Greenspace exposure and associated health outcomes: a systematic review of reviews

Search strategy

An electronic literature search was carried out using the following databases: PubMed, Embase, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), Scopus, and the Cochrane Database of Systematic Reviews. We restricted the search to peer-reviewed systematic reviews and meta-analyses conducted in humans and published in English between January 2010 and December 2020, including reviews first published online during this period (extended data: Supplementary Table S1). All databases were searched on 05/01/2021. To ensure that relevant systematic reviews were retrieved, we also conducted forward and backward reference searching on screened reviews, as well as manual searching of reference lists of relevant reviews. EndNote software programme was used for reference management (The EndNote Team, 2013).

Eligibility criteria

Title and abstract screening and full-text screening of identified systematic reviews were independently conducted by two authors. The following inclusion criteria were used: i) systematic review or meta-analysis; ii) written in English and published in a peer-reviewed journal; iii) greenspace exposure clearly defined using objective or subjective measures; iv) reported health outcome(s) were directly attributable to greenspace exposure. We excluded: i) systematic reviews that failed to follow a standardised, systematic review approach, for example, no mention of databases searched; no identification of the search terms used; or no assessment of original studies quality; ii) scoping reviews; iii) systematic reviews that did not consider the effects of greenspace independently of blue space. Reviews that mentioned health outcome(s) in the title and abstract but only reported health determinants such as body mass index (BMI), physical activity (PA), or diet were excluded during full-text screening (extended data: Supplementary File S1). Disagreements were resolved via discussions with the last author. Covidence was used to conduct all stages of study screening (Covidence systematic review software, 2022).

Data extraction

Data extraction was conducted independently by two authors. Pilot testing was conducted on 10 of the identified systematic reviews to build the data extraction form. The form included information on the design of original studies, types and measures of greenspace exposures, and types of health outcomes. Any disagreement in the data extraction process was resolved via discussions with the last author.

Analysis criteria

The objectives, methods, and conclusions of each systematic review included in this umbrella review were summarised. Quantitative data, reported as the effect estimate with associated confidence interval, were summarised for health outcomes if available. We could not conduct a meta-analysis as the primary aim of any umbrella review is to summarise existing systematic reviews and not to re-synthesise or conduct meta-analysis (Aromataris et al., 2015).

The number and types of original studies in the systematic reviews, as well as the measure(s) of greenspace exposure and health outcomes, were recorded. The International Classification of Diseases, 10^th Edition (ICD-10) was then used to allocate corresponding ICD-10 codes for each health outcome identified during data extraction (World Health Organization, 2019).

The methodological quality and risk of bias of included systematic reviews were evaluated independently by two authors using A MeaSurement Tool to Assess systematic Reviews 2 (AMSTAR-2) (Shea et al., 2017) and the Risk of Bias Assessment Tool for Systematic Reviews (ROBIS) (Whiting et al., 2016), respectively. Disagreements between researchers were solved via discussions with the last author. No systematic reviews were excluded based on the methodological quality or risk of bias assessment.

Study selection

Our initial systematic search yielded 4250 systematic reviews. After removing duplicates and screening titles and abstracts, 173 potentially relevant articles were selected for full-text assessment. After full-text evaluation, we excluded a total of 141 articles based on our inclusion criteria and retained a total of 36 systematic reviews. The PRISMA flowchart is available under the Reporting guidelines section at the end of this article.

Study type and characteristics

Of the 36 systematic reviews included in this umbrella review, most (n=29, 75%) were published after 2016, with only seven (25%) published before or during 2016. The number of original studies included in systematic reviews ranged from 5 to 143 ( Table 1), totalling 989. The majority of these were cross-sectional (n=418, 42.7%), followed by cohort (n=101, 10.3%), longitudinal studies (n=64, 6.5%), RCTs (n=61, 6.2%), and case-control studies (n=5, 0.5%) ( Table 2).

