Residential Surrounding Greenness and Cognitive Decline: A 10-Year Follow-up of the Whitehall II Cohort

Background: Evidence on beneficial associations of green space with cognitive function in older adults is very scarce and mainly limited to cross-sectional studies. Objectives: We aimed to investigate the association between long-term residential surrounding greenness and cognitive decline. Methods: This longitudinal study was based on three waves of data from the Whitehall II cohort, providing a 10-y follow-up (1997–1999 to 2007–2009) of 6,506 participants (45–68 y old) from the United Kingdom. Residential surrounding greenness was obtained across buffers of 500 and 1,000m around the participants’ residential addresses at each follow-up using satellite images on greenness (Normalized Difference Vegetation Index; NDVI) from a summer month in every follow-up period. Cognitive tests assessed reasoning, short-term memory, and verbal fluency. The cognitive scores were standardized and summarized in a global cognition z-score. To quantify the impact of greenness on repeated measurements of cognition, linear mixed effect models were developed that included an interaction between age and the indicator of greenness, and controlled for covariates including individual and neighborhood indicators of socioeconomic status (SES). Results: In a fully adjusted model, an interquartile range (IQR) increase in NDVI was associated with a difference in the global cognition z-score of 0.020 [95% confidence interval (CI): 0.003, 0.037; p=0.02] in the 500-m buffer and of 0.021 (95% CI: 0.003, 0.039; p=0.02) in the 1,000-m buffer over 10 y. The associations with cognitive decline over the study period were stronger among women than among men. Conclusions: Higher residential surrounding greenness was associated with slower cognitive decline over a 10-y follow-up period in the Whitehall II cohort of civil servants. https://doi.org/10.1289/EHP2875


Table of Contents
. Geographical distribution of the study population at three follow-ups. Table S1. Information on the exposure assessment: NDVI and EVI images used for the exposure assessment and information on missing values. Table S2. Description of Multiple Imputation. Table S3. Comparison between participants with complete data and participants with a missing value for at least one covariate. Median global cognition z-score and NDVI (1st quartile, 3 rd quartile). Table S4. Correlations between A. NDVI in the 500 meter buffers at different follow-ups and B. between different NDVI and EVI in different buffer sizes. Table S5. Sensitivity analyses. Difference (95% confidence interval) in global cognition and cognitive sub-z-scores at baseline and after 10 years of follow-up in association with a one interquartile range increase in residential surrounding greenness (NDVI). Table S6. Sensitivity analyses -further adjustment for neighborhood factors. Difference (95% confidence interval) in global cognition z score at baseline and after 10 years of follow-up in association with a one interquartile range increase in residential surrounding greenness (NDVI). Table S7. Sensitivity analysis -Inverse probability weighting. Difference (95% confidence interval) in global cognition z score over 10 years associated with a one interquartile range increase in residential surrounding greenness (NDVI). Table S8. Difference (95% confidence interval) in global cognition and cognitive sub-z scores at baseline and after 10 years of follow-up in association with a one interquartile range increase in residential surrounding greenness (EVI).

Number of imputed datasets created: 25
Variables included in the imputation procedure:

Variables used in the main analyses (complete observations of outcome and exposure, and covariates which could have missing values) together with other relevant covariates:
A cognitive score (global cognition, reasoning, fluency or short-term memory score), age, gender, ethnicity, marital status, educational attainment, household assets (including house), employment grade, IMD income domain, smoking, alcohol consumption, general health questionnaire depression subscale, network score, social support of close person 1, physical activity level, general health status, average PM2.5 in a 1000m buffer in year of follow-up, average PM10 in a 1000m buffer in year of follow-up, and a greenness indicator (NDVI in the direct neighborhood or EVI in the direct neighborhood) Treatment of non-normally distributed variables: matching.

Treatment of binary/categorical variables: logistic models.
Statistical interactions included in imputation models: none.   1997 -1999 2002-2004 2007-2009 1997 -1999 1 2002-2004 0  Table S 5. Sensitivity analyses. Difference (95% confidence interval) in global cognition and cognitive sub-z-scores at baseline and after 10 years of follow-up in association with a one interquartile range increase in residential surrounding greenness (NDVI).
All estimates are from linear mixed effect models that included NDVI, age, age squared, age x NDVI, gender, ethnicity (white, non-white), marital status (married/cohabiting, yes or no), alcohol use (frequency of consumption in the year prior to filling in the questionnaire; sometimes, daily, or never), diet (intake of fruit and vegetables; daily or less than daily), smoking status (current, past, or never), education (lower secondary school or less, higher secondary school, and university or higher), employment grade (high, middle, or low), and the Index of Multiple Deprivation (IMD) income and employment scores.    in global cognition z score over 10 years associated with a one interquartile range increase in residential surrounding greenness (NDVI). All estimates are from linear mixed effect models that included NDVI, age, age squared, age x NDVI, gender, ethnicity (white, non-white), marital status (married/cohabiting, yes or no), alcohol use (frequency of consumption in the year prior to filling in the questionnaire; sometimes, daily, or never), diet (intake of fruit and vegetables; daily or less than daily), smoking status (current, past, or never), education (lower secondary school or less, higher secondary school, and university or higher), employment grade (high, middle, or low), and the Index of Multiple Deprivation (IMD) income and employment scores.  All estimates are from linear mixed effect models that included EVI, age, age squared, age x EVI, gender, ethnicity (white, non-white), marital status (married/cohabiting, yes or no), alcohol use (frequency of consumption in the year prior to filling in the questionnaire; sometimes, daily, or never), diet (intake of fruit and vegetables; daily or less than daily), smoking status (current, past, or never), education (lower secondary school or less, higher secondary school, and university or higher), employment grade (high, middle, or low), and the Index of Multiple Deprivation (IMD) income and employment scores.