Streetview greenspace types and dementia risk in older adults: the Multi-Ethnic Study of Atherosclerosis.

Greenspace has been associated with lower dementia risk, but most studies have used satellite-derived measures that cannot distinguish vegetation types. We used streetview (SV)-based metrics of greenspace from an on-the-ground perspective to examine greenspace types that may be relevant for dementia risk. We followed 6761 participants from the Multi-Ethnic Study of Atherosclerosis, 2000-2018. Using deep learning algorithms applied to SV images, we measured %trees, %grass, %other green (plants/flowers/fields) in the 500 m area around participants' residences. We examined the controlled direct effect of greenspace with dementia accounting for death as a censoring event, using a weighted Kaplan-Meier survival estimator, with the product of exposure weights and inverse-probability-of-censoring by death. We ran models separately per greenspace type. Mean baseline age was 62 years (SD, 10). Participants exposed to lowest levels of %trees were more likely to have a household income < $24 999. We observed 585 (8.7%) dementia cases and 1437 (21.3%) deaths over follow-up. We observed a direct effect of high exposure to %grass on 19-year dementia risk had death been prevented of -4.57 (95% CI, -8.15, -0.40) percentage points. No associations were observed for %trees (-0.07; 95% CI, -2.29 to 2.11) and %other green (-0.03, 95% CI, -1.59 to 1.48). Higher percentage of grass may reduce dementia risk.