Investigating biological sex as a moderator of the association of nature exposure with brain health: a cross-sectional UK biobank analysis

The study investigates whether biological sex moderates the relationship between residential nature exposure and brain volume or cognitive function. Dementia poses a major global health burden with limited pharmaceutical treatments, highlighting the need for accessible prevention strategies. Prior observational evidence links residential greenspace to favorable brain structure and cognitive outcomes, but most studies have used greenspace rather than broader natural environment measures and have rarely examined moderators such as sex. Given known sex differences in dementia risk and brain aging, and sex-specific pathways (e.g., stress, mental health, physical activity, air pollution) potentially linking nature exposure to brain and cognitive health, the authors hypothesized that sex could moderate associations of nature exposure with brain volumes and cognitive performance. Using UK Biobank data, they aimed to assess these associations and formally test moderation by biological sex with comprehensive exposure measures and a large sample.

Prior studies suggest residential greenspace is associated with greater regional grey matter volume, cortical thickness, structural integrity, and better cognitive performance across domains. However, gaps include a focus on greenspace (vegetation) rather than broader nature (including water, rock), which may underrepresent exposure, small samples, and limited tests of moderation by sex. Evidence on sex differences is mixed: some studies report no sex differences for brain structure, while others find slower cognitive decline in females. Methodological differences across exposure measures, buffer sizes (100–2000 m), cognitive domains, and study designs (cross-sectional vs longitudinal) complicate interpretation. Theoretical frameworks propose pathways including stress reduction and attention restoration, mitigation of harmful stimuli (air pollution, noise), and facilitation of health behaviors (physical activity, social engagement, sleep, immune function), with potential sex-specific effects.

Design: Cross-sectional analysis using UK Biobank data (Application 69022). Participants: UK residents aged 37–73 recruited 2006–2010; analytic sample included 11,448 individuals with complete data for nature exposure (baseline 2006–2010), structural MRI (2014–2020), neuropsychological testing (2014–2015), and covariates; excluded those with dementia, Parkinson’s disease, or stroke. Ethics: Approved by NHS National Research Ethics Service (11/NW/0382); informed consent obtained. Nature exposure: Residential exposure estimated as percentage of land cover classified as “natural environment” within 1000 m and 300 m buffers around home locations using CEH 2007 Land Cover Map. Primary measure used land cover-based natural environment (vegetation, water, rock) rather than land-use greenspace. Buffer sizes chosen to capture broader neighborhood (1000 m) and immediate area (300 m). MRI outcomes: Structural MRI on Siemens Skyra 3T; T1-weighted images processed via UK Biobank pipeline. Outcomes: total grey matter volume, total white matter volume, and average hippocampal volume (across hemispheres). Volumes normalized for head size: total grey/white using T1-based skull surface estimate; hippocampal volume normalized by multiplying raw volume by head size scaling factor (field 25000). Cognitive outcomes: Online battery included Trail Making Test A and B; set-shifting indexed by TMT B-A (smaller difference indicates better performance). Symbol Digit Substitution Test measured complex processing speed (number of correct matches in 1 minute). Moderator: Biological sex recorded as male or female. Covariates: Age, BMI, education, household income, physical activity (IPAQ short form), time spent outdoors, smoking status, disability status, neighborhood socioeconomic status (Index of Multiple Deprivation), and home area population density; ethnic background included in crude models. Statistical analysis: Conducted in R 4.2.3. Descriptive statistics with t-tests and chi-squared tests. Linear regression models assessed associations of nature exposure (per 10% increment) with each brain and cognitive outcome, adjusted for covariates; crude models adjusted for age, sex, ethnic background. Moderation by sex tested via interaction term (nature exposure*sex); significant interactions followed by sex-stratified regressions. Sensitivity analyses: (1) alternative exposure measure using proportion of land classified as “greenspace” from GLUD (limited to England), for 1000 m and 300 m buffers; (2) restricting sample to participants without changes in residence between baseline and follow-up (excluding those whose time at current address was less than the interval between assessments). Alpha < 0.05; exploratory analyses.

