Global disparities in urban green space use during the COVID-19 pandemic from a systematic review

By 2050, the global urban population is projected to grow by 2.2 billion, intensifying challenges for green, healthy and resilient cities. Although outdoor exposure benefits urban physical and mental health, UGS are under pressure from densification and growth, with inequitable distribution disadvantaging less privileged groups. COVID-19 highlighted UGS as critical for mitigating social and health challenges. Yet, local studies reported mixed changes in UGS use during and after lockdowns, and it remains unclear how patterns varied across regions, user groups, and UGS types. Access disparities within cities and between Global North and South, and broader economic conditions (GDP per capita) may shape UGS availability and use. Policy responses to COVID-19, measured by the stringency index (closures, movement restrictions, travel bans), likely influenced UGS use, but effects across diverse UGS remain uncertain. This study conducts a systematic review to quantify global changes in UGS use during different COVID-19 phases, assess relationships with policy stringency and GDP per capita, examine equity in accessibility and adaptation strategies, and compare changes across UGS types. The authors included studies in five languages and from multiple disciplines and methods, with critical appraisal of study validity. The research questions were: (1) How did UGS use change globally and what is the geographic coverage of available studies? (2) How does change in UGS use relate to a COVID-19 policy stringency index and GDP per capita? (3) How do changes in use relate to equity in UGS accessibility and adaptation strategies? (4) How do changes in UGS relate to different UGS types?

Prior reviews partially addressed UGS use during COVID-19 but were limited by narrative syntheses and incomplete or early evidence bases while primary studies were still emerging. Evidence has shown both decreases during lockdowns and increases during and after, with substantial local variability. Broader literature indicates UGS access is unequally distributed within and across cities, with links to socio-economic status and a correlation between wealth (GDP per capita) and UGS availability outside the COVID-19 context. Existing global analyses highlight a North–South divide in exposure to greenspace and suggest that governance, socio-cultural contexts, and urban form influence UGS use. However, a comprehensive synthesis linking policy stringency, economic conditions, equity aspects, and UGS types at global scale was lacking, motivating this systematic review.

Design: Systematic review adhering to Collaboration for Environmental Evidence guidelines and PRISMA standards, with an a priori protocol. Scope and PECO: Population—global urban populations under COVID-19 restrictions; Exposure—use of green spaces; Comparators—before vs during and after first lockdowns; Outcomes—changes in use, behavior, and perception. Search strategy: Comprehensive searches (initial 13 June 2022; alerts until 27 August 2022) in Web of Science, Scopus, CNKI, and Google Scholar, covering Chinese, English, French, German, and Spanish. Included peer-reviewed and gray literature. Search strings combined terms for settlement type, green space type, COVID-19, and use-related terms; optimized with litsearchr; translated to other languages (with some adaptations). Restricted to publications from 2020 onward. Screening and eligibility: Two-stage screening (titles/abstracts, then full text) by two independent reviewers; disagreements resolved with discussion/third reviewer. Inclusion criteria: conducted in countries affected by COVID-19; examined changes in UGS use; referenced pre- vs during/after lockdown; included actual/stated uses or direct evidence; quantitative or qualitative data; languages within scope. Excluded reviews, anecdotal-only studies, and those assuming change without data. Gray literature screened similarly. PRISMA flow documented. Critical appraisal: Internal validity rated on a 0–5 Likert scale (quantitative: sampling, analysis adequacy, baseline controls, confounders considered and handled; qualitative: sampling plus transparency, credibility, reflexivity, transferability). Studies failing minimum criteria were excluded. Scores used as weights in analyses; validity displayed in figures. Data extraction: Predefined variables extracted for each study/city and UGS type: location, period (before/during/after first lockdown), UGS category (seven types), method category, sample size, study period, percentage change, proportions increasing/decreasing/no change, and qualitative direction (increase/decrease/no change). Multi-location and multi-UGS studies disaggregated. Numerical data extracted from text/tables/graphs (graphreader used when necessary). Qualitative data on equity and behavioral adaptations coded into themes. Contextual variables: Policy stringency from Oxford COVID-19 Government Response Tracker (mean 2020 values; national or subnational for US, Canada, UK, Australia). GDP per capita from a gridded 2015 dataset (5 arcmin resolution). Some modifiers (e.g., fine-scale UGS location) not consistently reported and thus not analyzed beyond qualitative themes. Data analysis: Geocoded cities and mapped in Equal Earth projection (QGIS). Primary analyses at city level; for conflicting UGS-type results within a city, averaged to avoid dominance by highly detailed studies. For UGS-type comparisons, analyzed each type separately. Direction of change coded as increase/decrease/no change based on majority responses or reported overall change. Logistic regressions (binomial GLM) compared stringency and GDP between decreased vs increased cases, weighted by study validity; cases of no change excluded from the regression but tested in sensitivity analyses. Magnitudes of change (subset of quantitative surveys with comparable percent change and sample sizes) plotted against stringency and GDP, weighted by sample size. UGS-type differences assessed via weighted chi-square tests (weights by study score) and visualized with mosaic plots. Sensitivity analyses included restricting to high-validity studies (scores 4–5) and alternative binary codings incorporating the no-change category. Publication bias discussed descriptively.

