Residential green spaces and neurodegenerative diseases in middle-aged and older adults: a systematic review and dose-response meta-analysis.

Introduction

Neurodegenerative diseases (NDs) are a heterogeneous group of disorders characterized by the progressive loss of neurons, with clinical manifestations spanning motor, sensory, and cognitive dysfunction [ 1 – 3 ]. Age is the most significant non-genetic risk factor for NDs [ 4 ]. With the accelerating aging of the global population—life expectancy is projected to exceed 85 years for both sexes in several countries and 80 years in 59 countries (including China) by 2040 [ 5 ]—the global burden of NDs is rising sharply. According to the latest 2021 Global Burden of Disease (GBD) data, the incidence of NDs is expected to continue climbing in regions with rapidly aging populations [ 6 ]. The prevalence of major NDs, such as Alzheimer’s disease (AD) [ 7 ]and Parkinson’s disease (PD) [ 8 ], increases exponentially with age. The number of people with dementia worldwide is projected to surpass 150 million by 2050, imposing an economic burden of up to $10 trillion [ 9 – 11 ]. Given that the timing of neuronal injury significantly impacts long-term survival [ 12 ], early preventive interventions are crucial for altering the natural course of NDs.

As there are currently no curative treatments for NDs, the medical paradigm is shifting from reactive treatment to proactive prevention. Beyond non-modifiable factors like aging [ 13 – 15 ]and genetic susceptibility [ 16 , 17 ], modifiable environmental factors (e.g., air pollution [ 18 ], noise exposure [ 19 ]) and lifestyle choices [ 20 ] have become key targets for prevention research. In this context, urban green space, as a modifiable component of the built environment, is gaining increasing attention in public health for its potential multi-faceted health benefits, including improving air quality, mitigating the urban heat island effect, promoting physical activity, fostering social cohesion, and reducing psychological stress [ 21 , 22 ].

Currently, residential green space exposure is commonly quantified using objective geographic methods. One of the most prevalent metrics is the Normalized Difference Vegetation Index (NDVI), which assesses the density and health of vegetation. This index is calculated using the reflectance values of near-infrared and red light, typically derived from satellite imagery datasets such as those provided by the NASA Landsat program. Another widely used approach involves analyzing Land Use/Land Cover (LU/LC) data to quantify the proportion of land area designated as green space (e.g., parks, forests) within a predefined buffer around a residence. These methods and the underlying data enable robust quantification of environmental exposure for large-scale epidemiological studies.

Existing epidemiological evidence has generally supported a protective role of residential green space on neurological health. For instance, a cohort study in Belgium [ 22 ] reported that higher levels of residential greenness were associated with lower mortality from neurodegenerative diseases, while a study in the United States [ 23 ] found that greater exposure to greenness, parks, and blue space was associated with reduced hospitalizations for neurodegenerative diseases. However, the findings are not entirely consistent. A study based on the Lothian Birth Cohort 1936 in the United Kingdom [ 24 ] and another study conducted in England [ 25 ] reported mixed or limited evidence regarding associations between green space and cognitive impairment or dementia.

Although some reviews have explored the link between environmental factors and specific NDs like AD, systematic reviews covering the full spectrum of NDs are scarce. To date, only the reviews by Zagnoli et al. [ 26 ] and Da et al. [ 27 ] have systematically examined the association between residential green space and dementia, with both reporting a protective effect. However, these studies have limitations concerning their inclusion criteria (e.g., restricted to specific ND types), exposure definitions (e.g., mixing subjective and objective measures), and lack of dose-response analysis. Therefore, this study aims to conduct a comprehensive systematic review and meta-analysis to thoroughly investigate the association between residential green space exposure and the risk of various NDs in middle-aged and older adults, including the first dose-response analysis, to provide robust evidence for urban planning and public health policy.

Methods

Registration of review protocol

This systematic review and meta-analysis adhered strictly to the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [ 28 , 29 ], analyzing observational studies that investigated associations between residential green space and NDs risk. The study protocol received prospective registration in the International Prospective Register of Systematic Reviews (PROSPERO; registration number: CRD420251020592).

Search strategy

We systematically searched PubMed, Embase, Web of Science, and Cochrane Library databases for relevant publications published through March 15, 2025, restricting our search to English-language publications. Our search strategy incorporated both controlled vocabulary (MeSH terms for PubMed, Emtree for Embase) and free-text terms. The search combined terms for (1) green space exposure (“urban greenness,” “park,” “vegetation index,” “NDVI”) and (2) neurodegenerative outcomes (“neurodegenerative diseases,” “Alzheimer’s disease,” “Parkinson’s disease”). The search strategy, adapted for each database, included a combination of subject headings (e.g., MeSH for PubMed and Emtree for Embase) and free-text words. The full search strategy for each database is provided in the Supplemental Material, where terms were optimized to capture the concept of green space most effectively within each specific database’s indexing system. Additionally, we performed manual searches of reference lists, review articles, and citation networks to identify potentially eligible studies not captured by database searches.

Eligibility criteria

Based on the PECOS framework [ 30 ], the following inclusion criteria were applied: (1) population: aged ≥ 45 years at baseline without pre-existing neurodegenerative conditions; (2) exposure: the study had to assess exposure to residential green space for a period relevant to the development of NDs, using objective methods such as the NDVI or LU/LC data. For prospective cohort studies, this required exposure assessment at baseline before the follow-up period. For case-control and retrospective cohort studies, we included studies where the exposure assessment was intended to represent a long-term or etiologically relevant period prior to diagnosis, even if perfect temporal precedence could not be guaranteed (e.g., using environmental data from a representative year for a multi-year study period). For cross-sectional studies, exposure assessment was required to reflect habitual or long-term residential green space conditions corresponding to the time of outcome assessment; (3) comparator: comparisons against clearly defined reference groups (lowest exposure quartile/no significant greenery) with quantitative thresholds; (4) outcomes: outcomes encompassing clinically diagnosed NDs (Parkinson’s disease, Alzheimer’s disease, mild cognitive impairment, and other dementia subtypes), excluding studies that reported only mortality endpoints without incidence or morbidity outcomes; (5) research design: observational studies (including longitudinal studies, cross-sectional studies, ecological studies, and case-control studies.)

The exclusion criteria were as follows: (1) duplicate publications or overlapping datasets from identical cohorts; (2) studies without full-text access or where data cannot be obtained. This included studies that did not report effect estimates (e.g., odds ratio, hazard ratio) with corresponding 95% confidence intervals for the meta-analysis; (3) studies examining non-relevant environmental exposures; (4) outcomes not specific to NDs (e.g., all-cause dementia or pure vascular dementia without neurodegenerative components); and (5) ineligible study designs (including animal research, clinical trials, protocols, reviews, commentaries, conference abstracts, or case series with fewer than 5 cases).

