The impact of greenspace or nature-based interventions on cardiovascular health or cancer-related outcomes: A systematic review of experimental studies

Significance Globally, cardiovascular disease (CVD) and cancer are leading causes of morbidity and mortality. While having different etiologies, CVD and cancer are linked by multiple shared risk factors, the presence of which exacerbate adverse outcomes for individuals with either disease. For both pathologies, factors such as poverty, lack of physical activity (PA), poor dietary intake, and climate change increase risk of adverse outcomes. Prior research has shown that greenspaces and other nature-based interventions (NBIs) contribute to improved health outcomes and climate change resilience. Objective To summarize evidence on the impact of greenspaces or NBIs on cardiovascular health and/or cancer-related outcomes and identify knowledge gaps to inform future research. Methods Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 and Peer Review of Electronic Search Strategies (PRESS) guidelines, we searched five databases: Web of Science, Scopus, Medline, PsycINFO and GreenFile. Two blinded reviewers used Rayyan AI and a predefined criteria for article inclusion and exclusion. The risk of bias was assessed using a modified version of the Newcastle–Ottawa Scale (NOS). This review is registered with PROSPERO, ID # CRD42021231619. Results & discussion Of 2565 articles retrieved, 31 articles met the inclusion criteria, and overall had a low risk of bias. 26 articles studied cardiovascular related outcomes and 5 studied cancer-related outcomes. Interventions were coded into 4 categories: forest bathing, green exercise, gardening, and nature viewing. Outcomes included blood pressure (BP), cancer-related quality of life (QoL) and (more infrequently) biomarkers of CVD risk. Descriptions of findings are presented as well as visual presentations of trends across the findings using RAW graphs. Overall studies included have a low risk of bias; and alluvial chart trends indicated that NBIs may have beneficial effects on CVD and cancer-related outcomes. Conclusions & implications (1) Clinical implication: Healthcare providers should consider the promotion of nature-based programs to improve health outcomes. (2) Policy implication: There is a need for investment in equitable greenspaces to improve health outcomes and build climate resilient neighborhoods. (3) Research or academic implication: Research partnerships with community-based organizations for a comprehensive study of benefits associated with NBIs should be encouraged to reduce health disparities and ensure intergenerational health equity. There is a need for investigation of the mechanisms by which NBIs impact CVD and exploration of the role of CVD biological markers of inflammation among cancer survivors.


Objective
To summarize evidence on the impact of greenspaces or NBIs on cardiovascular health and/or cancer-related outcomes and identify knowledge gaps to inform future research.

Methods
Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 and Peer Review of Electronic Search Strategies (PRESS) guidelines, we searched five databases: Web of Science, Scopus, Medline, PsycINFO and GreenFile. Two blinded reviewers used Rayyan AI and a predefined criteria for article inclusion and exclusion. The risk of bias was assessed using a modified version of the Newcastle-Ottawa Scale (NOS). This review is registered with PROSPERO, ID # CRD42021231619.