Greenspace exposure

Measures of greenspace exposure varied significantly between systematic reviews. Of the 36 systematic reviews included in this umbrella review, 22 (61%) reported more than one greenspace measure ( Table 3). The normalised difference vegetative index (NDVI) was the most common measure of greenspace, used in 21 systematic reviews (58%). This was followed by proximity to greenspace (n=12, 33%), land cover map (n=8, 22%), duration of stay in greenspace setting (n=7, 19%), quality of greenspace (n=6, 17%), tree canopy (n=5, 14%) and frequency of greenspace usage (n=4, 11%) ( Table 3).

Health outcomes

This review considered all health outcomes examined in relation to greenspace exposure. Given that over 110 health outcomes were identified ( Table 1), we categorised outcomes as: mental health and cognitive function, maternal health and birth outcomes, cardiovascular and metabolic outcomes, respiratory health and allergies, cancer, general health and quality of life (QoL), and all-cause and cause-specific mortality. Health outcomes which did not suit these categories were classified as other health outcomes ( Table 4).

Mental health and cognitive function. Overall mental health was investigated in seven (19%) systematic reviews (Carver et al., 2020; Gascon et al., 2015; Gritzka et al., 2020; Houlden et al., 2018; Shuvo et al., 2020; van den Berg et al., 2015; Zhang et al., 2017). van den Berg et al. (2015), Shuvo et al. (2020), Carver et al. (2020), and Gritzka et al. (2020) observed beneficial effects of greenspace exposure on overall mental health, but did not conduct meta-analyses due to heterogeneity in exposure and outcome measures. Similar findings were reported by Zhang et al. (2017), but the overall quality of original studies in the review was low and prone to a high risk of bias. In contrast, the relationship between surrounding greenness and overall mental health in children (Gascon et al., 2015) and adults (Gascon et al., 2015; Houlden et al., 2018) was inconclusive.

Impacts of greenspace exposure on psychological stress were also reported in seven (19%) systematic reviews (de Keijzer et al., 2020; Gritzka et al., 2020; Kondo et al., 2018; Shaffee & Abd Shukor, 2018; Shuda et al., 2020; Soga et al., 2017; Zhang et al., 2020). Greenspace (Kondo et al., 2018; Zhang et al., 2020) and nature (Shaffee & Abd Shukor, 2018; Shuda et al., 2020) exposure, as well as participation in gardening activities (Soga et al., 2017), were associated with reductions in stress. Greater long-term exposure to greenspace was also beneficial for stress (de Keijzer et al., 2020). Gritzka et al. (2020) reported only one original study on the perceived stress of employees.

A total of 11 (31%) systematic reviews investigated mental health indicators (Carver et al., 2020; Kondo et al., 2018; Oh et al., 2017; Roberts et al., 2019; Soga et al., 2017; Vanaken & Danckaerts, 2018; Wen et al., 2019; Yeo et al., 2019; Zhan et al., 2020; Zhang et al., 2020), although two of these reviews could not draw conclusions due to mixed results (Kondo et al., 2018) and inadequate evidence (Oh et al., 2017). Meta-analyses revealed that exposure to the natural environment was beneficial for depressive mood (SMD=0.38; 95%CI:0.16,0.56) (Roberts et al., 2019) and that higher prenatal greenspace exposure was associated with lower odds of mental disorders (OR=0.87; 95%CI:0.77,0.99) (Zhan et al., 2020). Other greenspace measures, such as gardening (Soga et al., 2017), urban greenspace (Kondo et al., 2018), long-term outdoor exposure (de Keijzer et al., 2020), and indoor nature interventions (Yeo et al., 2019) also had favourable effects on psychological wellbeing. In a narrative summary, Wen et al. (2019) found that negative emotions decreased after participating in forest bathing. Another review reported a positive association between mental health and greenspace exposure in older adults living in residential aged care facilities; however, health outcomes were primarily based on observations and perceptions of staff and relatives, which may be subject to information bias (Carver et al., 2020). Greenspace exposure was also beneficially associated with emotional and behavioural difficulties in children (Vanaken & Danckaerts, 2018) and psychological distress and depressive symptoms in adolescents (Zhang et al., 2020).

