Urban Greenness Associated with Fewer Mental-Health Hospital Admissions — But Effects Vary Across Countries

Highlight

– A pooled multicountry time-series analysis (11.4 million mental-health hospital admissions, 2000–2019) found a 0.1-unit increase in NDVI was associated with a 7% lower risk of hospital admission for all-cause mental disorders (RR 0.93, 95% CI 0.89–0.98) in pooled analyses.
– Associations were heterogeneous by country and disorder type; consistent protective findings in Brazil, Chile, and Thailand, but modest adverse associations in Australia and Canada for some outcomes.
– Benefits were most consistent in urban settings: an estimated 7,712 urban admissions annually were statistically attributable to observed greenness levels; scenario modelling suggested a 10% greenness increase could yield widely varying reductions in admissions between countries.

Background and clinical context

Mental disorders contribute a large and growing global burden of disease, affecting quality of life, healthcare utilisation, and productivity. Hospital admissions for mental disorders represent severe manifestations or crises requiring inpatient care and are sensitive to both health-system factors and underlying population risk. Interest in environmental determinants of mental health has increased, with green space exposure (quantified using satellite-derived metrics such as the normalized difference vegetation index, NDVI) proposed as a modifiable upstream factor that could reduce population mental-health burden via multiple pathways: stress reduction, increased physical activity, social cohesion, air-quality and heat-island mitigation, and restoration of cognitive resources.

Study design and methods

The BMJ multicountry time-series study analysed 11.4 million hospital admissions for mental disorders across 6,842 locations in seven countries (Australia, Brazil, Canada, Chile, New Zealand, South Korea, and Thailand) from 2000 to 2019. Outcomes were daily counts of hospital admissions for all-cause mental disorders and six specific categories: psychotic disorders, substance use disorders, mood disorders, behavioural disorders, dementia, and anxiety.

Greenness exposure was measured using NDVI values. The authors used quasi-Poisson regression models to estimate short-term associations between greenness and hospital admissions, controlling for temperature, relative humidity, major air pollutants, socioeconomic indicators, seasonality, and long-term trends. Models were stratified by sex, age group, urbanisation (urban vs non-urban), and season. The paper also performed exposure–response analyses and estimated attributable admissions under defined greenness-change scenarios (including a 10% increase in greenness) with results reported for urban settings.

Key findings

Overall pooled association

In the pooled, multicountry analysis, a 0.1-unit increase in NDVI was associated with a 7% reduction in the risk of hospital admission for all-cause mental disorders (relative risk 0.93, 95% CI 0.89 to 0.98). Exposure–response analyses suggested a generally monotonic and approximately linear relationship without clear thresholds.

Distribution of admissions by disorder category

Across the study period, of the mental-health-related hospital admissions analysed, 30.8% were for psychotic disorders, 24.7% for substance use disorders, 11.6% for mood disorders, 7.4% for behavioural disorders, 3.0% for dementia, and 2.5% for anxiety.

Heterogeneity by country and disorder

Associations varied substantially across countries and by disorder category. Brazil, Chile, and Thailand showed consistent protective associations across most disorder categories. In contrast, Australia and Canada showed modest adverse (harmful) associations for all-cause mental disorder admissions and for some specific categories. South Korea and New Zealand had mixed results, with some protective associations depending on outcome and model specification.

Urban vs non-urban differences

Analyses restricted to urban settings demonstrated more consistent protective associations. Using observed greenness levels, the authors estimated that 7,712 (95% CI 6,701 to 8,726) hospital admissions for mental-health disorders annually in urban areas were statistically attributable to greenness (that is, could be linked to current greenness exposures when compared with lower greenness counterfactuals used in the model).

Scenario modelling

Projected impacts of increasing urban greenness differed markedly between countries. The estimated reduction in hospital admissions for a 10% increase in greenness ranged from approximately 1 per 100,000 population in South Korea to roughly 1,000 per 100,000 in New Zealand, reflecting baseline greenness, admission rates, sociodemographic context, and the modelled exposure–response in each setting.

Subgroup findings

Results varied by sex, age, and season in location-specific analyses; however, the paper reports that the protective patterns were most robust in urban populations. The heterogeneity across strata and countries implies that local context strongly modifies the relationship between greenness and severe mental-health outcomes.