Introduction
Cardiovascular disease (CVD) is the leading cause of global morbidity and mortality [1,2]. In 2019, CVD accounted for approximately 18.6 million deaths globally [3]. In the 2020 Lancet global burden of disease (GBD) report, ischemic heart disease (IHD) and stroke, both CVD, were the top-ranked causes of disability adjusted life years (DALYs) in both 50-74 years and 75 years and older age groups [4]; and respectively responsible for 16% and 11% of the total global deaths, in 2019 [2]. In the US, 126.9 million adults had some form of CVD from 2015 and 2018 [3]. Costs associated with CVD from 2016 to 2017 totaled $363.4 billion ($216.0 billion in direct costs and $147.4 billion in lost productivity due to morbidity or mortality) [3]. In addition to the CVD burden, cancer was the sixth leading cause of global mortality in 2019, and a significant contributor to global morbidity [2]. Further, in 2020 alone, 19.3 million new cancer cases were diagnosed; and this number is expected to become 28.4 million cases in 2040, a 47% rise from 2020 [5]. There were almost 10.0 million cancer deaths in 2020 [5]. In 2017, the financial burden of cancer in the US was approximately 1.8% of gross domestic product or nearly $ 350 billion [6]. The cancer-related healthcare cost was $161.2 billion while the cost associated with premature mortality was $150.7 billion; and the cost of productivity loss from morbidity was $30.3 billion [6]. CVD and cancer have close co-morbid linkages due to multiple shared risk factors [7], which put cancer survivors at a disproportionate risk for CVD [1,8]. CVD and cancer are closely linked in a bidirectional causal relationship whereby having one of the diseases puts the patient at an increased risk of having the other [8,9]. With multiple common risk factors such as obesity, smoking, and inadequate or low physical activity (PA), co-occurrence of both diseases is a major clinical problem [8]. Each disease affects the treatment of the other, and therefore, has a detrimental impact on individual's quality of life (QoL) and survival [8]. For neighborhood disadvantages increased CVD risk in a socioeconomically diverse sample of African Americans [43]. For each standard deviation increase in neighborhood disadvantage, CVD risk increased by 25% (hazard ratio = 1.25; 95% confidence interval (CI) = 1.05, 1.49) [43]. In the Dallas heart study, a multilevel regression analysis with a sample of 1174  year-olds); found that residing in more deprived neighborhoods was significantly associated with increased BP and incidence of HTN over time during a 9-year period [51]. Individuals living in more deprived neighborhoods had 1.69 times greater odds of developing HTN (OR = 1.69, 95% CI 1.02, 2.02) [51]. Further, in another study, authors used census tract data to investigate the relationship between a 10-year change (1990 to 2000) in neighborhood SES and mortality among 288,555 participating individuals, aged 51-70 years, who enrolled in the National Institutes of Health-AARP Diet and Health Study in 1995-1996 (baseline) and did not move during the study [50]. Mortality data were assessed by linking census tract data to the Social Security Administration Death Master File between 2000 and 2011. Improvement in neighborhood SES was associated with a lower mortality rate, while SES deterioration was associated with a higher mortality rate for both cancer and CVD [50].
Neighborhood built or social environments have been linked with cancer outcomes [18] through multiple studies. In their "Multi-level Biological and Social Integrative Construct (MBASIC)" framework, Lynch and Rebbeck integrated macro-environment (i.e.: health care policy, neighborhood, or family structure), individual factors (i.e.: behaviors, carcinogenic exposures, socioeconomic factors, and psychological responses) and biological factors (i.e.: cellular biomarkers and inherited susceptibility variants) to represent the multifactorial and complex nature of cancer etiology [52]. This model has been deemed essential in cancer etiology research [18]. Subsequent research has linked poor neighborhood built and social environments to adverse health outcomes across the entire cancer control continuum including cancer risk [53,54], cancer incidence [55,56], cancer diagnosis [57], cancer treatment [58], cancer survivorship [59], cancer survival [57,60], and cancer mortality [18,61].
In addition to poorly built or social neighborhood environments, global climate change is also adversely impacting health, including poorer CVD and cancer outcomes [62,63]. Extensive literature reviews suggest that increased temperature is associated with higher extreme weather events-related morbidity and mortality, particularly cardiovascular (CV) and respiratory events [64,65]. The higher burden of warmer temperatures on CV health includes increased risk of myocardial infarction (MI) [66] and mortality for IHD in North America [67]. A 2008 study found that for every increase of 4.7˚C in mean daily temperature, there was a 2.6% increase in CV mortality in California (95% confidence interval (CI): 1.3, 3.9) [67].
Greenspace is a major component of the built neighborhood environment and has been linked with increased neighborhood property values [68][69][70]. Additionally, greenspace has been linked with many positive health outcomes [71], including lower odds of being overweight or obese, a major risk factor for both CVD and cancer [41]. Some of empirically investigated benefits of greenspace on CV health include increased angiogenic capacity [72], reduced CVD risk [17,73], decreased CVD morbidity [74], and decreased CVD mortality [19,75,76]. Similarly, some of the benefits of greenspace on cancer outcomes include enhanced cancer prevention initiatives [77,78], reduced cancer incidence [78,79], improved cancer survivorship [78,80], and reduced prostate cancer mortality [81]. Additionally, greenspace helps in sequestering carbon and contributing to greenhouse gases reduction, therefore is a viable intervention for the adverse impacts of climate change on both environmental and human health [82].
There is growing literature evidence on the impact of greenspace on improving clinical outcomes in CVD and cancer patients through different interventions such as "park prescription" programs and other nature-based interventions (NBI) [83][84][85][86]. Some of this evidence was found through experimental studies, suggesting possible causal relationships, and opportunities for specific interventions to improve CVD and cancer-related health outcomes. However these experimental studies have not yet been systematically reviewed to bring all existing evidence together [1]. In this review, we sought to systematically summarize findings from experimental studies with greenspace interventions and identify potential literature gaps for future research. We use an expanded definition of greenspace exposure that include forest bathing, nature viewing, nature visit, parks visits, gardening, etc. We conducted a systematic review of studies that have investigated the impact of greenspace or NBI on two main health outcomes: CVD and cancer. CVD outcomes include morbidity and mortality across different CVD conditions. Cancer-related outcomes include different measures across the cancer control continuum including cancer risk, prevention, detection, diagnosis, treatment, survivorship, end of life or mortality, as well as cancer-related QoL.

Methods
This review followed a pre-defined protocol that was developed following the preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) statement and checklist [87,88]; and was pre-registered with PROSPERO, ID # CRD42021231619. This review then followed the PRISMA 2020 reporting guidelines [89]. The PRISMA chart is illustrated in Fig 1; and the PRISMA 2020 27-items checklist is annexed in Appendix A.

Literature search
A comprehensive literature search was developed in collaboration with a medical librarian and peer reviewed using the Peer Review of Electronic Search Strategies (PRESS) guideline [105]. The following citation databases were searched on March 10 th , 2021: Web of Science, Scopus, Medline, APA PsycINFO, and GreenFile. Searches were limited to articles written in English. Databases were chosen because we sought to include all citation databases of peer-reviewed literature with comprehensive citation data for many different academic disciplines (Web of Science), source neutral literature curated by independent subject matter experts (Scopus), medical sciences from the National Library of Medicine's bibliographic database (Medline), literature in the field of psychology (PsycINFO) and literature focused on nature or greenspace (GreenFile). Search strategies were created using medical subject headings (MeSH) and keywords combined with database-specific advanced search techniques. MeSH terms and keywords were identified to represent greenspace interventions, CVD, and cancer. Keywords related to greenspace or NBI (i.e.: park prescription, wilderness therapy, forest bathing, forest therapy, green exercise, etc.), CVD outcomes (i.e.: heart failure (HF), stroke, coronary artery disease (CAD), MI, cardiac arrest, major adverse CV event (MACE), etc.) and cancer-related outcomes (i.e.: cancer prognosis, cancer incidence, cancer mortality, etc.). A full search strategy is annexed in Appendix B.
A total of 2565 results from literature searches (Medline: 348, PsycINFO: 68, Scopus: 1161, Web of Science: 972 and Greenfile: 16) were downloaded into EndNote where duplicate articles (n = 1126) were removed. 1439 unique publications were uploaded into Rayyan AI, an online tool for systematic review [90,91] available at https://www.rayyan.ai/. The web app facilitated article screening and eased collaboration between two independent reviewers.