Sample: N = 11,448; mean age 55.0 (SD 7.5); 51% female; 98% White. Sex differences in sample characteristics included males being older, more time outdoors, higher prevalence of hearing difficulty, diabetes, cardiovascular problems; females had lower residential nature and greenspace percentages, better TMT B-A and Symbol Digit Substitution performance, greater total grey matter and hippocampal volume, and lower total white matter volume. Associations (per 10% increment in natural environment exposure): - Total grey matter volume: 1000 m β = 629 mm³ (95% CI: 234, 1023; p = 0.002); 300 m β = 642 mm³ (95% CI: 286, 997; p < 0.001). - Total white matter volume: 1000 m β = 659 mm³ (95% CI: 229, 1089; p = 0.003); 300 m β = 527 mm³ (95% CI: 140, 914; p = 0.008). - Average hippocampal volume: 1000 m β = 0.39 mm³ (95% CI: -5.47, 6.25; p = 0.90); 300 m β = -3.76 mm³ (95% CI: -9.04, 0.00; p = 0.16) — not significant. - TMT B-A (sec): 1000 m β = -0.197 (95% CI: -0.395, 0.001; p = 0.05); 300 m β = -0.120 (95% CI: -0.299, 0.058; p = 0.19) — largely null. - Symbol Digit Substitution Test (correct matches): 1000 m β = 0.106 (95% CI: 0.057, 0.154; p < 0.001); 300 m β = 0.049 (95% CI: 0.006, 0.092; p = 0.03). Moderation by biological sex: - Grey matter volume: males showed greater increments than females per 10% exposure increase — 1000 m interaction β = 635 mm³ (95% CI: 53, 1217; p = 0.03); 300 m β = 634 mm³ (95% CI: 102, 1167; p = 0.02). Post-hoc stratified: significant in males (1000 m β = 797 mm³; 95% CI: 234, 1359; p = 0.006; 300 m β = 882 mm³; 95% CI: 374, 1391; p < 0.001), not in females (1000 m β = 444 mm³; 95% CI: -108, 996; p = 0.11; 300 m β = 447 mm³; 95% CI: -47, 941; p = 0.08). - TMT B-A: at 300 m only, males had greater improvement than females: interaction β = 0.284 s (95% CI: 0.016, 0.551; p = 0.04). Stratified: significant in males (β = -0.318 s; 95% CI: -0.564, -0.073; p = 0.01); females β = 0.064 s (95% CI: -0.193, 0.322; p = 0.62). - No sex moderation for white matter, hippocampal volume, or Symbol Digit Substitution outcomes. Sensitivity analyses: - Using greenspace (land use) instead of natural environment: results largely consistent; two changes lost significance — SDST association at 300 m (β = 0.031; 95% CI: -0.007, 0.069; p = 0.11) and sex moderation of TMT B-A at 300 m (β = 0.159 s; 95% CI: -0.086, 0.403; p = 0.20). - Restricting to non-movers: results largely unchanged; sex moderation for grey matter at 1000 m remained directional but was borderline/non-significant (β = 641 mm³; 95% CI: -11, 1292; p = 0.05); 300 m moderation remained robust.

The findings demonstrate that greater residential nature exposure is associated with larger total grey and white matter volumes and better processing speed, addressing the research question by showing beneficial associations between nature exposure and brain structure, with limited evidence for set-shifting. Biological sex moderates these relationships: males exhibited stronger associations for grey matter volume across buffer sizes and improved set-shifting at 300 m, suggesting sex-specific sensitivity to nature-related brain benefits. Potential mechanisms include sex differences in stress responses, mental health, air pollution effects, physical activity responses, and neuroendocrine changes (e.g., menopausal transition). The 300 m buffer appeared more sensitive to sex differences, indicating that immediate residential environments may be particularly relevant. However, the cognitive effect sizes observed (e.g., Symbol Digit Substitution) may be below clinically meaningful thresholds. Results’ sensitivity to exposure measures and residential stability underscores the importance of comprehensive exposure assessment and longitudinal address tracking. Overall, the study contributes evidence that nature exposure relates to brain health and that some benefits may vary by sex, informing hypotheses for future mechanistic and causal research.

Greater residential nature exposure may be linked to better brain structure and aspects of cognitive function in adults, with males potentially experiencing stronger benefits than females. These cross-sectional results add to emerging evidence on nature spaces as a community-level strategy for supporting brain health and potentially mitigating dementia risk. Future research should include longitudinal and experimental designs to establish causality, evaluate clinically meaningful cognitive impacts, test additional moderators (e.g., age, menopausal status, ethnicity), and investigate mediators such as stress, mental health, physical activity, and air pollution across diverse nature exposure measures and buffer sizes.

Cross-sectional design limits causal inference and allows potential reverse causality (e.g., healthier individuals choosing nature-rich areas). The binary, land cover-based exposure measure does not capture nature quality, actual engagement, or exposure beyond residence. The sample is predominantly cognitively healthy, White, and UK-based, limiting generalizability to other populations and regions. Analyses were exploratory without correction for multiple comparisons, increasing the risk of type I error. APOE ε4 genotype, a key dementia risk factor, was not available for inclusion as a covariate.