Association between urban green space and the risk of Alzheimer's disease: A longitudinal study


 BackgroundsAlthough studies have demonstrated the beneficial effects of green spaces on general well being, the benefits of urban green spaces on Alzheimer’s disease (AD) are still unclear. In this retrospective cohort study, we sought to ascertain the connection between the presence of green spaces in urban environments and risk of AD.MethodsThe study sample comprised 294,983 participants, aged 40–74 years, and was derived from the Korean National Health Insurance Service National Sample Cohort. We used data of area (m2) and amount of artificial parks in 2005, provided by the Ministry of Land, Infrastructure and Transport. Starting from 2009, after removing those who received a dementia diagnosis within 1–3 years of gathering the index data, participants received follow-up AD testing. The Cox proportional-hazards model was utilized to establish the adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs) of AD risk, based on the amount of urban green space area.ResultsCompared to the quartile with the least amount of green space, participants in the highest urban space quartile had a reduced risk of receiving an AD diagnosis (aHR 0.84, 95% CI 0.72–0.99). The mediating effect of green space presence on AD was stronger among those in higher household income brackets (aHR 0.80, 95% CI 0.65–0.99), with more comorbidities (aHR 0.78, 95% CI 0.64–0.96), and without disabilities (aHR 0.82, 95% CI 0.69–0.97).ConclusionsLiving in urban areas with increased green space coverage may ultimately result in a lower risk of being diagnosed with AD. Urban planning strategies capable of boosting the total coverage of urban green space may assist in reducing AD risk.

risks and protective factors that can be altered in order to reduce the risk of dementia (4). According to the Alzheimer's Association, there is su cient enough evidence to conclude that regular exercise, management of cardiovascular risk factors (obesity, diabetes, smoking, and hypertension), and cognitive training are capable of reducing risks associated with cognitive decline (5). Based on these ndings, a new focus on the importance of strategies to prevent incidence of AD within living environments has emerged.
Green space is recognized as a exible xture of the modern metropolitan environment often associated with the health of inhabitants. Studies have shown that people who reside in urban settings enjoy more healthy lifestyles, regardless of their demographic or economic status, when green space is more readily available (6,7). Due to increased urbanization, people are facing the reality of fewer green spaces in their living environments, which may negatively affect their health (6). It is believed that green space could reduce cardiovascular and premature overall mortality, while also improving mental health and overall cognitive functions (7). The presence of green spaces within local neighborhood environments is associated with decreased rates of poor health in self-reporting surveys (8), as well as all causes of mortality, respiratory disease, and circulatory diseases (9). According to a study from New Zealand, neighborhood green space was linked to increased cardiovascular and cognitive function, independent of individual risk factors (10). Prior studies have examined the connection between green spaces in urban settings and general health or health behaviors related to dementia. However, few studies have researched how green spaces affect the risk of AD diagnoses in urban areas.
Based on the evidence of the positive bene ts of green spaces on physical, mental, physical, cognitive, and social health, which would be considered critical elements of AD risk prevention, we sought to look at the prevalence of green space in urban areas and whether it can be an indicator for AD risk.

Study population
The research sample was obtained from the database of the National Sample Cohort of the Korean National Health Insurance Service (NHIS-NSC). The Sample Cohort database comprised bene ciaries' sociodemographic information; medical claims dataset, including diagnostic information from the 10th revision of the International Classi cation of Disease (ICD-10); admission, treatment; and the partner members' National Health Screening datasets. The information for the national health screenings was created by the NHIS and provides information through a health check-up database for all Korean inhabitants from 2002 to 2013; it also contains information on biological factors health exams and health habits. Korea's NHIS requires Koreans to join the national health insurance scheme, which covers nearly all available health services (11), including semiannual check-ups for people 40 years or older (12).
Multiple previous epidemiological trials have utilized the NHIS database, and outlines con rming their legitimacy have been de ned in detail elsewhere (13,14).
We extracted our data from data obtained in 2005 for 476,985 participants from seven cities in Korea. Of these, 294,983 participants aged < 40 or > 74 years were excluded. We also excluded 848 participants without green space values or readily available total land area data, as offered by the Ministry of Land, Infrastructure, and Transport (MOLIT). Then, 980 and 823 respondents were excluded due to death or diagnosis of any form of disease prior to the indexing date. The study population was comprised of 179,351 respondents. Starting on 1 January 2009, the respondents were followed until the date they rst received AD treatment, their death, or 31 December 2013.