Expert commentary, mechanisms, and limitations

Biological and social plausibility

Several plausible mechanisms link greenness exposure to improved mental health. Natural environments can reduce physiological stress responses (lower cortisol, blood pressure), encourage physical activity (a well-recognised antidepressant), foster social interaction and community cohesion, mitigate exposure to air pollution and extreme heat, and provide restorative cognitive experiences. These pathways are described in prior narrative and systematic summaries (for example, Hartig et al., Annu Rev Public Health 2014).

Interpretation of heterogeneity

The heterogeneity reported across countries and disorders merits careful interpretation. Differences likely reflect a combination of: variability in baseline greenness and its spatial distribution; differing healthcare utilisation and admission thresholds; heterogeneity in the NDVI metric’s ability to capture accessible or high-quality green space (NDVI measures vegetation density but not usability or types of vegetation); socioeconomic confounding that may remain despite model adjustments; cultural differences in care-seeking; and potential ecological bias in time-series analyses aggregated at location level.

Limitations of the study

Key limitations include exposure misclassification—NDVI is a proxy for greenness and does not differentiate parks from inaccessible vegetation or account for time spent by individuals in green spaces. Hospital admissions capture severe cases and are influenced by health-system capacity, admission criteria, and coding practices, so they are not direct measures of population mental-health prevalence or morbidity. Although models adjusted for many covariates (weather, air pollution, socioeconomic indicators), residual confounding cannot be excluded. The study’s time-series design is strong for short-term associations but limits causal inference about long-term residential greenness effects. Finally, observed adverse associations in some countries underline the possibility of unmeasured effect modifiers or data artefacts and argue for caution before generalising results.

Clinical and policy implications

For clinicians and health-system planners, these findings add to a growing evidence base that urban greening could be a population-level intervention with mental-health benefits, especially for severe outcomes that result in hospital care. However, the cross-country heterogeneity indicates that local evaluation is essential. Greening interventions should be coupled with equity-focused planning to ensure benefits reach disadvantaged communities and avoid unintended consequences such as green gentrification. Targeted interventions—improving access to safe, usable green spaces in densely populated urban neighbourhoods—may be the most pragmatic application.

Research implications

Further work is needed to: (1) disentangle short-term from long-term greenness effects on mental health; (2) refine exposure assessment to capture accessibility, quality, and use of green space; (3) use quasi-experimental designs (natural experiments, stepped-wedge evaluations) or randomised interventions where feasible to strengthen causal inference; and (4) explore mechanistic mediators (physical activity, social cohesion, air quality, noise, temperature) and effect modifiers (age, socioeconomic status, urban form).

Conclusion

The BMJ multicountry time-series analysis suggests that greater surrounding greenness is associated with lower rates of hospital admissions for mental disorders in pooled analyses and—most consistently—in urban settings. However, associations vary by country and disorder category, and some adverse associations were observed. Policymakers should consider context-specific evidence when planning urban greening policies, and further rigorous evaluation is needed to clarify causality, mechanism, and equitable implementation.

Funding and trial registration

Funding details and study registration, where applicable, are reported in the original BMJ article (Ye et al., 2025). Readers should consult the source publication for complete funding statements and any disclosures.

References

1. Ye T, Huang W, Xu Z, et al. Greenness and hospital admissions for cause specific mental disorders: multicountry time series study. BMJ. 2025 Nov 5;391:e084618. doi:10.1136/bmj-2025-084618. PMID: 41193238; PMCID: PMC12587678.

2. Hartig T, Mitchell R, de Vries S, Frumkin H. Nature and health. Annu Rev Public Health. 2014;35:207–228. doi:10.1146/annurev-publhealth-032013-182443.

3. World Health Organization. Mental health: strengthening our response. WHO Fact sheet. 2019 (updated 2022). Available from: https://www.who.int/news-room/fact-sheets/detail/mental-health-strengthening-our-response

AI image prompt for article thumbnail

“Aerial view of a contemporary urban neighbourhood with abundant trees, parks, and green roofs blending into residential blocks; in the foreground a subtle hospital silhouette and a translucent brain icon; soft calming greens and blues, high-detail, clean graphic composition suitable for a scientific magazine cover.”