Hi, my name is Richard Roven. My co-authors and I thank you for viewing our presentation entitled, The Association Between Residential Greenspace and Stroke. These are our disclosures. An increasingly recognized social determinant of mental and physical health is the use of social determinants. These are our methods. This is a one-to-four matched case control study of adults 18 years or older admitted to Aurora St. Luke's Medical Center in Milwaukee between 2016 and 2018. Cases were identified through our Get With The Guidelines database and included patients with ischemic stroke, TIA, subarachnoid hemorrhage, and intracerebral hemorrhage. Controls were identified through the electronic medical record. Matching variables included gender, five-year age group, admission month and year, race, hypertension, diabetes, smoking, and BMI. Potential subjects were excluded if they resided outside of Milwaukee County, had a prior stroke or TIA, or if there was a water feature within the residential zone. Ultimately, data from 5,870 patients were analyzed. This included 1,174 stroke cases and 4,696 non-stroke controls. We measured green space using the Normalized Difference in Vegetation Index, or NDVI. We averaged NDVI within a 250-meter radius of the patient's residence. NDVI was analyzed as both a continuous variable and as a categorical quartile. We also calculated walking distance from a patient's residence to the nearest public park. NDVI was treated as both a continuous variable and a binary variable, more or less than one kilometer distance. Our third metric is the Area Deprivation Index, a measure of socioeconomic status. This was analyzed as both an ordinal and a quartile variable. NDVI is a ratio that quantifies the difference between near-infrared light, which is strongly reflected by green vegetation, and red light, which is strongly absorbed by green vegetation. The near-infrared and red light reflectivity are obtained from satellite imaging and further processed using the ArcGIS geographical information software. The NDVI values range from negative one to positive one. A greater NDVI value means a greater density of green space. We determined the walking distance from a patient's residence to a public park, again using the ArcGIS software. The left panel shows a representative satellite image of Milwaukee County. The middle panel shows the NDVI transformed in ArcGIS. The right panel shows the area within a 250-meter radius of the patient's residence. We measured socioeconomic status using the Area Deprivation Index, or ADI. The ADI consists of 17 metrics obtained from the American Community Survey data. The ADI is accurate to the census block level. Every street address is associated with a national ADI percentile and a state ADI decile. National percentiles range from one or least disadvantaged to 100 or most disadvantaged. State deciles are similarly ordered from one to ten. These are our results. We first looked at the geographical distribution of stroke and non-stroke patients in Milwaukee County. Panel A is the deprivation map for the county, with red being the most deprived and yellow the least. Panel B shows the stroke cases. Using the Average Nearest Neighbor tool within ArcGIS software, we can see that stroke cases are clustered and not randomly distributed. Similarly, the non-stroke cases in Panel C are clustered. Panel D shows the overlap of cases and controls. NDVI is normally distributed for stroke cases and non-stroke controls. Stroke patients live in residences with greater surrounding green space than stroke patients. Using conditional logistic regression, this inverse relationship is statistically significant, with the odds of stroke decreasing as NDVI increases. When grouped into quartiles, patients living in the quartile with the greatest green space had a 19% lower odds of stroke compared to those living in the least green quartile. State ADI deciles are not normally distributed. For both non-stroke controls and stroke cases, the distribution is skewed towards the right or more disadvantaged. There was no difference in the median state ADI decile for stroke and non-stroke patients. However, when grouped into quartiles, patients living in the areas with greatest deprivation had a 28% increase in the odds of stroke compared to those patients living in the least deprived areas. In regard to national percentiles, there was no difference in the median ADI for stroke and non-stroke patients. When grouped into quartiles, it was the second-lowest national ADI quartile that showed a 23% increase in the odds of stroke. And lastly, this data isn't shown, but with walking distance to a public park, we found there was no impact on the odds of stroke, nor did residing with one kilometer of a park. We next looked at the correlation between green space and area deprivation. As expected, both state and national ADI were inversely correlated with NDVI. That is, as green space increases, area deprivation decreases. These tables show the result of conditional logistic regression looking for an interaction between area deprivation and NDVI as predictors for stroke. There is no interaction between the two variables, and because the regression coefficients for NDVI when used in a multivariate model with ADI and when used alone in a univariate model are similar, we conclude that NDVI is not simply a proxy for area deprivation and the explanatory power of NDVI is separate from that of ADI. There are limitations to the study. We do not know the mechanism of action of green space as a stroke risk factor. Prediction of the microclimate and promotion of social contact are two of the more popular theories. The dose and duration of green space exposure needed to impact health outcomes are similarly unknown, and we also do not have a method to capture how long a patient has lived at a particular address. To summarize, we found that as green space increases, the odds of stroke decreases, and patients living in homes with greater surrounding residential green space have a significantly lower odds of stroke. This effect of green space on stroke risk is independent of the traditional stroke risk factors on which we matched. This effect of green space on stroke is also a novel finding. We further found that patients living in the most disadvantaged neighborhoods have a significantly higher odds of stroke compared to those living in the least disadvantaged neighborhoods. And we also found that though NDVI and ADI are correlated, green space is not simply a proxy for ADI. In conclusion, our data add to the literature demonstrating a positive relationship between green space and health. This knowledge should inform public policy as urban green space reclamation and rejuvenation may be an effective intervention to improve health outcomes and lower health care costs. Green space is more and more quickly modifiable than are the factors that lead to socioeconomic inequalities. Thank you for watching our presentation, and please feel free to email me with any questions.

Association Between Residential Greenspace and Stroke

Aurora St. Luke's Medical Center

Green space and stroke risk

Normalized Difference in Vegetation Index (NDVI)

Area Deprivation Index (ADI)