Study corpus and coverage: From 3,310 initially screened articles, 189 were included; 12 excluded at appraisal, yielding 177 articles covering 237 cities across 60 countries. Languages: 223 locations from English-language studies, 7 Chinese, 5 German, 1 Spanish, 1 French. Validity: 80% of locations had low/medium validity scores (1–3); 20% scored 4–5. Global changes in use: Before–during lockdowns: 60% of locations reported decreased UGS use; 37% increased; 3% no change. Increases/unchanged concentrated in Central/Northern Europe, with additional cases in North America, East/Southeast Asia, Australia, New Zealand, and Chile. Before–after lockdowns: 34% decreased; 57% increased; 9% no change. Europe, East Asia, and Australia mostly increased; the Americas mixed. Geographic evidence skewed toward Europe, Australia, North America; few studies from Africa, South/Central Asia, Latin America. Policy stringency: Weighted logistic regression showed locations with decreased use had significantly higher COVID-19 policy stringency in 2020 than those with increased use before–during (P<0.001) and before–after (P<0.001). However, in high-validity subsets these differences were not significant, and 95% confidence intervals overlapped, tempering inference. GDP per capita: Higher GDP per capita associated with increased UGS use. Logistic regression indicated significantly higher GDP where increases were reported before–during (P<0.001; high-validity subset P=0.028) and before–after (P<0.001). Non-overlapping 95% CIs in GDP comparisons supported these trends. Magnitudes of change: Among 36 quantitative-survey studies, relative changes ranged from +50% to −95% (before–during) and +61% to −70% (before–after). Other methods (social media/GPS) reported even larger increases, up to +291% during and +240% after lockdowns. Visual trends suggested stronger decreases in lower-GDP contexts and greater increases in higher-GDP contexts; no clear visual pattern with stringency for magnitude plots. Equity and behavior: Lower-income, socially vulnerable, some non-white communities, women, and older adults faced greater barriers to UGS access. Behavioral adaptations included changes in frequency, duration, group composition, activity types, timing, travel distance, and UGS type; divergent behaviors observed within populations. UGS type differences: Before–during lockdowns, private gardens (100% increased/unchanged), UGS near home (78%), and forests/natural areas (71%) predominantly increased or stayed stable, while public parks (71%) and historic gardens (100%) mostly decreased. Before–after lockdowns, private/public gardens, forests/natural areas, and UGS near home largely increased or stayed stable (100%, 100%, 100%, 86% respectively), while decreases persisted in unspecified UGS and public parks (54% and 30%). Weighted chi-square tests confirmed significant differences across UGS types (P<0.001 before–during; P=0.015 before–after).

Findings reveal heterogeneous, location-specific changes in UGS use during COVID-19. Decreases were more frequent where policy stringency was higher, but this effect weakened in high-validity subsets, indicating that governance stringency was not the sole driver. Economic context mattered: higher GDP per capita correlated with increased UGS use, consistent with greater availability/quality of UGS and capacity to accommodate demand, while lower-GDP contexts more often saw decreases. Equity dimensions influenced who could access and benefit from UGS; vulnerable groups frequently faced greater constraints, potentially exacerbating health and well-being disparities. Behavioral adaptations mitigated some constraints, with shifts in timing, destinations, and activities, and increased reliance on nearby and less crowded UGS. UGS types mattered: public parks—especially central, crowded parks—saw more decreases during strict lockdowns due to closures and avoidance, whereas private gardens, UGS near homes, and forests/natural areas served as key refuges and experienced increases. After restrictions eased, many places saw rebounds or increases, suggesting compensatory use replacing other curtailed activities. The geographic distribution of evidence and outcomes underscores a North–South divide, both in study coverage and in patterns of UGS use, aligning with broader inequalities in greenspace access. Collectively, results support prioritizing diverse UGS provision—beyond public parks alone—to enhance urban resilience and equity in future crises.

The review provides a comprehensive global synthesis of how UGS use changed during COVID-19, linking patterns to policy stringency, economic conditions, equity, and UGS type. Overall, UGS use often decreased under strict policies and in lower-GDP contexts, while higher-GDP regions and certain UGS types (private gardens, nearby UGS, forests/natural areas) saw increases. Equity gaps were pronounced, with vulnerable groups facing greater barriers. Recommendations for resilient, equitable urban futures include: (1) increasing provision and recognition of small domestic and community gardens near housing (including rooftops); (2) enabling safe access to informal UGS near residences (vacant land, vegetated corridors); and (3) preserving and restoring forests and natural areas within/at city edges with appropriate access. These actions can help make cities greener, healthier, and more resilient to future crises (e.g., climate change, pandemics). Future research should monitor long-term post-pandemic trajectories of UGS use, expand high-validity studies in underrepresented regions, and standardize measures to enable stronger meta-analyses.

- Methodological heterogeneity across studies (qualitative/quantitative, varied metrics, units, sampling frames) limited formal meta-analyses and comparability. - Evidence base unevenly distributed geographically, with sparse coverage and few high-validity studies in the Global South; Africa and parts of Latin America/Asia were underrepresented. - Many studies had low-to-medium internal validity scores; associations with policy stringency were not consistently supported in high-validity subsets. - Reliance in some studies on broad data sources (e.g., Google Mobility “park” category) with generic definitions may diverge from other methods; Google Trends (intent) studies were excluded. - Limited availability of consistent contextual modifiers (e.g., fine-scale UGS location, individual demographics) constrained analysis of effect modification. - Timeframe focused on first lockdowns in 2020; long-term persistence of changes remains uncertain. - Potential publication bias toward reporting change vs no change, though no clear bias expected between increase vs decrease.