Abstract
The association between mental health symptoms and the quantity of green space surrounding the homes of people who recently experienced a cardiovascular event were examined while also assessing plausible mechanisms. Both objective- and perception-based metrics were used to characterize the amount of green space in a half-mile buffer around the homes of 142 participants enrolled in a cardiac rehabilitation program. Participants completed the Depression, Anxiety and Stress Scale to assess mental health symptoms. Increased quantity of green space around the home was associated with fewer depressive symptoms but was not clearly associated with symptoms of anxiety and stress. Physical activity in green space was associated with lower odds of all three outcomes, while having green space in view from the home was not associated with any outcome. These results suggest that increased accessible green space near the home may improve depression and promote recovery in this population. This may be due to physical activity in this space.
Introduction
Depression is one of the most common mental illnesses in the United States, with around 16 million Americans affected by major depressive symptoms each year. 1 Although depression is relatively common in the general population, certain clinical populations are even more predisposed to co-morbid depressive symptoms. Patients who have had a recent cardiovascular event, such as myocardial infarction (MI) or cardiac surgery, are one such population that is especially affected by depression.2–7
Depression in cardiovascular event patients has been associated with higher risk of morbidity and mortality, poorer clinical and other behavioural or psychological outcomes, reduced overall quality of life and subsequent cardiovascular events.8,9 Finding ameliorating factors for depression in this population could result in more effective recoveries and reduced risk of developing additional health problems. 10 Because in-hospital recovery after a cardiovascular event is relatively brief, averaging only three to five days,11,12 the greatest opportunity for reducing depressive symptoms is to identify preventive factors during the home recovery phase.
Increased quantity of green space near the home, defined as land that is partly or completely covered with grass, trees, shrubs or other vegetation (e.g. parks, community gardens, cemeteries), has been shown to be associated with less depression in the general population and may therefore be a preventive target for reducing depressive symptoms in cardiovascular event patients. 13
Potential mechanisms by which green space can reduce depression include the psychological benefits of viewing nature and improved accessibility for physical exercise.14–16 The aesthetic value of green space may play a role in the reduction of depression because of improved stress restoration and cognitive functioning associated with viewing natural areas.17–20 Green spaces near the home also provide increased access and opportunity for physical activity, which has been shown to reduce depressive symptoms. 21 In addition, previous literature has shown increased mental health benefits of exercising in green areas compared to built environments. 22
Although there has been evidence of an association between green space and reduction of depression in general populations, there have been no studies to date assessing this association in a cardiac rehabilitation population. Our study addresses this gap by examining the association between the quantity of green space near the home and mental health symptoms (depression, as well as anxiety and stress) in a population recovering from a recent cardiovascular event, as well as exploring potential mechanisms that may explain such an association.
Methods
Study population
We surveyed 142 patients from two cardiac rehabilitation sites in Rochester, NY (Monroe County): The University of Rochester Medical Center (URMC) and Rochester Regional Health at Unity Cardiac Rehabilitation (Phase II) centres. Generally, cardiac rehabilitation is a medically supervised program that helps improve the health and well-being of people who have had a recent cardiovascular event. Phase II of cardiac rehabilitation programs begin when the patient leaves the hospital and recovers at home.
The total rehabilitation population in our clinic population consists of participants with: recent MIs, heart surgeries (such as a coronary bypass), cardiac transplants, angioplasty or stenting, stable angina and placement of a ventricular assist device (data on the distribution of cardiovascular event types experienced was not available). There were no exclusions made based on cardiovascular event type. Recruitment protocols for both sites were reviewed and approved by the University of Rochester Research Subjects and Review Board.
Green space measurement
Three methods were used to quantify residential green space: A manual (study-specific) geographical information system (GIS)-based method, the Normalized Difference Vegetation Index (NDVI) and self-reported quantity of green space near the homes. The process and rationale for quantifying these three different measures of green space are described in detail below.
Objective green space measurement using a manual GIS-based method
Although the NDVI has been commonly used in epidemiologic literature as an exposure measurement tool for green space, it likely overestimates publicly accessible green spaces by capturing all areas that are green in colour. Green space near the home that is accessible to the public has been shown to be associated with increased likeliness of physical use.16,23 To quantify publicly accessible green space, we calculated objective green space quantity around a participant’s home through a manual exposure measurement using the GIS software, ArcMap 10.1(ESRI, Redlands, CA). We used a definition of green space, based on the definition provided by the US Environmental Protection Agency, 24 that included all parks, playgrounds, school fields, cemeteries and other open spaces that are publicly accessible.