Greening exposure index

A “green space exposure index” is a collective term for metrics or indices used to quantify exposure to residential green space [ 31 ]. Among these, the NDVI and LU/LC classifications are the most commonly used indicators. Over the past decade, most studies have used NDVI to quantify vegetation density. NDVI data are available from multiple satellite platforms, such as the Landsat program by the National Aeronautics and Space Administration (NASA). NDVI is a remote sensing index that estimates vegetation density by calculating the difference between near-infrared and visible light reflectance, divided by their sum. The NDVI is calculated for each pixel of a satellite image, yielding values that range from − 1 to + 1. A value close to 0 indicates the absence of greenness, a value approaching + 1 signifies the highest density of vegetation, and a value near − 1 suggests the presence of water bodies. As a reliable estimator of greenness, NDVI captures both the proximity and extent of vegetated areas. This assessment method is objective as it does not rely on self-reported data, making it suitable for evaluating the impact of green space on NDs [ 32 – 34 ]. Land Use (LU) refers to how land is utilized by humans, while Land Cover (LC) describes the physical and biological surface characteristics of the land. Studies based on LU/LC data typically quantify the proportion of specific land cover types within a predefined buffer zone around a participant’s residence. Residential green space exposure is then estimated by calculating the percentage of green land cover types (e.g., parks, forests) within these buffers (e.g., 300–500 m radius) surrounding participants’ addresses [ 35 , 36 ].

Study selection

Two independent reviewers (SCY and LJ) conducted literature screening using EndNote X9 (Clarivate Analytics), performing duplicate removal followed by title/abstract review against the predefined inclusion/exclusion criteria. Potentially eligible studies underwent full-text assessment for final inclusion determination. Inter-reviewer discrepancies were resolved via consultation with a senior researcher (SJ), with consensus required for final decisions.

Data extraction

Two investigators (SCY and LJ) independently extracted data, with discrepancies resolved through consultation with a third reviewer (SJ). When information was unclear or missing, we contacted corresponding authors for clarification or additional data. Extracted variables included:

study identification: first author, publication year, country; study design: data source, duration, mean follow-up; population characteristics: sample size, sex distribution, mean age; exposure metrics: residential green space assessment method, specific index, exposure categories; outcomes: neurodegenerative disease endpoints; effect estimates: effect measures (e.g., odds ratios [ORs], hazard ratios [HRs], or relative risks [RRs]) with their corresponding 95% confidence intervals (CIs); dose-response data: the number of cases and participants/person-years, and the mean/median exposure level for each category; adjusted covariates: variables adjusted for in the statistical model.

Data analysis and synthesis

This systematic review and meta-analysis synthesized evidence from observational studies investigating associations between residential green space exposure and ND risk. We focused specifically on incidence outcomes, excluding studies reporting only mortality endpoints [ 22 ]. Under the rare disease assumption, OR and RR are considered valid approximations of the HR [ 37 ]. Given the low incidence of NDs, and consistent with methodologies used in prior meta-analyses [ 38 , 39 ], we pooled all three effect measures. Consequently, we adopted the HR and its 95% Confidence Interval (CI) as the summary effect measure. When multiple buffer radii were reported, we selected estimates from the most extreme exposure contrasts for primary analysis [ 26 ]. Specifically, for studies reporting quantiles (e.g., quartiles), we extracted the data for the highest quantile versus the lowest. For studies that presented results for multiple buffer zones (e.g., 300 m, 500 m, 1000 m), we established an a priori selection hierarchy: we prioritized data from the 500 m buffer, followed by the 300 m and 1000 m buffers, respectively. This hierarchy was based on the frequency of reporting in the literature and the representation of a plausible neighborhood walking distance. Some cohorts contributed effect estimates for more than one neurodegenerative disease outcome (e.g., both AD and PD). To address the statistical non-independence of these data, we performed a sensitivity analysis by excluding studies contributing multiple outcomes to evaluate their impact on the overall pooled estimate.

We employed DerSimonian-Laird random-effects models [ 40 ]to compare ND risk between extreme exposure categories and examine potential dose-response relationships. Heterogeneity was evaluated using Cochran’s Q test (significance threshold p < 0.10) and I² statistics, with results visualized in forest plots (95% CIs). To investigate potential sources of heterogeneity and explore possible effect modification, we conducted several pre-specified subgroup analyses. The variables were selected based on their clinical and methodological importance. Disease Characteristics: NDs type. Population Characteristics: Key demographic and behavioral factors, including sex, age, and smoking status, where sufficient data were available. Methodological Characteristics: Key parameters of exposure assessment, including the buffer zone radius used to define residential green space and the increment of NDVI reported in the primary studies (e.g., per 0.1 unit increase). Meta-regression analysis was performed to evaluate the association of covariates with greening and ND risk. Sensitivity analyses employed leave-one-out cross-validation to evaluate individual study influence. Publication bias was assessed through funnel plot symmetry and Egger’s regression test [ 41 ].

Dose-response analyses were restricted to studies quantifying green space exposure using the NDVI. This restriction was applied to ensure methodological homogeneity, as NDVI provides a standardized and continuous measure of greenness, which is essential for a valid quantitative synthesis of the dose-response relationship across different studies. For the dose-response meta-analysis, we extracted the effect estimates and their corresponding 95% confidence intervals for at least three quantitative exposure categories. If a study provided both continuous (e.g., per 0.1-unit increase in NDVI) and categorical estimates, we used the categorical data for the dose-response analysis. Studies that only reported a continuous effect were not included in the quantitative dose-response synthesis.

Linear trends were estimated via generalized least squares for trend (GLST) methods [ 42 ], accounting for within-study correlation. To model exposure-response relationships that weren’t linear, restricted cubic splines with knots at Harrell’s recommended percentiles (10th, 50th, and 90th) were used. This method combines slopes from different studies into a single exposure-response curve, which makes it possible to find patterns that don’t follow a straight line [ 43 ]. We assessed nonlinearity by testing whether the second spline coefficient differed significantly from zero (Wald test). This methodology has demonstrated reliability in prior meta-analyses [ 39 ]. Statistical analyses were performed using Stata 16 for meta-analysis and R software (version 4.4.1; R Foundation for Statistical Computing) with the “dosresmeta” and “mvmeta” packages for dose-response analysis, along with the “ggplot2” package for visualization.

Quality assessment of studies

The quality assessment of included studies was conducted based on the methodology proposed by Da et al. (Supplementary table S2) [ 27 ]. For case-control, cohort, and cross-sectional studies, we utilized the Newcastle-Ottawa Scale (NOS) and its adapted versions [ 44 , 45 ]. Studies were categorized by score as low (0–3), moderate (4–6), or high (7–9) quality. The detailed criteria for the adapted cross-sectional tool are available in Supplemental Table 2. For the single retrospective cohort study with a semi-ecological exposure assessment [ 46 ]. We acknowledge that, at the time of our assessment, we applied a preliminary version of the ROBINS-E tool (Supplementary table S3) [ 47 ], as the finalized version had not yet been fully operationalized. This approach has now been explicitly stated to improve transparency. The results from ROBINS-E were categorized as low, moderate, serious, or critical risk of bias. For studies with mixed design characteristics (e.g., retrospective cohort studies with semi-ecological exposure assessment), the ROBINS-E tool was applied based on the primary study design, while additional consideration was given to potential bias arising from aggregated exposure assessment. All assessments were conducted independently by two investigators, with disagreements resolved through discussion with a senior researcher.