Article selection process
The following PICO framework [92] of inclusion and exclusion criteria was followed: P ( 2. Cancer-related outcomes include lifestyle changes (i.e., gardening continuation after intervention) and QoL during cancer survivorship, and cancer outcomes (i.e.: cancer prognosis, cancer incidence, cancer mortality, etc.). We used the National Cancer Institute definition of cancer survivorship in defining the cancer survivor's population which proposes that survivorship starts the first time the patient was told by a healthcare provider that they have cancer until the end of life [96].
Since the overall goal of the review is to look at the impact of interventions on outcomes, Using Rayyan, search results were systematically screened by two reviewers (J.C.B and J.S. B) to determine eligibility. Reviewers first screened articles' titles against eligibility criteria, excluding any article that did not clearly meet the PICO criteria by reading articles' titles. Conflicts were resolved and the process was repeated, screening full abstracts, and then article's methods section. If a conflict could not be resolved between the two reviewers, a third mediator (KMMB) was consulted. Finally, the reference lists of all included articles were screened to identify relevant publications not retrieved by electronic database searches.

Data extraction and reporting
Extracted data are summarized in Tables 3 and 4 and include: (1) Studies' geographical information (City, state, country); (2) Studies' urbanicity setting (rural, semi-urban, or urban) where applicable; (3) type of greenspace or nature-based interventions + controls description where applicable, (4) assumptions made or hypotheses; (5) Measures of any CVD related outcome (incidence, morbidity, or CVD related mortality); (6) Measures of any cancer-related outcome (anything from the cancer control continuum, cancer-related quality of life (QOL), or cancer-related mortality; (7) cancer type under investigation (specific or any type); (8) Covariates adjusted for including (a) individuals level variables such as demographic information (when available); socioeconomic information (when available); comorbidity information (when available); and (b) neighborhood factors (when available) such as social environment factors, and other neighborhood-built environment characteristics; (9) Statistical analyses conducted; (10) Studies strengths and weaknesses". We used the following information to create alluvial charts as a visual representation of trends across studies by outcomes of interest, a method that was previously used in previous systematic reviews [ Beneficial effect, or no statistically significant difference between control and experimental groups: Not significant; or whether beneficial changes were observed in the control group instead of in the interventional or experimental group: Significant in controls.
Two excel datasets used to create alluvial charts for (1) CVD, and (2) cancer-related outcomes are respectively annexed in Appendices C1 and C2.

PRISMA 2020 chart illustrating our articles screening process
From 2,565 articles initially retrieved from database searches, 31 articles meeting our predefined criteria remained after screening, as illustrated in Fig 1. At the abstract screening stage, 45 articles were excluded because they did not meet at least one of our pre-defined inclusion criteria. Each one of the excluded studies was either not experimental, or not looking at one of the outcomes of interest.

Risk of bias assessment
The risk of bias for 31 included studies was assessed using a modified version of Newcastle-Ottawa Scale (NOS). Two reviewers (J.C.B. and J.S.B.) independently assessed articles on eight pre-defined items including representativeness of exposed cohort, similarity of cohorts' origins, similarity of exposed vs non-exposed cohorts (compatibility), ascertainment of exposure, baseline differences, outcome assessment, exposure duration (enough to observe outcome), Item assessment description: Representativeness of exposed group: One star was given if the study population reflected the title or abstract of the article (i.e., the group is representative (or somewhat representative of the community average). For example, a study that only used male subjects, but the title/abstract did not specify that the 'community' was males (leaving room for confusion), did not receive a star. However, a study that said in the title they were assessing results in a "population of healthy young males" and then used healthy young males, did receive a star. Non-exposed group: One star was given if two groups (exposed or not exposed) were drawn from the same population. Half of a star was given if the same group served as the control group (on a different day or time in which they were not exposed to the intervention). No star was given if it was a simple pre-exposure and post-exposure measurement with no control group OR if the two groups (exposed or not exposed) were not drawn from the same population.
Ascertainment of exposure: One star was given for studies where participants were randomly assigned to be in control or exposure group. Half of a star was assigned if the same sample was the control one day then the experiment another day or time, or if the two groups were similar but not random. No star if there was no control group.
Baseline difference: One star was given if there was a control group, and there was no baseline difference. Half a star was given if the same group served as their own controls, by repeating the experiment twice, once with exposure and once without (as a control) and there was no difference at baseline. No star given if there was no control group, there were differences between the group at baseline, or if this was not reported.
Compatibility and controlling factors between groups: If the study design controlled for two or more factors in both groups that may have impacted the outcome (i.e., diet, caffeine, sleep) they were given 2 stars. If they had two compatible groups but controlled for only one or no factors, they were given one star. No star was given for simple preexposure/postexposure tests with no control group.
Outcome assessment: One star was given if the study clearly defines outcomes and how they were assessed.
Exposure duration: One star was given if raters perceive that exposure duration was long enough to observe differences in outcomes. Cohorts follow up: One star was given if all subjects were followed up until completion or if there if raters perceive the number of subjects lost to follow as small enough to not introduce any bias https://doi.org/10.1371/journal.pone.0276517.t002 The control group was set to run without being exposed to any visuals.
There may be a synergistic benefit to engaging in physical activities while being directly exposed to nature.