Using this definition of green space, we manually outlined polygon features to border local green spaces and estimated the area of green space in a circular buffer with a half-mile radius around each participants’ home. We confirmed green space areas using Google Earth Street View (also known as ‘ground truthing’). Similar buffer analyses have been used previously in green space and health outcomes research.25–29 The percentage of designated green space area was calculated as a continuous measurement (% of buffer area). Green space was also categorized into ‘low’, ‘medium’ and ‘high’ groups based on tertile distributions of the continuous GIS-based green space measurement. Because of low case numbers in each group, the group having the highest amount of green space was analysed (‘high’ (≥17.5%)), compared to low or medium green space groups combined (‘low’ (X = 0%–7.4%) and ‘medium’ (7.5%–17.4%)).
To account for any exposure measurement error by the reviewer, validation of our method was conducted. A separate researcher was asked to independently identify and measure the green space areas around 10 homes that had already been processed and measured. The Pearson correlation between the continuous green space measures of the two researchers was 0.72 (p: 0.01), which indicates that this method was reliable and replicable.
Objective measurement using NDVI
The amount of green space was also estimated using the NDVI as a comparison to our method. An NDVI map of our catchment area was previously developed by the GIS Services Division of the Monroe County local government. NDVI-based green space metrics were available for 121 homes of participants in the study, the remaining 21 homes of study participants were excluded from this exposure measurement based on the low quality of available data. The NDVI was scaled from 100 to 200 (with 200 being the highest reflectance of green colour). An average NDVI score was calculated from all NDVI pixels in a half mile circular buffer in the 121 participants for whom data were available.
Perceived green space assessment
In addition to the objective measurements, green space was also measured based on the participants’ self-reported quantity of green space near their homes recognizing that perceived quantity might better reflect and be affected by their ability and willingness to access green spaces and their use of these resources.30–32 Participants were provided our definition of areas that we considered green spaces before these questions in our survey.
To address perceived residential green space quantity, we asked participants, ‘How many parks, recreational fields, or other green spaces do you have within a 10-minute walk from your home?’ The possible responses for this question were: ‘None’, ‘Very Little’, ‘Some’ and ‘Very much’. This question was designed for this study using similar questions that were previously presented by the Trust for Public Land. 33 The time of 10 minutes was used in this question because that is what is believed to be an adequate time to walk a half mile, which is comparable to our primary buffer zone radius in our GIS-based measurement.33,34 Since case numbers were low in the individual perceived quantity groups, we decided to dichotomize this variable into ‘none/very little’ and ‘some/very much’.
Assessment of potential mechanisms
In addition to our primary analyses with quantity of green space, we also examined the potential influence of two proposed mechanisms (view from the home and physical activity) potentially causing the association between green space and depression via survey questions. The following sections further describe the assessment of each mechanism.
Mechanism 1: Perceived view of green space
To assess how participants perceived the amount of green space in view from their home we asked, ‘How much of the view from your home window is of nature (trees, plants, etc.)?’ Responses included: ‘none at all’, ‘very little’, ‘some’, or ‘very much’. This question was previously developed for a study of aggressive behaviour and green space view in inner city Chicago adults. 35
Mechanism 2: Physical activity
A question from the Physical Activity in Parks Setting questionnaire was also included to assess physical activity intensity. 36 We asked participants, ‘On a typical visit (in general) to a park or other green space near your home which of the following best describes your level of activity when visiting these areas?’ The participant may indicate: ‘do not visit parks’, ‘mostly sitting’, ‘mostly light activities’ (standing, walking or strolling), ‘mostly moderate activities’ (walking at moderate pace and tennis) and ‘mostly vigorous activities’ (jogging, team field sports and basketball).
There were 126 participants (89%) who provided responses to this question. To best address missing items and the distribution of cases among these groups, we collapsed the ‘mostly sitting’ and ‘mostly light activities’ groups as a light intensity group and the ‘moderate’ and ‘vigorous’ activity groups as a heavy intensity group. We excluded those that do not visit parks to assess differences in mental health outcomes with physical activity intensity among those that visited local green spaces.
Mental health assessment
The Depression, Anxiety and Stress Scale (DASS) was used because of its psychometric properties, and because it has been previously used in both green space research and in studies of cardiac rehabilitation populations.37–40 The DASS consists of a set of three self-report scales designed to measure the negative emotional states of depression, as well as anxiety and stress. Due to their mechanistic similarities in a potential association with green spaces and to compare our results in a clinical population to the existing body of literature in the general population, 41 results from the anxiety and stress scales were also included as outcomes in this study.