Cognitive function was also explored in 10 (28%) reviews (de Keijzer et al., 2020; de Keijzer et al., 2016; Gritzka et al., 2020; Islam et al., 2020; Kabisch et al., 2017; Kondo et al., 2018; Twohig-Bennett & Jones, 2018; Vanaken & Danckaerts, 2018; Whear et al., 2014; Yeo et al., 2019). Greenspace exposure was associated with improved cognitive function and development in adults (Gritzka et al., 2020; Kondo et al., 2018) and children (Islam et al., 2020; Kondo et al., 2018). Garden use was also associated with decreased agitation in people with dementia (Whear et al., 2014). Other reviews concluded that evidence on cognition was limited (de Keijzer et al., 2020; de Keijzer et al., 2016; Vanaken & Danckaerts, 2018), inconsistent (Yeo et al., 2019), or insufficient (Kabisch et al., 2017; Twohig-Bennett & Jones, 2018).

Maternal health and birth outcomes. The impact of greenspace exposure on pregnancy outcomes such as birthweight (BW), low birth weight (LBW), small for gestational age (SGA), and preterm birth (PTB), was assessed in eight (22%) systematic reviews (Akaraci et al., 2020; Dzhambov et al., 2014; Islam et al., 2020; Kabisch et al., 2017; Kondo et al., 2018; Lee et al., 2020; Twohig-Bennett & Jones, 2018; Zhan et al., 2020). Kondo et al. (2018) found insufficient evidence to support a statistical association between greenspace exposure and BW, PTB, and SGA, while Kabisch et al. (2017) reported inconsistent evidence on BW. All other reviews found weak but positive associations between higher surrounding greenness and increased BW (Akaraci et al., 2020; Dzhambov et al., 2014; Islam et al., 2020), decreased risk for LBW (Islam et al., 2020) and PTB (Twohig-Bennett & Jones, 2018), and lower odds of SGA (Akaraci et al., 2020) within different greenspace buffer sizes. For example, a meta-analysis of ten studies by Zhan et al. (2020) demonstrated protective effects of residential greenness on LBW measured by NDVI at 100 (OR=0.80; 95%CI:0.75,0.99), 300 (OR=0.82; 95%CI:0.71,0.93), and 500 (OR=0.85; 95%CI:0.77,0.93) metre buffers. Lee et al. (2020) also found that BW was associated with overall greenness. The association was determined via NDVI buffer sizes of 100 (β [beta standardised regression coefficient]=0.003; 95%CI:0.001,0.005), 250 (β=0.001; 95%CI:0.000,0.002), and 500 (β=0.002; 95%CI:0.001,0.004) metres after adjusting for air quality and civil environment.

Cardiovascular and metabolic outcomes. The risk of cardiovascular outcomes following greenspace exposure was assessed in ten (28%) systematic reviews (de Keijzer et al., 2020; Gritzka et al., 2020; Kabisch et al., 2017; Kondo et al., 2018; Lachowycz & Jones, 2011; Oh et al., 2017; Twohig-Bennett & Jones, 2018; Wen et al., 2019; Yeo et al., 2019; Yuan et al., 2021). A review of observational studies by Yuan et al. (2021) reported beneficial effects of greenspace on the risk of cardiovascular disease (CVD) events in older adult populations. The authors did not conduct a meta-analysis due to the small number of original studies on CVD events. Wen et al. (2019) reported an improvement in overall cardiovascular function and hemodynamic index in response to forest bathing. Limited evidence was found on coronary heart disease (CHD) (Lachowycz & Jones, 2011) and hypertension (de Keijzer et al., 2020; Twohig-Bennett & Jones, 2018). Gritzka et al. (2020) reported inconclusive findings on blood pressure (BP), while other reviews found only one (Kabisch et al., 2017; Kondo et al., 2018; Yeo et al., 2019) or two (Oh et al., 2017) original studies on the outcome. However, the meta-analyses by Twohig-Bennett and Jones (2018) demonstrated decreased diastolic blood pressure (DBP) (MD=-1.97; 95%CI:-3.45,-0.49), heart rate (HR) (MD=-2.57; 95%CI:-4.30,-0.83), and heart rate variability (HRV) (MD=91.87; 95%CI:50.92,132.82) following exposure to increased greenery.