PLOS ONE
and follow up after greenspace intervention. Two assessors discussed discrepancies between scores until a consensus was reached through a joint re-evaluation of the article, a method that has been used in previous studies [98]. The process resulted in a maximum of 9 possible points for each article; whereby 9 points represents the least risk of bias, and the risk of bias went up as the score went down. Following a cut-off point used in previous studies, score equal or greater than 5 was considered as "low-risk of bias" while score below 5 was considered as representing a high-risk of bias [99]. Our assessment suggested that 21 out of 31 studies (68%) had a low risk of bias; and the overall average score for all studies combined suggest a low risk of bias with a modified NOS score of mean (±SD) = 6.0 (±1.8). The risk of bias assessment is summarized in Table 2.

Summary characteristics of 31 articles included in the review
Data from included studies is summarized in two tables (3 and 4). Table 3 summarizes 26 studies with CV outcomes; and Table 4 summarizes 5 studies with cancer-related outcomes.
Reported items include citation, study location, urbanicity setting, sample size, study type, follow up/duration, covariates, age, interventions, greenspace exposure type, CV health or cancer-related outcomes, statistical analyses conducted, main findings, study strengths and weaknesses, and conclusions. Impact of greenspace or nature based interventions on CV health outcomes. The first column represents articles, the second column represents geographical settings of studies, the third column represents specific interventions used in the studies, the fourth column represents the measures of CV health, while the fifth column represents the conclusion in terms of protective effects (Beneficial effect), or no significant results (Not significant). This graphical representation shows an overall trend in findings across all studies included. Acronyms: SBP 1 : Systolic blood pressure; DBP 2 : Diastolic blood pressure; BNP 3 : Brain natriuretic peptide; HRV 4 : Heart rate variability; RAS 5 : Reninangiotensin system components; PNSA 6 : Parasympathetic Nervous System Activity; SNSA 7 : Sympathetic Nervous System Activity; hsCRP 8 : High sensitivity C-reactive protein; TNFα 9 : Tumor necrosis factor alpha; HR 10

Study design and demographics
All included studies used some kind of experimental designs. Thirteen (13) studies used simple pre-post study designs, some studies used the same group as the control and experimental group (on a different day/time) and measured statistical differences with paired sample t-tests [103,106,107,111,112,115,118,119,123,126,[128][129][130], and eight (8) studies used randomized control and experimental groups and measured statistical differences with independent sample t-tests [100,104,105,108,109,113,124,129]. Sample sizes ranged from 7 [120] to 155 [104] with an average sample size of 33.5. Study participants' mean age ranged from 10 years [103] to 80.3 years [122]. Twenty (20) studies included both male and female participants, 7 included males only [106,107,112,116,123,129,130], and 4 included females only [111,117,122,125]. No study specified nonbinary gender conforming or transgender identity.

Statistical analyses
Various statistical approaches were used in describing data and testing effects of NBI on outcome measures of CV health and cancer-related QoL. Descriptive statistics reported means and standard deviations as well as frequency distributions [105,111,116,125,127]. In addition to descriptive statistics, inferential statistics were used to determine statistical differences observed pre and post intervention. Some studies used specific tests for normality and homogeneity of variances such as Kolmogorov-Smirnov and Levene's tests [100,108,109] or Shapiro-Wilk test [104,105]. Studies with normally distributed data used parametric tests such as t-tests, chi square, spearman correlation or regression [104]. Studies with categorical outcome variables used Chi squared test for statistical independence or association between samples [104,108,122,124,126,128]; and some studies with dichotomous outcome variables incorporated McNemar's test [126,128] to determine if there are differences between two related groups. Other studies used regression models to test predictions of interventions effects on dependent continuous outcomes variables [104,125]; and studies with more than two groups to compare during interventions used ANOVA to test for statistical differences between groups' means [101,103,110,114,117,121,129]. Other studies used post adjustment tests such as Bonferroni post-hoc pairwise comparisons or partial eta squared (η2) test [103] or Cohen's d test [118,119] for effect size estimation. In addition to parametric tests, studies with non-normally distributed outcome variables used nonparametric tests such as Mann Whitney U test for between subjects' comparisons or Wilcoxon Signed Rank tests for within subjects' comparisons to compare statistical differences between samples [100,102,116,122,123,[104][105][106]108,109,112,113,115] or Kruskal-Wallis test for multi-group comparisons with post hoc Bonferroni adjustment [108]. One study used schematic views in representing their findings and did not specify the statistical test used [120].

Geographic distribution and urbanicity setting
Sixteen (16) studies were carried out in Asia, mostly in Japan and China, 12 in Europe, and 3 in North America. No study from other parts of the world (Africa, South America, and Australia) was identified. 22 studies were conducted in urban areas while 9 studies did not specify their urbanicity setting; and no study reported a rural setting for the experiment.

Summary of findings
Of 31 studies included in this review, 26 examined CV health related outcomes (Table 3) while 5 examined cancer-related outcomes ( Table 4). Results of these studies are described separately for CV and cancer outcomes.