DASS scores (0–21) were based on seven items with a Likert scale of 0–3, with 0 indicating the participant never experienced the emotional state and 3 indicating they almost always experienced the emotional state. Because distributions of the continuous scores were skewed towards 0, and due to our limited sample size, we decided to dichotomize the continuous DASS scores into the following groups: (1) ‘normal’ or (2) ‘minor or above’ (minor, moderate, or severe scores). This categorization was based on DASS standard score cut-offs for minor depression (10), anxiety (8) and stress (15). 42
In addition to using the DASS scores, psychotropic medication use (i.e., antidepressants, assessed via survey) was considered as a proxy for mental health symptoms. To incorporate medication-use in our case ascertainment, we included those who responded as taking anti-depressant medication in the 30 days prior to the survey date, and not scoring minor or above on the mental health scales, in our final case definitions.
Our primary dependent variable was a dichotomous variable that incorporated information from both the DASS and self-reported medication use (i.e. either scoring above the DASS threshold or taking medication vs neither).
Statistical analyses
We conducted crude and adjusted Poisson regressions with counts of our dichotomous outcomes for depression, stress and anxiety. Total numbers in each model varied based on availability for the specific exposure variable and/or covariate data. Adjusted Poisson regressions calculated risk ratios (RRs), controlling for age, sex and employment.
Crude and adjusted Poisson regressions were also conducted to explore the association between our dichotomous mental health outcomes and the categorical data for the two proposed mechanisms, view of green spaces and physical activity in green spaces. These models also controlled for age, sex and employment.
All analyses were performed in SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).
Results
The study population was primarily male (69%), white (96%), had at least some college education (75%) and overweight or obese (95%) (Table 1). The mean age of the entire study population was 67 years (range: 39–92 years). Participants were, on average, one month into their rehabilitation when given the study survey. Alcohol consumption (mean of 1 drink per month) and current smoking (5%) were low, which are expected from a population in rehabilitation. The study population was also primarily married (66%) and unemployed/retired (72%). Overall, the distributions of our study population are comparable to the entire cardiac rehabilitation population from which the study population was recruited and to national statistics of cardiac rehabilitation enrollees.43,44
Covariate distributions by depression, anxiety and stress status (minor or above a ).
BMI: body mass index.
Classification for ‘minor or above’ depression, anxiety and stress was determined using scores and cutoffs from the DASS-21 scale, as well as including those with antidepressant medication use (regardless of DASS-21 score).
When including medication use in our mental health case ascertainment, our study population consisted of 30% (n = 42) with minor or above depression (15% from the DASS alone), 32% (n = 46) with minor or above anxiety (24% from the DASS alone) and 26% (n = 37) with minor or above stress (11% from the DASS alone).
Objective manual green space measurement
Adjusted and unadjusted Poisson regression results for the objective green space analyses were nearly identical (Table 2). Compared with those with low or medium green space near the home, those with a high amount of green space showed a trend of reduced risk of having minor or above depression, (RR: 0.58 (0.27–1.25)). We also found a trend for slight reductions in risk of depression with every 1% increase in green space area (RR: 0.97 (0.94–1.01)). Although both the categorical and continuous analyses of green space and depression showed some reduced risk in the hypothesized direction, neither were statistically significant. There was no association between green space (categorical or continuous) with anxiety or stress.
Crude and adjusted Poisson regressions of GIS-based measurements of green space and mental health.
GIS: Geographical Information System; RR: risk ratio; DASS: Depression, Anxiety and Stress Scale.
Adjusting for age, gender and employment (y/n).
Classification for ‘minor or above’ depression, anxiety and stress was determined using scores and cutoffs from the DASS-21 scale, as well as including those with antidepressant medication use (regardless of DASS-21 score).
NDVI objective measurement
NDVI measurements of green space around 121 of the participants homes significantly differed from our GIS method and the self-reported green space amounts. For instance, means of the GIS-based measured green space did not correspond with low (12.21%), medium (17.69%) and high (12.65%) green space based on the NDVI. In addition, mean NDVI values were not different between those who perceived more green space compared to those who perceived less (134.74 vs. 134.06). In terms of an association with a continuous NDVI measure of green space, we did not observe any association with depression (RR: 0.99 (0.98–1.02)), anxiety (RR: 1.00 (0.98–1.02)) or stress (RR: 1.00 (0.98–1.02)) (Table 3).