We assessed the overall certainty of evidence for each main outcome using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework. The certainty of evidence reflects the extent of our confidence that the estimated effect is close to the true effect. According to the GRADE approach, a body of evidence originating from observational studies starts with a “low” level of certainty. The certainty level can be downgraded based on five domains: (1) risk of bias in the included studies, (2) inconsistency of results (i.e., unexplained heterogeneity), (3) indirectness of evidence, (4) imprecision of the effect estimate, and (5) publication bias. Conversely, the certainty level can be upgraded based on three domains: (1) a large magnitude of effect, (2) evidence of a dose-response gradient, and (3) if all plausible residual confounding would reduce a demonstrated effect. The final certainty of evidence was categorized as high, moderate, low, or very low.

Results

Study selection

Our systematic search identified 20,725 records from databases and citation sources (Fig. 1 ). Following duplicate removal and title/abstract screening, 32 potentially eligible articles underwent full-text review. Twenty studies were excluded during full-text assessment (see Supplementary Table S1 for exclusion rationale and study details). Twelve studies met all inclusion criteria and were included in the final meta-analysis of green space exposure and neurodegenerative disease risk.

Study characteristics and quality

Table 1 summarizes the main features and results of the 12 included observational studies [ 46 , 48 – 58 ]. These studies were published between 2019 and 2024. Among them, five studies were conducted in North America (four in the United States [ 46 , 52 , 54 , 58 ] and one in Canada [ 49 ]), six in Asia (five in China [ 48 , 50 , 51 , 53 , 57 ] and one in South Korea [ 55 ]), and one in the United Kingdom [ 56 ]. In terms of study design, six studies utilized a cohort design [ 53 – 58 ], three utilized a cross-sectional design [ 50 – 52 ], one was a case-control study [ 48 ], one combined a cross-sectional and cohort design [ 49 ], and one was an ecological study [ 46 ]. In addition, the total sample size of the included studies exceeded 1.7 million, in which the minimum baseline age of the studies was 48.9 and the maximum was 100.3, which, according to the World Health Organization’s definition of age, suggests that our participant group was predominantly middle-aged and older.

Among the 12 studies included in the systematic review, 9 utilized the NDVI to assess greening exposure [ 49 – 55 , 57 , 58 ], while 3 employed LU/LC datasets to classify green space [ 46 , 48 , 56 ]. The evaluation of greening buffers surrounding residential areas across studies was performed at differing buffer radii: one study quantified greening exposure within a 100-meter (m) buffer around residences [ 49 ], another assessed residential green space within a 200-m radius [ 54 ], two studies examined exposure at 500 m [ 50 , 51 ], and two others used 300-m and 1,000-m radii [ 53 , 56 ]. Additionally, one study measured exposure at multiple scales (250 m, 500 m, and 1,000 m) [ 58 ]. For statistical synthesis, the effect estimate corresponding to the highest greening exposure level was extracted for meta-analysis.

As shown in Table 1 , this systematic review included only studies investigating the association between green space exposure and the risk of NDs. We excluded studies on all-cause dementia, vascular dementia, and those reporting solely mortality from neurological diseases [ 22 ]. Among the included studies, 8 examined the association between residential green space and AD risk [ 46 , 48 , 49 , 52 , 54 , 56 – 58 ], four focused on PD [ 49 , 53 , 55 , 57 ], and two assessed cognitive impairment [ 50 , 51 ]. Additionally, one study [ 49 ] investigated the relationship between residential green space and multiple sclerosis risk. Furthermore, one study [ 48 ] evaluated the association between residential green space and overall neurodegenerative disease risk.

Among the 12 observational studies included in this meta-analysis, a single study was rated as moderate quality, while the remaining 11 were rated as high quality, as documented in the Supplementary Material.

Residential green space and risk of neurodegenerative disease

To provide a general overview, an exploratory pooled analysis combining AD, PD, and cognitive impairment were conducted. Our pooled analysis of 12 studies demonstrated that high residential green space exposure was associated with a 15% lower risk of NDs versus low exposure (HR = 0.85, 95% CI: 0.79–0.91; Fig. 2 ). However, substantial heterogeneity was observed (I² = 91.08%, P < 0.001). To explore potential sources of heterogeneity, we conducted a series of sensitivity analyses. The protective association remained robust after excluding analyses to studies using land use/land cover-based exposure assessment (HR = 0.76; Supplementary Figure S1), and when limited to cohort studies (HR = 0.76; Supplementary Figure S3). Similarly, restricting the analysis to studies reporting hazard ratios did not materially change the results (HR = 0.78; Supplementary Figure S20). In the leave-one-out sensitivity analysis, sequential exclusion of individual studies—including the largest study—did not substantially alter the pooled estimates, with effect sizes remaining within a similar range (HR = 0.85, 95% CI: 0.79–0.92) and consistently indicating a protective association (Supplementary Figure S19). These findings suggest that no single study disproportionately influenced the overall results. In addition, we assessed the impact of studies contributing multiple disease endpoints from a single cohort. After excluding these studies, the pooled estimate remained largely unchanged (HR = 0.83, 95% CI: 0.77–0.91, Supplementary Figure S17), while heterogeneity remained high (I² = 90.35%). This indicates that inclusion of correlated effect sizes did not materially bias the overall findings.

The persistence of high heterogeneity across all analyses, despite the robustness of the primary finding, suggests it is not attributable to a single outlier study or a specific methodological choice (e.g., study design, exposure metric). Instead, it likely originates from underlying clinical and methodological diversity across studies, such as differences in population characteristics (age, genetics, SES), varied definitions of residential green space exposure, and discrepancies in covariate adjustment models.

Subgroup analyses by disease type

Given the substantial heterogeneity observed in the main analysis (I² = 91.08%), we conducted several pre-specified subgroup analyses to explore potential sources of this variation (Fig. 3 ). Subgroup analyses stratified by disease type indicated differential associations between residential green space exposure and the risks of specific NDs. A pooled analysis of seven studies demonstrated that higher levels of green space exposure were significantly associated with a 14% lower risk of AD (HR = 0.86, 95% CI: 0.79–0.95). In contrast, the pooled associations for PD (HR = 0.75, 95% CI: 0.56–1.00; based on four studies) and cognitive impairment (HR = 0.79, 95% CI: 0.61–1.02; based on two studies) did not reach statistical significance (see Supplementary Figures S4-6). The nonsignificant findings for PD and cognitive impairment may be partially explained by the limited number of available studies and smaller sample sizes, which might reduce statistical power and contribute to wider confidence intervals.

