School of Public Health

Erik Slawsky

An Evaluation of Greenspace Exposure as a Protective Factor in Dementia Risk among U.S. Adults 75 Years or Older


Green spaces such as parks and forests have been identified to potentially confer health benefits, such as protective neurocognitive effects, to those individuals regularly exposed to them. Possible mechanisms for greenspace exposure to act on cognition range from facilitation of cognitive restoration to supporting lower levels of air pollution. Some literature to date has demonstrated neurocognitive health benefits for greenspace exposure among children. There is limited information on the role of residential greenspace exposure among the elderly in risk of dementia.


The Gingko Evaluation of Memory Study (GEMS) is a well-characterized cohort of U.S. adults aged 75 years or older evaluated for dementia. After geocoding participant residences, three greenspace metrics were combined to create a composite residential greenspace exposure. Normalized difference vegetative index (NDVI) values were combined with polygon and buffer based assessments of percent park overlap and linear distance to nearest park. Dementia and its subtypes Alzheimer’s disease (AD) and Vascular Dementia (VaD) were evaluated prospectively over 8 years using neuropsychiatric evaluations, MRI, and consensus adjudication by study neurologists. Risk of all-cause dementia, AD (without VaD) and VaD (with or without AD) was determined using Cox proportional hazards regression, unadjusted and adjusted for potential confounders including demographics, health behaviors, and ApoE genotype.


Of 3069 GEMS participants enrolled in 2000, 3048 (99%) were successfully geocoded. Analyses using Cox proportional hazards modeling showed that high residential greenspace (~25% of the sample) was associated with a reduced risk of dementia (HR=0.77, 95% CI: 0.59–0.99) in unadjusted models. With the model adjusted for demographic and behavioral characteristics, treatment arm, average neighborhood deprivation index, and Apolipoprotien E4 status, the association was attenuated (HR=0.82, 95% CI: 0.59–1.13), and did not remain statistically significant. When evaluating subtype, associations between greenspace and AD were not significant for the unadjusted model (HR=0.91, 95% CI: 0.65–1.27). Greespace was found to be significantly protective for risk of Mixed/VaD dementia in the unadjusted model when comparing high vs. low exposure. (HR=0.61, 95% CI: 0.41–0.93). Models for both subtypes were attenuated by adjustment. We did not find evidence of effect modification by sex, apolipoprotien E4 status, or neighborhood deprivation index.


We observed a modest relationship between residential greenspace exposure and all-cause dementia among the elderly although adjustment for confounders attenuated the relationship. Models evaluating subtype suggested a vascular pathway. Regular residential greenspace exposure may play a role in reducing the risk of dementia among the elderly in the United States.