Greenspace or NBIs on cardiovascular health.
Twenty-six (26) out of 31 studies included in the review looked at measures of CV health. Out of those 26 studies, 8 studies were conducted in Japan [104,106,107,112,115,116,120,123], 4 in China [100,108,109,124], 4 in the UK [103,114,117,121], and 2 in Taiwan [111,119]. One study was conducted in each of the following countries: Korea [122], Austria [110], Hungary [113], Poland [118], Lithuania [105], Finland [125], US [101] and Norway [102] (Fig 2). The most widely used intervention was forest bathing, quite common in Japan and China, followed by green exercise, nature viewing and gardening (Fig 2). The most reported outcomes were DBP, SBP, and HR, measured in 18 out of 26 studies. HRV was next and was measured in 5 out of those 26 studies, followed by measures of both the parasympathetic and sympathetic nervous systems, measured in 4 out of 26 studies. Few outcomes looked at stress measures of the cardiac myocyte such as the brain natriuretic peptide (BNP), Endothelin-1 (ET-1) and some components of the Renin-angiotensin system (RAS). Other outcomes investigated are measures of cholesterol such as high-density lipoprotein (HDL) and low-density lipoprotein (LDL). Most statistical tests conducted across all studies found that greenspace or NBI led to beneficial CV health outcomes (Beneficial effect), and some found no statistically significant difference (Not significant) (Fig 2).

Greenspace or nature-based interventions on cancer-related outcomes.
Five (5) out of 31 studies looked at cancer-related outcomes. Of those 5 studies, 3 were conducted in the US [126][127][128] while 2 were conducted in Japan [129,130]. Three US studies focused on vegetable gardening interventions while two Japanese studies focused on forest bathing interventions. Japanese studies looked at number of natural killer (NK) cells and their activity while US studies examined more diverse outcomes. Four of the outcome measures were related to positive health behaviors such as improved vegetable consumption habits [126][127][128], improved fruit consumption habits [127,128], increased PA [126,127] and gardening continuation [126]. Other outcomes were related to measures of physical fitness including strength [127], endurance [127], agility [127], and the two-minute-step test [126]. Three outcome measures were focused on overall health including weight loss [127], overall QoL [127,128], and reassurance of worth [128]. Three outcome measures were related biological markers including cortisol, a measure of stress [128], telomerase activity, a measure of aging [126,128], and interleukin-6 (IL-6), a pro-inflammatory biomarker and measure of systemic inflammation [128] (Fig 3). Observed trend suggests NBI's health protective effects on cancer outcomes (Beneficial effect) with few exceptional outcomes that were not statistically significant (Not significant) or significant only in control groups whereby control groups had better outcomes than the experimental groups (Significant in controls) (Fig 3). The 'significance in control groups' does not, in any way, suggest negative effect of the intervention. It is also not same as "not significant".

Greenspace interventions and outcomes
This review focused on NBIs or greenspace interventions. Diverse types of experimental exposure to greenspace were identified, including forest bathing, green exercise, vegetable gardening, and nature viewing (Figs 2 and 3). Outcomes investigated were related to CV health or cancer. Study locations were distributed across three continents including Asia, Europe, and North America. As hypothesized, observed trends suggest overall beneficial effects of greenspace interventions on both CV health and cancer-related outcomes, with some exceptions on few outcome measures.