Other types of subgroup analyses and meta-regression analyses

Subgroup analyses were conducted by sex, age, smoking status, buffer zone radius, and NDVI increments (per 0.1 unit). Sex-stratified analyses revealed comparable protective associations for male (HR = 0.75, 95% CI: 0.70–0.81; Supplementary Figure S8) and female (HR = 0.74, 95% CI: 0.63–0.87; Supplementary Figure S9). Age-specific analyses indicated an inverse association between green space exposure and ND risk that attenuated with increasing age, being most pronounced in participants < 65 years (HR = 0.69, 95% CI: 0.58–0.82; Supplementary Figure S10). Smoking status modified the association, with stronger protective effects observed in nonsmokers (HR = 0.75, 95% CI: 0.65–0.85; Supplementary Figure S12) than smokers (HR = 0.87, 95% CI: 0.76–0.98; Supplementary Figure S13). Six studies that looked at NDVI as a continuous variable found that for every 0.1-unit increase, the risk went down by 12% (HR = 0.88, 95% CI: 0.83–0.94; Supplementary Figure S7). Unlike previous studies, we stratified green buffer zones by radius (250, 500, and 1,000 m). The strongest protective association was observed for the 250-m buffer zone (HR = 0.69, 95% CI: 0.61–0.78; Supplementary Figure S14).

We used multivariate meta-regression to find possible modifying factors because the observed heterogeneity was high (I² = 91.08%). Despite assessing multiple covariates, including ND subtype, demographic characteristics (age, sex, ethnicity), socioeconomic factors (education, income), lifestyle factors (smoking, alcohol use), comorbidities, and urban/rural residence, none significantly accounted for the observed heterogeneity in the green space-ND association (Supplementary Table S6). Publication bias was evaluated using funnel plots and Egger’s test (all p > 0.05), suggesting no substantial bias.

However, heterogeneity remained high within most subgroups (I² > 70%), indicating that these characteristics could only explain a small fraction of the inter-study variance. Therefore, the results of these exploratory analyses should be interpreted with considerable caution.

Dose–response analysis

Our dose-response analysis, incorporating five studies using NDVI, revealed a significant overall association (HR = 0.69, 95% CI 0.59–0.81, Supplemental Figure Dose-response 1). While a test for linearity was significant (χ² = 17.62, P < 0.001), the non-linear model provided a better fit to the data (χ² = 7.37, P = 0.03), suggesting a more complex relationship. As depicted in Fig. 4 B, the curve demonstrates a steep initial decline in the risk of NDs as NDVI increases from low to moderate levels. The risk was reduced by 22% at an NDVI of 0.4 (HR = 0.78, 95% CI 0.73–0.82) and by 46% at an NDVI of 0.6 (HR = 0.54, 95% CI 0.47–0.62). However, the curve then begins to plateau, and the 95% confidence intervals widen considerably at higher NDVI values, indicating greater uncertainty in the effect estimate. Notably, the protective effect was no longer statistically significant at the highest end of the exposure range (HR at NDVI = 1.0: 0.53; 95% CI: 0.27–1.03). This pattern suggests a potential saturation effect, where the marginal health benefits may diminish beyond a certain threshold of green space exposure. Therefore, while an inverse dose-response relationship is evident, it appears to be non-linear and is characterized by significant uncertainty at the highest exposure levels. (Linear dose–response: Fig. 4 A; nonlinear dose–response: Fig. 4 B).

GRADE evidence quality assessment

The certainty of the summary evidence was graded according to the GRADE (Grading of Recommendations Assessment, Development and Evaluation) framework [ 59 ]. We assessed five domains for each outcome: risk of bias, inconsistency, indirectness, imprecision, and publication bias. The overall certainty of the evidence for the association between residential green space and the risk of neurodegenerative diseases was judged to be Very Low according to the GRADE criteria (Supplementary Table S6). Although the evidence originated from observational studies (starting at “Low” certainty) and we found a dose-response gradient, the certainty was downgraded by two levels due to serious risk of bias across the studies and very serious inconsistency (I² > 90%). Therefore, although our meta-analysis shows a consistent protective trend, the generalizability of the conclusion should be interpreted with caution.

Discussion

This meta-analysis included 12 studies, which comprehensively assessed the association between residential greening and the risk of NDs among over 1.7 million participants. The results indicated that higher levels of residential green space were associated with a reduced risk of NDs in middle-aged and older adults. Moreover, this association remained consistent in subgroup analyses stratified by ND category, sex, age, smoking status, buffer size, and per 0.1-unit increase in greenness value.

Our dose-response analysis further suggested an inverse association between residential green space and NDs risk. The precision and generalizability of this finding are limited by the small number of eligible studies. Nevertheless, the result was robust in a sensitivity analysis, as the exclusion of any individual study did not materially affect the dose-response relationship’s shape or significance. Therefore, while our analysis points to a notable trend, these findings should be considered preliminary, requiring validation in larger studies that utilize standardized exposure assessments.

The substantial heterogeneity in our analysis persisted despite rigorous sensitivity analyses, suggesting it originates from the intrinsic diversity of the primary studies rather than statistical artifacts. One likely source is our aggregation of different NDs into a single outcome. While various NDs share age-related neurodegenerative processes and overlapping environmental or lifestyle risk factors, we must also acknowledge their distinct pathophysiological and diagnostic differences. Therefore, we emphasize that the pooled result for overall NDs should be interpreted as a broad, cross-diagnostic indicator rather than a conclusion specific to any single disease.

To further elucidate potential disease-specific differences, subgroup analyses were performed by type of neurodegenerative outcome. The results revealed heterogeneous associations across disease categories. Specifically, a pooled analysis of seven studies indicated that higher green space exposure was significantly associated with a 14% lower risk of AD (HR = 0.86, 95% CI: 0.79–0.95). This finding is consistent with growing epidemiological and experimental evidence suggesting that greener living environments may protect against cognitive decline and dementia. Moreover, as green space can mitigate loneliness—a key risk factor for cognitive decline—by facilitating social engagement and cognitive activation [ 60 ].

In contrast, no statistically significant associations were observed between residential green space and the risks of PD (HR = 0.75, 95% CI: 0.56–1.00) or cognitive impairment (HR = 0.79, 95% CI: 0.61–1.02). These nonsignificant findings should be interpreted with caution, as the number of studies and sample sizes for these outcomes were limited, potentially reducing statistical power and increasing uncertainty in the pooled estimates. Moreover, differences in exposure assessment methods (e.g., types of greenness indicators, buffer distances, and temporal exposure windows) and variations in demographic or lifestyle factors across study populations may also contribute to inconsistent results.