Ethical Considerations
Seoul National University Hospital's institutional review board approved this study (IRB number: E-1803-045-928). They removed the request for informed approval, as the NHIS-NSC database is anonymized and follows rigorous con dentiality rules. We did not include natural green space as de ned in our study, such as elds, mountains, and forests, into urban green space. Seven Korean metropolitan cities have been divided to 72 administrative districts that match the metropolitan areas where participants live. The area of districts areas varied between 2.8 km 2 and 755.0 km 2 , with a mean of 55.1 km 2 (standard deviation, SD = 79.9 km 2 ). The coverage of green spaces in each urban district was de ned as a proportion (%). All green spaces were classi ed into quartiles with speci c threshold values 0.0-0.34, 0.34-0.92, 0.94-2.33, 2.38-15.3. In our study, we de ned Alzheimer's Disease (AD) based on a positive AD diagnosis (ICD-10 codes F00, G30) with the prescription of dementia-related drugs, including donepezil, galantamine, rivastigmine, and memantine (15).

Statistical analysis
The Cox proportional-hazard model was utilized to ascertain adjusted hazard ratios (aHRs) and 95% con dence intervals (CIs) for AD risk. We also conducted a strati ed analysis of the connection between green spaces in urban settings with AD based on age, gender, physical activity, household income, comorbidity, and disability subgroups. Finally, we conducted an analysis of sensitivity to the effects of green spaces in urban environments on AD; however, participants who received a dementia diagnosis within 5 years of the indexing date were excluded. The statistical signi cance was treated as P < .05 from a two-sided approach. We collected data using SAS 9.4 (SAS Institute, Cary, NC, USA) and statistical analysis with STATA 13.0 (Stata Corp LP, College Station, TX, USA). Table 1 offers insights into the characteristics of the study participants. The average green space value is 1.67% overall, and the individual values of the four quartiles, from smallest to greatest, are 0.17%, 0.64%, 1.58%, and 4.54%, respectively. The number of participants who received health screenings was 80,170, representing 44.7% of all participants. The group with the highest household income tended to reside in neighborhoods containing larger amounts of green space available (13,613, 14,854, 14,997, and 15,935 participants from the lowest to the highest quartile). On the other hand, the group with the lowest household income tended to live in areas with less urban green space (8,204,7,802,7,120, and 6,157 participants from the lowest to the highest quartile). Table 2 illustrates the reduced risk associated with green space coverage of developing AD inversely. By comparison to those in the rst (smallest) quartile, those in the second, fth, and fourth (largest) quartiles tended to experience a decreased risk of AD (aHR 0.98, 95%, CI 0.84-1.14; aHR 0.96, 95%, CI 0.82-1.11; aHR 0.84, 95%, CI 0.72-0.99, respectively). The risk of AD tended to be reduced with increasing green space coverage with signi cance as well (p for trend = 0.045). Table 3 shows a strati ed analysis of the impact urban green spaces have on AD based on the subgroups of age, sex, household income, comorbidity, and disability. When comparing the respondents in the rst (smallest) green space quartile to other quartiles, those with greater household income (aHR 0.80, 95%, CI 0.65-0.99, p for trend = 0.027), with more comorbidities (aHR 0.78, 95%, CI 0.64-0.96, p for trend = 0.021), and those without any impairments (aHR 0.82, 95%, CI 0.69-0.97, p for trend = 0.022) tended to have decreased risk of AD. An inverse relationship between the presence of green space and the prevalence of AD was discovered during the subgroup analysis based on age and sex, but without signi cance. Table 4 reveals the impact of urban green spaces on AD prevalence in participants after receiving dementia diagnoses within 5 years of the indexing date. After excluding participants who received diagnoses of dementia within 5 years after the indexing date, the risk of AD tended to be consistent (aHR 0.83, 95%, CI 0.69-1.00, p for trend = 0.038 ). Supplement Table 1 represents the association of urban green space on AD among individuals who underwent health screening examinations. We found a trend of reduced AD risk with increasing green space coverage, but with no signi cance (aHR 0.81, 95%, CI 0.62-1.04, p for trend = 0.139). After adjusting all covariates, including biological factors and lifestyle behaviors, age, gender, household income, housing region, comorbidity, BMI, total cholesterol, blood pressure, fasting blood sugar, smoking status, level of physical exercise, and alcohol intake were still not signi cant, but the trend towards lower danger of AD (aHR 0.81, 95%, 0.62-1.06, p for trend = 0.161) was maintained.