Forest bathing.
Forest bathing "Shinrin-yoku" is a conscious and contemplative practice of being immersed in the sights, sounds, touches, tastes and smells of the forest [131]. This practice was developed in Japan in the 1980s as a physiological and psychological exercise and part of the national health program [132,133]. Its purpose was in twofold: (1) reduce burnout from the stressful work environment; and (2) inspire residents to reconnect with and protect the country's forests [133]. Scientists have then investigated its benefits on physical, mental, emotional, and social health outcomes [134]. Forest bathing is known to boost immunity [113,130,135], a plausible central pathway between nature exposure and human health benefits [136].
In this review, forest bathing was the most common intervention (15 out of 31 studies). Forest bathing was deployed in different forms including short forest recreation programs [113,118], forest therapy programs [111,112], longer slow walks in forests [100,104,[107][108][109]124,129,130], forest viewing vs urban viewing [106], and full forest immersion experience, comprised of sessions of slowly moving in silence through woodland, stopping to observe using all of senses (sight, smell, touch, hearing, and taste) and engaging in slow and relaxing breathing to ensure discovery and mindful appreciation of the woodland [119,121]. In 15 studies with forest bathing intervention, 6 were conducted in Japan [104,106,107,112,129,130], 4 in China [100,108,109,124], 2 in Taiwan [111,119], and one in Hungary [113], Poland [118], and UK, respectively [121].
Most statistical tests conducted found beneficial effects of forest bathing on outcome measures for CV health with few exceptions that did not find statistically significant associations. Few non-significant associations included some outcome measures including diastolic blood pressure (DBP) [107,111,113,118], systolic blood pressure (SBP) [107,124], HR [104,109,111,113,124], pulse pressure [109], and HRV [119]. One study found no statistical significance in both PSNA and SNSA [119]. Other remaining statistical tests conducted across various studies found significant beneficial effects. The first beneficial outcome observed is in measures of heart function such as reduced DBP [100,104,112,119,124], reduced SBP [100,104,111,112,118,119], lower HR [106,107,109,112,118,119], and increased HRV [121,124]. Another measured outcome that can impact CV health was stress. Stress reduction is salutogenic and was empirically observed with a decrease in stress hormones levels including urinary dopamine [107], adrenaline [112], and serum cortisol [109,112] after the intervention. Stress reduction was also observed with indicators of autonomic nervous system, such as enhanced parasympathetic nervous system activity (PNSA) [106] and suppressed sympathetic nervous system activity (SNSA) [106].
Improved systemic inflammatory profile is another beneficial outcome that was observed through reduction in both pro-inflammatory biomarkers and increase in anti-inflammatory biomarkers after forest bathing interventions. Reduced pro-inflammatory biomarkers include endothelin (ET-1) [100,108,109], IL-6 [100,108,109], tumor necrosis factor alpha (TNF-α) [109], homocysteine (Hcy) [100], and high sensitivity C-reactive protein (hsCRP) [124]. Increased anti-inflammatory biomarkers include serum adiponectin [107]. There were numerical differences between pre and post measures for two measured biomarkers of inflammation within the intervention groups, but no statistically significant differences were observed. Those non-statistically significant tests were for TNF-α [100,108] and hsCRP [108], and were reported in the alluvial chat as "Not significant".
Measures of oxidative stress were also improved after forest bathing interventions, as observed through lower levels of malondialdehyde (MDA) in experimental group post-intervention [108,109]. Last but not least, measured CVD pathological factors biomarkers were improved after forest bathing interventions as observed though serum reduction of constituents of the renin angiotensin system (RAS) (renin [108], angiotensin II (Ang II) [108], angiotensinogen (AGT) [100,108], angiotensin II type 1 receptor (AT1) [100,108], and angiotensin II type 2 receptor (AT2) [100,108]) and the brain natriuretic peptide (BNP), a biomarker of HF [108]. One study found mild reduction in renin and angiotensin II (Ang II) in the experimental group, although changes were not statistically significant [100]; and this was reported as "Not significant" in alluvial charts.
Most statistical tests conducted found beneficial effects of forest bathing on cancer-related measured outcomes including enhanced immune functioning observed through increase in number of NK cells [129,130] and their activity [129,130]. The forest bathing 'outcome-conclusion' chart is illustrated in Fig 4. Forest bathing is a promising intervention to improve CV health and QoL, particularly during cancer survivorship. Clinical practitioners, particularly those working in cardio-oncology specialties should examine more closely these non-invasive interventions and incorporate them in the standard of care to optimize CV health outcomes for cancer survivors through increased use of nature prescription programs, in addition to the clinical standards of care.

Green exercise.
Another commonly used intervention was green exercise (8 out of 31 studies). Green exercise has been defined as any PA occurring in a natural environment [114]. In this study, exercising with a view of nature through a window, on pictures, or on televisions was also considered "green exercise". Diverse green exercise interventions were used in studies included in this review, but most of them used nature visual stimuli. Duncan et al., 2014 had participants in the intervention arm of their study cycle for 15 min whilst watching a film of cycling in a forest environment [103]. Like Duncan et al., 2014, Pretty et al., 2005, had participants watch different scenes of videos projected on a wall whilst exercising on a treadmill [114] while Song et al., 2018's participants viewed Bonsai, small plants in container with restriction to roots or food storage capability [137]. The Bonsai used as a visual stimulus had characteristic mimicking natural landscapes that has been historically used in daily life in Japan [115]. White [110].
Green exercise has been shown to improve both physical and mental health [114,138] and higher enjoyment of exercise [139]. Some of the positive health outcomes previously associated with green exercise include greater feelings of revitalization and positive engagement [140], and improvement in measures of mood and self-esteem [141] such as depression, tension, and anger [142,143]. Green exercise has been suggested by previous scholars as a potential workplace intervention to reduce job stress and promote restoration [144]. Chronic stress has been linked to increased CVD risk [145][146][147], including a 40-50% increase in the occurrence of coronary heart disease in prospective observational studies [145,148].
In our review, interventions with green exercise were conducted in different countries, including the UK [103,114,117], Lithuania [105], Finland [125], Austria [110], Japan [123] and the US [101]. Green exercise was found to be positively associated with many outcome measures related to CV health with few statistical tests that found no significant associations or no Forest bathing intervention effects on both CV health and cancer-related outcomes: Trends of associations among all statistical tests conducted. The first column represents outcome measures while the second column represents summary conclusions in terms of protective effects (beneficial effect) or no significant results (not significant). Acronyms: SBP 1 : Systolic blood pressure; DBP 2 : Diastolic blood pressure; BNP 3 : Brain natriuretic peptide; HRV 4 : Heart rate variability; RAS 5 : Renin-angiotensin system components; PNSA 6 : Parasympathetic Nervous System Activity; SNSA 7 : Sympathetic Nervous System Activity; hsCRP 8 : High sensitivity C-reactive protein, TNFα 9 : Tumor necrosis factor alpha; and NK 10 : Natural killer cells. numerical difference at all. Green exercise's beneficial CV health outcomes include observed reduction in SBP [103,114,117], DBP [105,114,117], HR [117,123,125], and increase in HRV [123,125]. Another significant change observed was a reduction in cortisol, a measure of stress [105]. Some studies did not find a significant difference on measures of SBP [101,105,110,125], DBP [101,103,110,125], and HR [101,110,114], including one study that found no association between green exercise and one measure of CV health, HR [103]. For studies that looked at cancer-related outcomes, none used a green exercise intervention. The green exercise 'outcomes-conclusions' chart is illustrated in Fig 5. No study in this review investigated the impact of "green exercise" on cardiotoxicity among cancer survivors. This literature gap suggests the need for empirical investigation on the role of greenspaces in reducing risks for cardiotoxicity in this highly vulnerable population and testing the use of such interventions in Cardio-oncology clinics to optimize CV health and improve cancer survivorship care. Additionally, only one statistical test investigated the impact of green exercise on CV biological markers by looking at cortisol. Future studies should investigate more biomarkers, including additional stress biomarkers and CVD pathological factors such as the components of the renin angiotensin system and inflammatory biomarkers such as IL-6, hsCRP, TNF-α etc..