Notably, we observed that the protective effect of residential green space levels against NDs was more pronounced in participants younger than 65 years, whereas no significant association was found in those older than 75 years (HR = 0.97, 95% CI: 0.80–1.18). This age-dependent disparity may be attributed to two factors: (1) The accumulation of aging-related pathologies (e.g., increased DNA damage [ 8 ]), which may attenuate the benefits of environmental interventions; (2) reduced outdoor activity among older adults, leading to lower effective residential green space exposure. These findings underscore the importance of early ND prevention, ideally during the preclinical or early neuropathological stages, aligning with the emerging concept of “preclinical intervention” for AD [ 61 ]. Intriguingly, the protective effect of greenness was more prominent in nonsmokers, despite epidemiological evidence suggesting a negative association between smoking and PD risk [ 62 , 63 ]. While nicotine has been hypothesized to exert neuroprotective effects (e.g., via dopamine release stimulation), this does not preclude the harmful impact of other tobacco constituents. Due to the limited number of eligible studies, we were unable to conduct BMI-stratified analyses to explore its potential interaction with residential green space-ND associations. However, buffer-based subgroup analyses revealed that residential green space within a 250–m radius of residences demonstrated the strongest protective effect against ND. This distance likely reflects the typical activity range of older adults and aligns with evidence suggesting that synergistic interactions between trees and herbaceous vegetation at this scale optimize ecological benefits [ 64 ].

We hypothesize that vegetation-specific differences in photosynthetic efficiency (quantified via NDVI) may modulate neuroprotective effects through microenvironmental regulation (e.g., temperature, humidity, and biogenic volatile organic compounds) [ 65 ]. This mechanistic insight warrants further investigation.

While our review focuses on the epidemiological association, it is useful to briefly consider the potential pathways through which green space may confer neuroprotective benefits. The benefits of green space are likely mediated through a combination of pathways, including promoting physical activity [ 66 – 68 ] and social engagement [ 69 ], facilitating psychological restoration and stress reduction [ 70 ], and mitigating harmful environmental exposures like air pollution and noise. These broad pathways, rather than specific biomolecular interactions, provide a plausible framework for the observed associations.

Furthermore, a major limitation, inherent in the primary studies included in this review, is the potential for significant exposure misclassification, which likely biases our pooled estimates toward the null. The exposure assessment methods used in the included studies were predominantly based on static, residence-centered metrics such as NDVI or land-use categories. This approach fails to capture the true, multi-dimensional nature of an individual’s interaction with green space, leading to several forms of misclassification [ 71 , 72 ].

First, we must consider spatial misclassification. The reliance on residential addresses overlooks daily mobility and the time individuals, especially active older adults, spend in non-residential locations like parks, community centers, or workplaces. This discrepancy between residential and activity-space exposure, often termed the Uncertain Geographic Context Problem, can lead to a substantial under- or over-estimation of an individual’s true green space exposure [ 73 ]. Furthermore, broad metrics like NDVI do not differentiate between green space types (e.g., tree canopy vs. low-lying grass) or account for green space quality, accessibility, and amenities [ 31 ]. A high-quality, accessible park with benches and walking paths offers vastly different opportunities for physical activity, social engagement, and stress reduction compared to an inaccessible patch of dense vegetation with the same NDVI value.

Second, temporal misclassification is a critical concern, especially given the long latency periods of neurodegenerative diseases. Most included studies used cross-sectional green space data, which may not accurately reflect historical exposure during the etiologically relevant developmental window of the disease, potentially spanning decades. Changes in urban greening or residential history are rarely accounted for, introducing a significant temporal mismatch between exposure and outcome [ 74 ].

To overcome these limitations, promising alternatives offer greater spatiotemporal resolution and are gaining traction [ 75 – 77 ]. These include longitudinal NDVI averages, mobility-based metrics derived from high-resolution GPS data, and ground-level indices from street-view imagery. Future research should prioritize these dynamic, mobility-based exposure assessments and incorporate sensitivity analyses, such as using multiple buffers or measurement error correction methods, to validate the robustness of their results.

Comparison with previous studies

To the best of our knowledge, this meta-analysis is the first to specifically synthesize the evidence on the association between residential green space and the risk of NDs such as Alzheimer’s and Parkinson’s disease. Previous reviews have often subsumed these conditions under the broader category of dementia. For instance, a related meta-analysis by Zagnoli et al. [ 26 ] explored the association between greenness and dementia. While their main pooled analysis did not show a statistically significant overall association, their dose–response analysis suggested a non-linear relationship, with evidence of a potential protective effect at intermediate levels of greenness exposure. However, specific effect estimates varied across studies and exposure metrics, and were primarily derived from modeled dose–response curves rather than uniform exposure units, limiting direct comparability. Our study builds upon this by focusing on specific NDs and confirming a similar, albeit stronger, protective non-linear dose-response pattern, thereby filling a critical knowledge gap. Furthermore, a recent review by Da et al. [ 27 ] also assessed the impact of environmental factors on cognitive decline and dementia, reporting findings consistent with ours regarding the protective role of green space. Our review distinguishes itself through a more stringent methodology. We applied stricter inclusion and exclusion criteria to ensure the independence of each study cohort and, critically, we included only studies that utilized objective measures of green space exposure (e.g., NDVI, LU/LC). This focus on objective metrics was intended to reduce potential bias from exposure misclassification, thereby increasing the robustness of our quantitative synthesis. We also recommend that future studies adopt standardized methods for assessing green space exposure to improve comparability and facilitate a more comprehensive evaluation of the dose-response relationship.

Strengths and limitations

Key strengths of this study include: (1) A systematic review and meta–analysis of residential green space exposure and ND risk in middle–aged/older adults, incorporating the latest evidence and dose–response analysis; (2) Rigorous quality appraisal, risk–of–bias assessment, and sensitivity analyses to ensure result reliability and robustness; (3) Multivariable meta–regression confirming residential green space as an independent protective factor against ND in this population; (4) First dose–response quantification of ND risk gradients across residential green space levels, addressing prior limitations of dichotomous (“high vs. low”) exposure comparisons; and (5) ND–specific focus (vs. broader dementia studies) enhanced study comparability and enabled rigorous dose–response evaluation. These findings comprehensively elucidate the residential green space–ND relationship.

Our study is subject to several limitations: (1) Methodological Heterogeneity: The most significant limitation of this meta-analysis is the extremely high level of statistical heterogeneity observed (I 2 > 90%). It stated that this high heterogeneity severely limits confidence in the pooled effect estimate. The summary estimate should therefore be regarded not as a single true effect size, but rather as a mathematical average of findings from highly disparate studies. Consequently, the summary estimate should not be interpreted causally or as a precise quantification of the risk reduction. While our subgroup analyses attempted to explore the sources of this heterogeneity, they were largely unsuccessful in explaining the majority of the variance. This suggests that the variability likely stems from a complex interplay of methodological factors that differ across studies. (2) Limited Data Availability: A paucity of studies prevented us from conducting meta-analyses for specific NDs, including Amyotrophic Lateral Sclerosis(ALS) and Multiple System Atrophy (MSA). (3) Geographic Bias: The predominance of studies from China and the US limits the generalizability of our findings. Variations in urban planning, environmental profiles, and sociocultural factors across different regions could modulate the residential green space-ND association. Future work should focus on low- and middle-income countries and other underrepresented areas to build a more globally relevant evidence base. (4) Low Evidence Quality: The overall quality of evidence was rated as moderate for our primary outcome and low to very low for others, potentially compromising the robustness of our findings. (5) The potential for temporal ambiguity in exposure assessment: Several included studies, particularly case-control and retrospective designs, used environmental data from a single time point to represent long-term exposure. This approach, while pragmatic, may introduce non-differential exposure misclassification if green space levels changed over time, potentially biasing our results towards the null. Future prospective studies with longitudinal exposure data are needed to better clarify the temporal relationship. (6) Effect Measure Aggregation: Pooling ORs, HRs, and RRs may have introduced imprecision into our summary estimate of risk. (7) Language Restriction: By including only English-language publications, we may have introduced language bias and overlooked relevant studies from non-English-speaking regions. This could impact the global applicability of our conclusions. Future updates should incorporate non-English and grey literature to test the robustness of the findings. We also advocate for broader language criteria in future systematic reviews to improve global representativeness and mitigate bias.