Respiratory health and allergies. We identified nine (25%) systematic reviews that assessed exposure to residential greenspace on respiratory health (de Keijzer et al., 2020; Hartley et al., 2020; Islam et al., 2020; Kabisch et al., 2017; Kondo et al., 2018; Lambert et al., 2017, 2018; Oh et al., 2017; Twohig-Bennett & Jones, 2018). Lambert et al. (2017) demonstrated no association between residential greenness and allergic respiratory diseases in children and adolescents. Hartley et al. (2020) undertook an updated review of Lambert et al. (2017) and found similar results. Meta-analysis showed that high greenspace areas were associated with lower odds of asthma, although this result was based on only two original studies with substantial heterogeneity (Twohig-Bennett & Jones, 2018). Islam et al. (2020) and Lambert et al. (2018) investigated the respiratory health of children and adolescents, which revealed limited evidence to support the effects of greenspace exposure on asthma and atopic sensitisation, respectively. Kabisch et al. (2017), Kondo et al. (2018), de Keijzer et al. (2020), and Oh et al. (2017) reported insufficient original studies to draw conclusions on respiratory diseases, pulmonary function, and aeroallergens.

Cancer. Only four (11%) systematic reviews investigated the relationship between greenspace exposure and cancer (de Keijzer et al., 2020; Kabisch et al., 2017; Kondo et al., 2018; Twohig-Bennett & Jones, 2018) due to a small number of original studies on the outcome. Kondo et al. (2018) and Twohig-Bennett and Jones (2018) suggested that greenspace exposure is beneficially associated with the risk of prostate cancer after controlling for individual factors, including PA, smoking, and medical history. de Keijzer et al. (2020) and Twohig-Bennett and Jones (2018) also suggested a harmful association between greenspace exposure and skin cancer. Inconclusive findings were reported for the effect of park availability on lung cancer (Kabisch et al., 2017).

General health and Quality of life (QoL). We found 11 (31%) systematic reviews that investigated the role of greenspace exposure on self-reported general health, perceived wellbeing, and QoL (de Keijzer et al., 2020; Gritzka et al., 2020; Houlden et al., 2018; Kondo et al., 2018; Shuvo et al., 2020; Soga et al., 2017; Twohig-Bennett & Jones, 2018; van den Berg et al., 2015; Vanaken & Danckaerts, 2018; Whear et al., 2014; Yeo et al., 2019). Some reviews reported inconclusive findings on the effect of greenspace exposure on perceived (de Keijzer et al., 2020; Gritzka et al., 2020) and objective (Houlden et al., 2018) wellbeing and QoL, while another found a positive association (Vanaken & Danckaerts, 2018). Shuvo et al. (2020) concluded that there is an association between greenspace exposure and self-reported general health, but the methodological quality of studies was low. Earlier reviews revealed mixed evidence on the relationship between greenness and general health (Kondo et al., 2018; van den Berg et al., 2015) and falls (Whear et al., 2014). A meta-analysis by Twohig-Bennett and Jones (2018) demonstrated an association between greenspace exposure and self-reported health conditions (OR=1.12; 95%CI:1.05,1.19). Another two reviews considered greenspace exposure via participation in gardening activities and reported beneficial effects on self-perceived health and QoL (Soga et al., 2017; Yeo et al., 2019).