Vegetable gardening.
Gardening interventions provide individuals with hands-on experience planting, growing, and harvesting fruits and vegetables, which may promote consumption of fruits and vegetables [149,150]. Individual benefits of gardening activities include increased PA, access to fresh air, landscape beautification and enjoyment [151]. Gardening interventions have been linked to many health benefits [152] including improved physical health [153,154] and mental health [155][156][157]. Gardening has been proposed as a strategy for health promotion in aging women [158] and its prescription, along with other conservation activities are recommended to improve health and wellbeing in aging population [159]. In the cancer care continuum, gardening interventions have been linked to positive health outcomes and improved survival [160,161]. Some specific benefits of gardening during cancer survivorship include improved dietary habits, improved PA, and improved QoL [162].
In this review, vegetable gardening interventions were conducted in Japan [120], South Korea [122] and the US [126][127][128]. Two studies looked at CV health related outcomes [120,122] while three studies looked at cancer-related outcomes [126][127][128]. Most studies found beneficial effects of gardening interventions on outcome measures related to CV health and cancer-related QoL, with some exceptions that found no statistically significant changes (not significant), or significant only among controls. Those 'not significant' exceptions include some statistical tests on outcome measures of weight loss and overall QoL in one feasibility study in cancer survivors [127], some biomarkers including the monocyte chemoattractant protein-1 (MCP-1), NADPH oxidase p47, and the inducible nitric oxide synthase protein (iNOS) [122], stress hormone cortisol and IL-6 [128], and low-density lipoprotein (LDL) [122]. Two tests found significance among controls, one on overall QoL [128] and another one on telomerase activity [126], an enzyme responsible for maintenance of telomeres length by addition of guanine-rich repetitive sequences in both gametes and stem and tumor cells [163].
Included studies in this review showed beneficial effects of gardening interventions on stress [120], total cholesterol and HDL [122], BP [122], dietary habits [126][127][128], positive selfcare behaviors [126,127], physical performance [127], increased reassurance of worth [128] and improved aging process [128]. Stress reduction benefits were observed through proxy measures with enhanced parasympathetic nervous system activity (PNSA) [120] and suppression of sympathetic nervous system activity (SNSA) [120]. Benefits on blood cholesterol level were measured through improved high-density lipoprotein [HDL], or good cholesterol profile [122]. Beneficial outcomes in BP were measured with both decreased SBP [122] and decreased DBP [122]. Improvement in dietary habits was observed through improved vegetable and fruit consumption [126][127][128]. Positive self-care behaviors were observed through improved PA [126,127] and gardening continuation [126]. Physical performance improvement was observed through improvement in the 2-minute-step test [126] and other measures including improved strength [127], improved endurance [127] and improved agility [127]. Increased reassurance of worth was measured with self-reported assessments of psychosocial measures [128]. Improvement in aging process was observed through a decrease in telomerase activity [128]. The gardening interventions impact on both CV health and cancer-related outcomes, along with the overall conclusion are graphically illustrated in Fig 6. Observed trends (Fig 6) suggest that gardening is a promising intervention to improve outcomes related to CV health and QoL during cancer survivorship. Cardio-oncologists should keep close collaborations with primary care providers in optimizing the cancer survivorship care by including these innovative interventions to improve CV health and survivorship experience. Community leaders, including local government and other community-based organizations should work together to ensure presence, accessibility, and use of community gardens. In addition to supporting positive healthy gardening behaviors, those gardens also have potential to increase access to healthier foods options for residents in "food deserts" and "food swamps" neighborhoods [164][165][166]. Such gardens could also enhance biodiversity, local ecosystem, water management and contribute to local climate change resilience strategies [167]. Additionally, continuous targeted messaging campaigns should be in place to remind those at increased risk of the benefits associated with gardening. Academic partners should come in to continuously evaluate impact and suggest best practices to ensure maximum benefits from all the resources set aside for such a community wide intervention to support intergeneration equity. Future studies should incorporate more biological measures including pathological factors for CVD such as biomarkers of oxidative stress and more inflammatory biomarkers in addition to IL-6, the only pro-inflammatory biomarker that was investigated in vegetable gardening interventions studies included. 4.1.4. Nature Viewing. Exposure to natural environments including viewing them has been linked with improved restoration and cognitive capacity [102] and autonomic function recovery after acute-mental stress [168]. In this review, we found studies that tested nature viewing effects on measures of CV health including HR, SBP, DBP, PSNA and SNSA. Those studies were carried out in two countries, Japan [115,116] and Norway [102]. Statistical tests found beneficial effects of nature viewing on CV health including reduction in HR [102,116], enhanced PSNA [115,116] and suppressed SNSA [115]. Tests on measures of SBP and DBP were not statistically significant [116]. The nature viewing interventions on both CV health outcomes, along with the overall conclusion are graphically illustrated in Fig 7. Contrary to other NBI in this review (forest bathing, green exercise, and vegetable gardening), nature viewing intervention did not measure a single biological marker of inflammation. Future studies should investigate nature viewing's impact on biomarkers including CVD pathological factors such as the components of the renin angiotensin system and pro-inflammatory biomarkers such as IL-6, hsCRP, and TNFα. Such knowledge would complement current behavioral self-care plans particularly for cancer survivors in reducing risk for cardiotoxicity; and nature viewing is a relatively harmless intervention, amenable to change and relatively easier to implement.
Presented all together, this review suggests that forest bathing and gardening interventions have the most beneficial outcomes (Figs 4 and 6) compared to other interventions (nature viewing or green exercising) which are also beneficial, but to a less extent (Figs 5 and 7). Intervention specific alluvial charts suggest more thickness for "beneficial effect" for forest bathing and gardening compared to nature viewing or green exercise. These findings have implications for increasing use of forest bathing and/or gardening interventions to improve CVD and/or cancer outcomes. Nature viewing and green exercise interventions remain also very important in improving outcomes. The clinical use of these interventions would be best assessed with patient preference and what interventions they are most likely to adhere to.