Conclusions

This systematic review and meta-analysis provides suggestive evidence that higher exposure to residential green space is associated with a lower risk of NDs in middle-aged and older adults. Our dose-response analysis indicates a non-linear inverse association, though the effect estimates are characterized by considerable uncertainty at higher levels of residential green space. While the evidence is promising, particularly for exposures measured within a 250-meter buffer, the findings should be interpreted with caution due to the very low overall certainty of evidence and the significant heterogeneity across studies. Our results support the potential of residential green space as a public health strategy to promote healthy aging and mitigate the risk of NDs. However, these findings underscore the need for future, large-scale longitudinal studies with standardized exposure assessments to confirm these associations and to more precisely delineate the dose-response relationship. Such research would be vital for informing evidence-based urban planning and public health interventions.

Supplementary Information

Supplementary Material 1.

Supplementary Material 2.

Acknowledgements

We thank all the authors who contributed to this study.

Abbreviations

Neurodegenerative diseases

Alzheimer’s disease

Parkinson’s disease

Amyotrophic Lateral Sclerosis

Multiple System Atrophy

Normalized Difference Vegetation Index

land use/land cover

Confidence interval

Hazard ratio

Odds ratio

Relative risk

Grading of Recommendations, Assessment, Development, and Evaluation

Newcastle-Ottawa Scale

Authors’ contributions

S.C.Y. and S.J. were responsible for the study design and conception, wrote the first draft, and revised and integrated the contributions of the remaining authors. S.C.Y., L.J., and J.Y.J. conceptualized the study and were responsible for direct data acquisition and analysis. S.C.Y., Z.X.W. and J.Y.J. were responsible for calculating and analyzing the dose–response relationship. L.L.M., W.Z.H. and L.J. were responsible for the design, conception, and drafting of the entire study. Z.D.X. and S.J. were responsible for the article quality review and finalization. The corresponding author attests that all listed authors meet the authorship criteria and that no others meeting the criteria have been omitted. All authors reviewed and approved the final manuscript.

Funding

The author did not receive any organizational support for the submission.

Data availability

All data generated or analyzed during this study are included in this published article and its Supplementary Information file. However, raw data (Excel spreadsheets) used in the statistical analyses will be made available upon request through the corresponding author (Zhao Dexi).

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Chunyu Shen, Jing Su and Liming Liu are the first authors.

Contributor Information

Yanjie Jiang, Email: jiangyanjie9415@163.com.

Dexi Zhao, Email: dexizhao1006@163.com.

References

1. Ou G, Lin W, Zhao W. Neuregulins in Neurodegenerative Diseases. Front Aging Neurosci. 2021;13:662474. 10.3389/fnagi.2021.662474.

2. Kovacs GG. Chap. 21 - Concepts and classification of neurodegenerative diseases. In: Kovacs GG, Alafuzoff I, editors. Handbook of Clinical Neurology. Elsevier; 2018. pp. 301–7. 10.1016/B978-0-12-802395-2.00021-3.

3. Brettschneider J, Del Tredici K, Lee VM-Y, Trojanowski JQ. Spreading of pathology in neurodegenerative diseases: a focus on human studies. Nat Rev Neurosci. 2015;16:109–20. 10.1038/nrn3887.

4. Kritsilis M, Rizou SV, Koutsoudaki PN, et al. Ageing, Cellular Senescence and Neurodegenerative Disease. Int J Mol Sci. 2018;19:2937. 10.3390/ijms19102937.

5. Foreman KJ, Marquez N, Dolgert A, et al. Forecasting life expectancy, years of life lost, and all-cause and cause-specific mortality for 250 causes of death: reference and alternative scenarios for 2016-40 for 195 countries and territories. Lancet (London England). 2018;392. 10.1016/S0140-6736(18)31694-5.

6. Wang S, JiangY, Yang A, et al. The Expanding Burden of Neurodegenerative Diseases: An Unmet Medical and Social Need. Aging disease. 2024;16(5):2937–52. 10.14336/AD.2024.1071.

7. Rocca WA, Petersen RC, Knopman DS, et al. Trends in the incidence and prevalence of Alzheimer’s disease, dementia, and cognitive impairment in the United States. Alzheimers Dement. 2011;7:80–93. 10.1016/j.jalz.2010.11.002.

8. Hou Y, Dan X, Babbar M, et al. Ageing as a risk factor for neurodegenerative disease. Nat Rev Neurol. 2019;15:565–81. 10.1038/s41582-019-0244-7.

9. Collaborators G. 2019 DF. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. The Lancet Public Health. 2022;7:e105. 10.1016/S2468-2667(21)00249-8.

10. Nandi A, Counts N, Chen S, et al. Global and regional projections of the economic burden of Alzheimer’s disease and related dementias from 2019 to 2050: A value of statistical life approach. EClinicalMedicine. 2022;51:101580. 10.1016/j.eclinm.2022.101580.

11. Wu Y-T, Beiser AS, Breteler MMB, et al. The changing prevalence and incidence of dementia over time - current evidence. Nat Rev Neurol. 2017;13:327–39. 10.1038/nrneurol.2017.63.

12. Aperghis M, Johnson IP, Patel N, et al. Age, diet and injury affect the survival of facial motoneurons. Neuroscience. 2003;117:97–104. 10.1016/S0306-4522(02)00762-5.

13. Alty J, Farrow M, Lawler K. Exercise and dementia prevention. Pract Neurol. 2020;20(3):234–40. 10.1136/practneurol-2019-002335.

14. Wahl D, Solon-Biet SM, Cogger VC, et al. Aging, lifestyle and dementia. Neurobiol Dis. 2019;130:104481. 10.1016/j.nbd.2019.104481.

15. Costa LG, Cole TB, Dao K, et al. Effects of air pollution on the nervous system and its possible role in neurodevelopmental and neurodegenerative disorders. Pharmacol Ther. 2020;210. 10.1016/j.pharmthera.2020.107523.

16. Lavezzi AM, Colizzi M, Lein PJ. Editorial: Advances in understanding the pathogenetic mechanisms of neurodevelopmental disorders and neurodegenerative disease — The environment as a putative risk factor. Front Neurol. 2023;14:1259772. 10.3389/fneur.2023.1259772.