Crude and adjusted Poisson regressions of NDVI-based measurements of green space and mental health.
NDVI: Normalized Difference Vegetation Index; OR: odds ratio.
Adjusting for age, gender and employment (y/n).
Classification for ‘minor or above’ depression, anxiety and stress was determined using scores and cutoffs from the DASS-21 scale, as well as including those with antidepressant medication use (regardless of DASS-21 score).
Perceived green space measurement
Crude and adjusted Poisson regression results were also similar when evaluating participant’s perceived amount of green space (Table 4). Those who perceived more green space within a 10-minute walk from their home (‘some/very much’) had a lower risk of minor or above depression compared to those with very little or no green space (OR: 0.52 (0.28–0.97)) in unadjusted analyses. After adjustment for age, gender and employment, the effect estimate remained similar and protective but was no longer statistically significant (RR: 0.57 (0.31–1.08)). Reduced risk of anxiety (RR: 0.74 (0.41–1.35)) and stress (RR: 0.68 (0.35–1.34)) was found with higher perceived green space, but these estimates were not statistically significant.
Crude and adjusted Poisson regressions of perceived amount of green space (within 10-min walk) and mental health.
RR: risk ratio; DASS: Depression, Anxiety and Stress Scale.
Adjusting for age, gender, and employment (y/n).
Classification for ‘minor or above’ depression, anxiety and stress was determined using scores and cutoffs from the DASS-21 scale, as well as including those with antidepressant medication use (regardless of DASS-21 score).
Mechanisms
Table 5 shows the results from Poisson regressions exploring both potential mechanisms: view of green space and physical activity in green spaces. There were 52 (38%) participants with ‘very little/some’ green space in view from their home and 85 (62%) with ‘very much’. Having more green space in view from the home was not associated with any of the three mental health outcomes.
Results from Poisson regressions exploring the association between mental health and two mechanisms potentially explaining green space benefits.
OR: odds ratio.
Classification for ‘minor or above’ depression, anxiety and stress was determined using scores and cutoffs from the DASS-21 scale, as well as including those with antidepressant medication use (regardless of DASS-21 score).
Adjusted ORs (controlling all models for employment, gender and age).
Participants responded to the green space physical activity intensity question as follows: None or light activity (n = 90, 71%) and moderate to vigorous intensity (36, 29%). In those with moderate/vigorous exercise in local parks and green spaces compared to those performing ‘none/light’ exercise, a 51% reduction in the odds of having minor or above depression was found; (OR: 0.49 (0.20–1.19)) that was borderline significant and a 59% reduction in odds of having minor or above anxiety (OR: 0.41 (0.17–0.97)). Increased physical activity intensity was also associated with reduced odds of stress but this estimate was not statistically significant (OR: 0.53 (0.22–1.29)).
Discussion
Our results show that having higher quantity of accessible green spaces around the home measured objectively or based on participant’s perceived quantity was associated with lower levels of depression but not with stress or anxiety. While these estimates were not statistically significant, mostly due to our smaller sample size, they were in the hypothesized direction, particularly for depression; and the effect estimates were relatively large. Physical activity intensity in green spaces was associated with reduced symptoms of all three outcomes but these estimates were also not statistically significant.
Interestingly, the NDVI measure was not correlated with our measures of green space and was not associated with mental health symptoms. These results suggest that our method of determining accessible green space may be more appropriate to use in studies of green space exposure and mental health. Although one of the potential mechanisms between green space and mental health is the viewing of green spaces, 45 where NDVI would be appropriate to use because it captures public and private green spaces that can be viewed from the home. Our study did not find an association between quantity of green space in view from the home and mental health.
Secondary analyses were also conducted to compare our objective and perceived measures. We found that those who perceived more green space in a 10-minute walk from the home had a higher mean of green space percentage in a half-mile buffer zone around the residence (objectively measured with GIS) compared to those who perceived less green space (16.16% vs. 11.03%). This shows that these two measures are related, but our difference in results using the objective and perceived measures may indicate a difference in how these measures could provide different measures of effect on mental health in epidemiologic studies.