Limitations
While this review is methodologically rigorous, it has some limitations. First, in the risk of bias assessment, we used a modified version of the Newcastle Ottawa Scale because there was no validated tool that accurately assessed all types of studies included in our review. While the official NOS has been validated for case-control and cohort studies, the scoring guide created for this study by modifying the scale to capture factors related to experimental or pre-post studies has not been validated, and it's scoring can be subjective. This subjectivity was attenuated by ensuring that two reviewers (J.C.B. and J.S.B.) independently assess all studies. Secondly, we reported trends across all relevant statistical tests conducted in all included studies with alluvial charts to visualize our results summary, but no meta-analysis was done to suggest any statistical inference for all the articles if taken altogether. Therefore, our trend across all studies should be seen as a descriptive summary of findings; and any inference made should consider all studies collectively. Thirdly, not all included studies measured, adjusted for, or reported the same variables. This is why we used alluvial chart to summarize similar trends instead of conducting a meta-analysis to deduct any statistical inference for all studies combined. Last, this review is not immune to other limitations discussed by the authors of the included studies, which may include small sample size, lab errors, potential misclassification, or other measurement errors. Regardless of the limitations, this review is an outstanding summary of impact of greenspace or nature based interventions on both CV health and cancerrelated outcomes and highlight benefits with direct implication for clinical and public health practice.

Conclusion
This review sought to assess the impact of greenspace or NBI on: (1) CV health, and (2) cancer-related outcomes.
Interventions used included a Japanese tradition of forest bathing or "shinrin-yoku," green exercise, gardening, and nature viewing. CV health related outcomes include measures of BP, HR, HRV, autonomic nervous system activity, stress biomarkers including cortisol, oxidative stress measures such as iNOS, RAGE, and NADPH oxidase p47, CVD pathological factors including lipid profile, components of the renin angiotensin system, pro-inflammation biomarkers including IL-6, hsCRP, TNFα, ET-1, Hcy, MDA, and MCP-1 and anti-inflammatory biomarkers including adiponectin. Cancer-related outcome measures include measures of physical performance such as physical strength, endurance, and agility; personal behaviors such as vegetable and fruits consumption, PA, and weight loss; biological markers including stress markers (cortisol), inflammatory markers (IL-6), some components of the renin angiotensin system (RAS), and some immune function markers including both the count of natural killer cells as well as their activity.
An overall trend across studies suggests beneficial effects of greening and NBI on both CV health and cancer-related outcomes, although not all studies found a significant benefit. Cardio-oncologists, along with primary care providers should incorporate these innovative interventions in the standard of care to optimize both CV and cancer-related health outcomes.
Future studies should combine multiple measures of CVD pathological factors including components of the renin angiotensin system (renin, Ang II, AGT, AT1 and AT2), multiple markers of oxidative stress, multiple measures of both pro and anti-inflammatory biomarkers, and multiple biomarkers of stress. Other direct and relatively easier measures such as BP, HR, pulse pressure and HRV would be important to add to this line of investigation. Additionally, future studies should pay more attention to some populations with higher CVD risk such as cancer survivors to order to investigate the premise of such innovative population-based approaches in reducing cardiotoxicity from cancer treatment therapies and optimize the survivorship experience.
Existing conceptual models such as the "Greenspace and Health Equity model" [169] or the "Greenspace in Cardio-Oncology model" [1] can be very useful in future research on greenspace and CardioOncology disparities. There is a need for increased research funding from relevant organizations such as the American Heart Association, the American Cancer Society, and National Health Institutes including the National Cancer Institute. This knowledge will promote a more robust understanding of the role of greenspace and NBI on CV and/or cancerrelated outcomes as well as their critical contribution to climate resilient neighborhoods. The focus on biomarkers is particularly relevant for clinical practice as more biomarkers can clinically be measured and greenspace interventions impact on CV health can be continuously assessed during all stages of the cancer care continuum. Such practice can help reduce risks for MACE, reduce mortality, and improve cancer survivorship quality and survival.
Supporting information S1 Checklist. The PRISMA 2020 checklist: Appendix A.
(DOCX) S1 File. The full databases search strategy and alluvial charts data files: Appendices B and C. (DOCX)