17. Belloy ME, Napolioni V, Greicius MD. A Quarter Century of APOE and Alzheimer’s Disease: Progress to Date and the Path Forward. Neuron. 2019;101:820–38. 10.1016/j.neuron.2019.01.056.

18. Kuntić M, Hahad O, Münzel T, Daiber A. Crosstalk between Oxidative Stress and Inflammation Caused by Noise and Air Pollution—Implications for Neurodegenerative Diseases. Antioxidants. 2024;13:266. 10.3390/antiox13030266.

19. Al-Chalabi A. Preventing neurodegenerative disease. Brain. 2021;144:1279–80. 10.1093/brain/awab151.

20. Renalds A, Smith TH, Hale PJ. A Systematic Review of Built Environment and Health. Fam Community Health. 2010;33:68–78. 10.1097/FCH.0b013e3181c4e2e5.

21. Lambert KG, Nelson RJ, Jovanovic T, Cerdá M. Brains in the City: Neurobiological effects of urbanization. Neurosci Biobehav Rev. 2015;58:107–22. 10.1016/j.neubiorev.2015.04.007.

22. Rodriguez-Loureiro L, Gadeyne S, Bauwelinck M, et al. Long-term exposure to residential greenness and neurodegenerative disease mortality among older adults: a 13-year follow-up cohort study. Environ Health: Glob Access Sci Source. 2022;21:49. 10.1186/s12940-022-00863-x.

23. Klompmaker JO, Laden F, Browning MHEM, et al. Associations of greenness, parks, and blue space with neurodegenerative disease hospitalizations among older US adults. JAMA Netw Open. 2022;5:e2247664. 10.1001/jamanetworkopen.2022.47664.

24. Cherrie MPC, Shortt NK, Mitchell RJ, et al. Green space and cognitive ageing: a retrospective life course analysis in the lothian birth cohort 1936. Soc Sci Med. 2018;196:56–65. 10.1016/j.socscimed.2017.10.038.

25. Wu Y-T, Prina AM, Jones AP, et al. Community environment, cognitive impairment and dementia in later life: results from the cognitive function and ageing study. Age Ageing. 2015;44:1005–11. 10.1093/ageing/afv137.

26. Zagnoli F, Filippini T, Jimenez MP, et al. Is Greenness Associated with Dementia? A Systematic Review and Dose-Response Meta-analysis. Curr Environ Health Rep. 2022;9:574–90. 10.1007/s40572-022-00365-5.

27. Da L, Song X, Jia Z, et al. Objectively measured environmental features and their association with cognition and dementia: A systematic review and meta-analysis. Ageing Res Rev. 2025;104:102630. 10.1016/j.arr.2024.102630.

28. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA Statement for Reporting Systematic Reviews and Meta-Analyses of Studies That Evaluate Health Care Interventions: Explanation and Elaboration. PLoS Med. 2009;6:e1000100. 10.1371/journal.pmed.1000100.

29. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. 10.1136/bmj.n71.

30. Morgan RL, Whaley P, Thayer KA, Schünemann HJ. Identifying the PECO: A framework for formulating good questions to explore the association of environmental and other exposures with health outcomes. Environ Int. 2018;121:1027–31. 10.1016/j.envint.2018.07.015.

31. Markevych I, Schoierer J, Hartig T, et al. Exploring pathways linking greenspace to health: Theoretical and methodological guidance. Environ Res. 2017;158:301–17. 10.1016/j.envres.2017.06.028.

32. Pettorelli N, Vik JO, Mysterud A, et al. Using the satellite-derived NDVI to assess ecological responses to environmental change. Trends Ecol Evol. 2005;20:503–10. 10.1016/j.tree.2005.05.011.

33. Vilcins D, Sly PD, Scarth P, Mavoa S. Green space in health research: an overview of common indicators of greenness. Rev Environ Health. 2024;39:221–31. 10.1515/reveh-2022-0083.

34. Hart JL, Yeager RA, Riggs DW, et al. The Relationship between Perceptions and Objective Measures of Greenness. Int J Environ Res Public Health. 2022;19:16317. 10.3390/ijerph192316317.

35. James P, Banay RF, Hart JE, Laden F. A review of the health benefits of greenness. Curr Epidemiol Rep. 2015;2:131–42. 10.1007/s40471-015-0043-7.

36. Dadvand P, Villanueva CM, Font-Ribera L, et al. Risks and benefits of green spaces for children: a cross-sectional study of associations with sedentary behavior, obesity, asthma, and allergy. Environ Health Perspect. 2014;122:1329–35. 10.1289/ehp.1308038.

37. Greenland S, Thomas DC, Morgenstern H. The rare-disease assumption revisited. A critique of estimators of relative risk for case-control studies. Am J Epidemiol. 1986;124:869–83. 10.1093/oxfordjournals.aje.a114476.

38. Jiang Y, Zhang S, Chen Y, et al. Physical activity and risk of Parkinson’s disease: an updated systematic review and meta-analysis. J Neurol. 2024;271:7434–59. 10.1007/s00415-024-12672-y.

39. Jiang Y, Jin Z, Wang H, et al. A dose–response meta-analysis of physical activity and the risk of alzheimer’s disease in prospective studies. J Neurol. 2025;272:256. 10.1007/s00415-025-12960-1.

40. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7:177–88. 10.1016/0197-2456(86)90046-2.

41. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997;315:629–34. 10.1136/bmj.315.7109.629.

42. Greenland S, Longnecker MP. Methods for Trend Estimation from Summarized Dose-Response Data, with Applications to Meta-Analysis. Am J Epidemiol. 1992;135:1301–9. 10.1093/oxfordjournals.aje.a116237.

43. Crippa A, Discacciati A, Bottai M, et al. One-stage dose–response meta-analysis for aggregated data. Stat Methods Med Res. 2019;28:1579–96. 10.1177/0962280218773122.

44. Wells GA, Shea B, O’Connell D et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. 2014. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp .

45. Herzog R, Álvarez-Pasquin MJ, Díaz C, et al. Are healthcare workers’ intentions to vaccinate related to their knowledge, beliefs and attitudes? A systematic review. BMC Public Health. 2013;13:154. 10.1186/1471-2458-13-154.

46. Wu J, Jackson L. Greenspace Inversely Associated with the Risk of Alzheimer’s Disease in the Mid-Atlantic United States. Earth (Basel). 2021;2:140–50. 10.3390/earth2010009.

47. Higgins JPT, Morgan RL, Rooney AA, et al. A tool to assess risk of bias in non-randomized follow-up studies of exposure effects (ROBINS-E). Environ Int. 2024;186:108602. 10.1016/j.envint.2024.108602.

48. Liu C-C, Li C-Y, Kung S-F, et al. Association of Environmental Features and the Risk of Alzheimer’s Dementia in Older Adults: A Nationwide Longitudinal Case-Control Study. Int J Environ Res Public Health. 2019;16:2828. 10.3390/ijerph16162828.