All-cause and cause-specific mortality. A total of 11 (31%) systematic reviews and meta-analyses considered all-cause or cause-specific mortality (Akaraci et al., 2020; Gascon et al., 2016; Islam et al., 2020; Kabisch et al., 2017; Kondo et al., 2018; Lachowycz & Jones, 2011; Rojas-Rueda et al., 2019; Twohig-Bennett & Jones, 2018; van den Berg et al., 2015; Yuan et al., 2021; Zhan et al., 2020). The meta-analysis by Gascon et al. (2016) found that living in areas of high greenery reduced the risk of all-cause mortality (RR=0.92; 95%CI:0.87,0.97) but not lung cancer mortality. Three meta-analyses performed after Gascon et al. (2016) presented similar findings (Rojas-Rueda et al., 2019; Twohig-Bennett & Jones, 2018; Yuan et al., 2021), except for CVD mortality (Yuan et al., 2021). Yuan et al. (2021) also reported that increased NDVI is associated with reduced risk of stroke mortality (HR=0.77; 95%CI:0.59,1.00). Narrative syntheses summarised moderate to strong effects of increased greenness on the risk of the following mortalities: non-accidental (van den Berg et al., 2015), CVD (Kondo et al., 2018), respiratory (Kondo et al., 2018), heat-wave (Kabisch et al., 2017) and overall cancer (Kondo et al., 2018; Twohig-Bennett & Jones, 2018). Insufficient evidence was found on neonatal (Akaraci et al., 2020), infant (Islam et al., 2020; Zhan et al., 2020), and circulatory disease (Lachowycz & Jones, 2011) mortality for synthesis.

Other health outcomes. We identified 16 (44%) systematic reviews that evaluated the effect of greenspace on other health outcomes. Inconsistent associations (Kabisch et al., 2017; Kondo et al., 2018; Lachowycz & Jones, 2011; Yeo et al., 2019) or no association (Gritzka et al., 2020) were reported for greenspace exposure and obesity-related health indicators. In contrast, a meta-analysis by Luo et al. (2020) reported higher NDVI to be associated with lower odds of being overweight or obese (OR=0.88; 95%CI:0.84,0.91). Other reviews found inadequate evidence on weight status (de Keijzer et al., 2020; Schulz et al., 2018; Shuvo et al., 2020). Kondo et al. (2018) reported mixed results on the association between urban greenspace exposure and cortisol concentration, while other reviews found insufficient evidence on serum (Gritzka et al., 2020; Oh et al., 2017) and salivary (Shuda et al., 2020) cortisol. In comparison, a meta-analysis by Twohig-Bennett and Jones (2018) showed that salivary cortisol was lower (MD=-0.05; 95%CI:-0.07,-0.04) when exposed to higher levels of greenspace. Narrative synthesis by Wen et al. (2019) also revealed a decrease in cortisol and adrenaline in response to forest bathing.

Quantitative data summary

Quantitative data collected from meta-analyses were collated according to outcome category, outcome, and measure of greenspace exposure ( Table 5). Effect measures reported in meta-analyses which were based on a singular original study were excluded from the quantitative data summary table. No meta-analyses were reported for cancer outcomes.

Quality of Evidence and the risk of bias

Half of the systematic reviews (50%) had a high risk of bias, and three (8%) reviews were categorised as having ‘unclear risk’. The remaining 15 (42%) reviews were deemed low risk using the ROBIS tool ( Figure 1; extended data: Supplementary Table S2).

Figure 1. Risk of bias assessment of included reviews using ROBIS.

Rows 1-4 show judgements on individual criteria while the bottom row shows the overall risk of bias.

Of the 36 systematic reviews, eight (22%) were rated as high quality, 15 (42%) as moderate quality, and 13 (36%) as low quality according to the AMSTAR-2 tool ( Table 6; extended data: Supplementary Figure S1). No reviews were rated as critically low-quality.