensions of the social environment has also been highlighted by other authors (Barnett & Casper, 2001;Braveman & Gottlieb, 2014;Kepper et al., 2019).It may also be important to specifically consider sex effects of social relationships, given the findings in this review.Another aspect to consider in future research is the dilution of the effects of offline social relationships by online relationships.Indeed, studies into the effects of the online social environment on CMD risk are rare but should be the topic of future research given the increasing use of social media and other online platforms that allow for interpersonal interactions.
While we aimed to capture the totality of social environment exposures and their association with CMD risk factors, most studies only studied a single aspect of the social environment in a cross-sectional setting or with modest follow-up at most.This hampers the estimation of the total "life course" (Ben-Shlomo et al., 2014) or "exposome" (Beulens et al., 2022) influence of the social environment on CMD risk and future studies would benefit from the integration of multiple dimensions of the social environment over the life course, and their combined effects on CMD risk.
Finally, referring back to the conceptual framework we used as the basis for this systematic review (Abreu et al. submitted elsewhere), future studies should assess the extent to which dimensions of the social environment explain the impact of structural socioeconomic factors on health outcomes.Indeed, there is a very strong link between poverty and ill health (Marmot et al., 2020) and part of this association may be explained by the adverse social environments individuals become exposed to when living in poverty.For example, in a cross-European study, we demonstrated that neighborhood-level social capital explained large parts of the association between neighborhood income inequality and BMI (Mackenbach et al., 2016-a).Regardless of the extent to which the links between structural socioeconomic factors and CMD is explained by the dimensions of the social environment, it is likely that both of those aspects should be targeted through upstream policies to improve population level cardiometabolic health.
In conclusion, the findings from vote counting and meta-analyses suggest that exposure to adverse social environments is associated with unfavorable cardiometabolic risk factors profiles.The evidence for the dimension "Economic and Social Disadvantage" is most robust while other dimensions of the social environment such as "Civic Engagement and Participation" require more evidence from well-designed prospective studies.

Fig. 2a .
Fig. 2a.Quality of cross-sectional studies included based on the New-Ottawa Castle tool Quality Assessment Scale adapted for cross-sectional studies, review on social environmental determinants of cardiometabolic risk factors.

Fig. 2b .
Fig. 2b.Quality of cohort studies included based on the New-Ottawa Castle tool Quality Assessment Scale, review on social environmental determinants of cardiometabolic risk factors.

Fig. 3 .
Fig. 3. Overview of vote counts based on direction of effect for associations between dimensions of the social environment and cardiometabolic risk factors.

Fig. 4 .
Fig. 4. Summary of forest plots (random-effects model) for the meta-analyses of social environment dimensions and cardiometabolic risk factors.

Fig. 5a .
Fig. 5a.Summary of forest plots (random-effects model) for the meta-analyses of economic and social disadvantage and cardiometabolic risk factors among men.

Fig. 5b .
Fig. 5b.Summary of forest plots (random-effects model) for the meta-analyses of economic and social disadvantage and cardiometabolic risk factors among women.

Fig. 6a .
Fig. 6a.Summary of forest plots (random-effects model) for the meta-analyses of social relationships and norms and cardiometabolic risk factors among men.

Table 1
Characteristics of included studies, review on social environmental determinants of cardiometabolic risk factors.
(continued on next page) T.C.Abreu et al.

Table 1
(continued ) (continued on next page) T.C.Abreu et al.

Table 1
(continued ) (continued on next page) T.C.Abreu et al.

Table 2
Summary pooled effects and between-study variance estimates with 95% confidence intervals from meta-analyses covering the social environmental determinants of cardiometabolic risk factors.