49. Yuchi W, Sbihi H, Davies H, et al. Road proximity, air pollution, noise, green space and neurologic disease incidence: a population-based cohort study. Environ Health. 2020;19:8. 10.1186/s12940-020-0565-4.

50. Zhu A, Yan L, Shu C, et al. APOE ε4 Modifies Effect of Residential Greenness on Cognitive Function among Older Adults: A Longitudinal Analysis in China. Sci Rep. 2020;10:82. 10.1038/s41598-019-57082-7.

51. Jin X, Shu C, Zeng Y, et al. Interaction of greenness and polygenic risk score of Alzheimer’s disease on risk of cognitive impairment. Sci Total Environ. 2021;796:148767. 10.1016/j.scitotenv.2021.148767.

52. Aitken WW, Lombard J, Wang K, et al. Relationship of Neighborhood Greenness to Alzheimer’s Disease and Non-Alzheimer’s Dementia Among 249,405 U.S. Medicare Beneficiaries. JAD. 2021;81:597–606. 10.3233/JAD-201179.

53. Yu Z, Wei F, Zhang X, et al. Air pollution, surrounding green, road proximity and Parkinson’s disease: A prospective cohort study. Environ Res. 2021;197:111170. 10.1016/j.envres.2021.111170.

54. Slawsky ED, Hajat A, Rhew IC, et al. Neighborhood greenspace exposure as a protective factor in dementia risk among U.S. adults 75 years or older: a cohort study. Environ Health. 2022;21:14. 10.1186/s12940-022-00830-6.

55. Jung J, Park JY, Myung W, et al. Association between Residential Greenness and Incidence of Parkinson’s Disease: A Population-Based Cohort Study in South Korea. IJERPH. 2022;19:3491. 10.3390/ijerph19063491.

56. Hu H-Y, Ma Y-H, Deng Y-T, et al. Residential greenness and risk of incident dementia: A prospective study of 375,342 participants. Environ Res. 2023;216:114703. 10.1016/j.envres.2022.114703.

57. Zhu Z, Yang Z, Yu L, et al. Residential greenness, air pollution and incident neurodegenerative disease: A cohort study in China. Sci Total Environ. 2023;878:163173. 10.1016/j.scitotenv.2023.163173.

58. Brown SC, Aitken WW, Lombard J, et al. Longitudinal Impacts of Precision Greenness on Alzheimer’s Disease. J Prev Alzheimer’s Disease. 2024;11:710–20. 10.14283/jpad.2024.38.

59. Langendam MW, Akl EA, Dahm P, et al. Assessing and presenting summaries of evidence in Cochrane Reviews. Syst Rev. 2013;2:81. 10.1186/2046-4053-2-81.

60. Lara E, Caballero FF, Rico-Uribe LA, et al. Are loneliness and social isolation associated with cognitive decline? Int J Geriatr Psychiatry. 2019;34:1613–22. 10.1002/gps.5174.

61. Sperling RA, Aisen PS, Beckett LA, et al. Toward defining the preclinical stages of Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 2011;7:280–92. 10.1016/j.jalz.2011.03.003.

62. Mappin-Kasirer B, Pan H, Lewington S, et al. Tobacco smoking and the risk of Parkinson disease: A 65-year follow-up of 30,000 male British doctors. Neurology. 2020;94:e2132–8. 10.1212/WNL.0000000000009437.

63. Cheng Y, Wang Y-J. Tobacco smoking and the reduced risk of Parkinson disease: A puzzle of 60 years. Neurology. 2020;94:860–1. 10.1212/WNL.0000000000009431.

64. Martinez ADLI, Labib SM. Demystifying normalized difference vegetation index (NDVI) for greenness exposure assessments and policy interventions in urban greening. Environ Res. 2023;220:115155. 10.1016/j.envres.2022.115155.

65. Cristiano P, Madanes N, Campanello P, et al. High NDVI and Potential Canopy Photosynthesis of South American Subtropical Forests despite Seasonal Changes in Leaf Area Index and Air Temperature. Forests. 2014;5:287–308. 10.3390/f5020287.

66. Gong Y, Gallacher J, Palmer S, Fone D. Neighbourhood green space, physical function and participation in physical activities among elderly men: the Caerphilly Prospective study. Int J Behav Nutr Phys Act. 2014;11:40. 10.1186/1479-5868-11-40.

67. Iso-Markku P, Kujala UM, Knittle K, et al. Physical activity as a protective factor for dementia and Alzheimer’s disease: systematic review, meta-analysis and quality assessment of cohort and case-control studies. Br J Sports Med. 2022;56. 10.1136/bjsports-2021-104981.

68. Mahalakshmi B, Maurya N, Lee S-D, Bharath Kumar V. Possible Neuroprotective Mechanisms of Physical Exercise in Neurodegeneration. IJMS. 2020;21:5895. 10.3390/ijms21165895.

69. De Vries S, Van Dillen SME, Groenewegen PP, Spreeuwenberg P. Streetscape greenery and health: Stress, social cohesion and physical activity as mediators. Soc Sci Med. 2013;94:26–33. 10.1016/j.socscimed.2013.06.030.

70. Banay RF, James P, Hart JE, et al. Greenness and Depression Incidence among Older Women. Environ Health Perspect. 2019;127:027001. 10.1289/EHP1229.

71. Helbich M. Toward dynamic urban environmental exposure assessments in mental health research. Environ Res. 2018;161:129–35. 10.1016/j.envres.2017.11.006.

72. Kwan M-P. The Uncertain Geographic Context Problem. Ann Assoc Am Geogr. 2012;102:958–68. 10.1080/00045608.2012.687349.

73. Kwan MP. The limits of the neighborhood effect: contextual uncertainties in geographic, environmental health, and social science research. Ann Am Assoc Geogr. 2018;108:1482–90. 10.1080/24694452.2018.1453770.

74. Nieuwenhuijsen MJ, Khreis H. Car free cities: pathway to healthy urban living. Environ Int. 2016;94:251–62. 10.1016/j.envint.2016.05.032.

75. Freymueller J, Schmid HL, Senkler B, et al. Current methodologies of greenspace exposure and mental health research-a scoping review. Front public health. 2024;12. 10.3389/fpubh.2024.1360134.

76. James P, Hart JE, Hipp JA, et al. GPS-Based Exposure to Greenness and Walkability and Accelerometry-Based Physical Activity. Cancer Epidemiol Biomarkers Prev. 2017;26:525–32. 10.1158/1055-9965.EPI-16-0925.

77. Wilt GE, Roscoe CJ, Hu CR, et al. Minute level smartphone derived exposure to greenness and consumer wearable derived physical activity in a cohort of US women. Environ Res. 2023;237:116864. 10.1016/j.envres.2023.116864.

Associated Data

Supplementary Materials

Supplementary Material 1.

Supplementary Material 2.

Data Availability Statement

All data generated or analyzed during this study are included in this published article and its Supplementary Information file. However, raw data (Excel spreadsheets) used in the statistical analyses will be made available upon request through the corresponding author (Zhao Dexi).