The perception of higher quantities of green space near the home was associated with lower levels of symptoms in all three mental health illnesses, while the actual measurement of green spaces in roughly the same catchment area was only associated with depression and minimally with stress. These findings were similar to those in an Australian cohort, where those who perceived a lot of green space, compared to those who perceived low amounts of green space, had a 60% increase in the odds of having a better mental health score (OR: 1.60 (1.26–2.04)) on the Short-Form Health Survey (measuring composites of depression and anxiety). 46 Recent studies in general populations have found increased quantity of measured green space near the home to be associated with lower few depression symptoms in adults and children.47–49 The difference between perceived and objective measurements may be due to participants using ‘symbolic proximity’ to green spaces, where they perceive more parks that are closer to home and those that they regularly use, rather than all parks in the exact distance covered in an objective measure. 50 This relational view may explain the differences in the results of our two measurement approaches.
There are several potential limitations to this study. The first limitation was the potential for exposure misclassification. The subjective nature of our exposure measurement may be limited by error from the researcher in identifying green space areas that are not truly green space, and not identifying areas that are truly green spaces. Nevertheless, in the validation of our method, we found a Pearson correlation of 0.72 (p value: 0.01) between the two continuous measures of greens space around those homes.
Limitations may also exist with our mental health outcome assessment. In this study, we included anti-depressant and anti-anxiety medication as indicators of asymptomatic mental health outcomes in our case ascertainment which can sometimes be prescribed for indications other than depression, stress or anxiety; however, other studies have indicated medication use a fair proxy for mental health outcomes.51–53 This suggests that it is acceptable to include medication use as a proxy for underlying mental health symptoms, even if these symptoms are currently controlled.
There may also be a limitation with this study based on the temporality of depression and cardiovascular events. It is possible that some patients may have had depression that preceded the event. We were unable to collect data on history of depression, which is a predictor for depression, anxiety and stress. However, previous literature has indicated that depression is more likely to develop after a cardiovascular event. One study found that 22% of patients that were not depressed at the time of hospitalization for a cardiac event had developed moderate to severe depression three months after the event. 54 Future studies can address this potential limitation through longitudinal data collection (i.e. ascertaining mental health status before, during and after a cardiac event).
Finally, this study was limited by a small sample size. Therefore, most of our point estimates lacked statistical precision. Nevertheless, their magnitude and consistency should be considered suggestive of an association and are consistent with hypothesized directions.
Despite the above limitations, this study also has several strengths. A major strength of our study was the use of individual-level green space exposure measurements. Previous studies have used geographic areas, such as census tracts, to measure a participant’s green space availability. In addition to individual-level exposure assessment, our study also utilized a novel approach to green space quantification that improves upon existing methodology. When we study how green space may impact health, we must also consider use and accessibility of these areas in our quantification methods. NDVI, used in much of the previous literature, measures all land areas having a green colour of reflection. Although this method is accurate for measuring publicly accessible green spaces in very urban areas (where NDVI is likely to capture all green spaces), in areas that are more sub-urban or rural (having more privately owned green spaces), this method likely overestimates usable green space area. 55 The method used in our study did not include private fields, lawns or other green spaces on private properties because these are not appropriate to include in research examining the underlying concepts of green space usability, such as physical activity, found to be very important in our study.
Conclusions
This is the first study to date that examined green space quantity near the home specifically in those with recent cardiovascular events, a population at high-risk for mental health problems. Results from this study may be generalized to other clinical populations with no existing research, where patients are recovering from a recent surgery, adverse health event or traumatic event from home. In these such populations, these health events or trauma are risk factors for mental illness and green space near the home may be associated with reduced mental health symptoms. 56
Future studies could improve on the existing literature in several ways. First, the use of a longitudinal cohort with a larger sample size would be very important for understanding how green space impacts the development of mental health in this type of population. Ideally, studies should examine participants before, during and after (recovery period) a cardiovascular event to understand how mental health is influenced by the cardiovascular event and by exposure to green space over time. Finally, our findings (if confirmed) can advocate the use of green space physical activity in cardiac rehabilitation and add to existing literature on mental health benefits of green space to leverage the development of green spaces in new areas or maintenance of existing green spaces for the benefit of public health.
Footnotes
Authors’ contributions
All authors designed the study and developed methodology. KK assisted in the literature review, exposure assessment methodology, and interpretation of the green space exposures. RB assisted in the literature review of cardiac populations and mental health, the study design (including the selection of the study population), and interpretation of results. BC assisted in the literature review of mental health (with respect to green space and cardiac events) as well as provided expertise and content with the DASS exposure measurement as well as the statistical analyses. DP and EvW developed the study design, executed the study recruitment and data analyses, and assisted in literature review as well as results interpretation.
Acknowledgements
The authors gratefully acknowledge the assistance of staff from both cardiac rehabilitation centres (URMC and Unity).
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.