Discussion
We have shown that participants in neighborhood settings with a greater degree of green space available have less risk of developing AD. As far as we know, ours is the rst study to show how signi cant the relationship between residential green space and AD risk is for general populations in Asia. Some modi able risk factors for AD, physical inactivity, depression, and social isolation in late life (16), could be attenuated for people living with green space. Several feasible theories may support the preventive effects of the presence of green space in urban settings on AD risk, improving physical, mental, and cognitive health, which are known attenuators of AD risk. Physical activity is also shown to increase when greener surroundings are readily available (17)(18)(19). Green space availability in neighborhoods could also promote exercise, such as walking, for middle-age and older adults (17). Respondents living in low proximity to local green spaces were shown to be less prone to visiting green spaces than respondents living in close proximity to green spaces (20). Some Western studies have previously demonstrated that activity levels increase for those living in neighborhoods with larger areas of green space (21,22). Likewise, for Korea, increased levels of physical activity were linked to areas with more public parks (18). Many prospective studies have shown that the likelihood of developing AD signi cantly decreases in participants who engage in exercise activities more often (23)(24)(25)(26)(27)(28). Another 17year-long study of elderly Japanese subjects has also shown physical activity prevent AD (29). Recently, a randomized, controlled study of elderly people who had not been diagnosed with dementia revealed that exercise could increase hippocampal volume and enhance overall cognitive function (30).
Individuals who were more physically active exhibited lower levels of Aβ deposits in Aβ screenings, which are associated with the clinical conditions of AD patients (31).
The neighborhood environment, as a mechanism for urban green spaces to bene t mental health, has been introduced (32-34). It has also been shown that access to green spaces over the long term potentially plays a protective role, bene ting mental health and alleviating anxiety and depression (35).
People residing in environments with less access to green spaces reported more feelings of solitude and lack of social connection compared to respondents who lived in neighborhoods with more green space access (36). Green space was directly associated with reduced levels of stress among adults in older age groups (37) and could alleviate the negative health effects of stress (38). There was a protective association between green spaces and depression and suicide predictors, which are potential indicators of future AD risks, according to the study on the Korean population (39). According to two meta-analyses, depression can signi cantly increase the risk of a dementia diagnosis, by as much as twofold (40,41). Several studies also suggested that symptoms of depression were associated with AD risk (42)(43)(44). There is a neuropathological correlation between dementia and depression, and this can be observed via alterations in glucocorticoid steroids and hippocampal atrophy, increased deposits of β-amyloid plaques, in ammatory changes, and de cits of nerve growth factors or regulators like neurotrophins (45). According to these ndings, it is plausible to assume more urban green spaces are necessary for assisting the elderly population, who are among the most vulnerable groups when it comes to certain mental health issues that are known AD risks.
Urban green space may play a protective role for cognitive function, which could in turn reduce the overall risk of AD. A greater likelihood of cognitive decline and dementia has been linked to a lower proportion of nearby green spaces and private garden spaces (46). Neighborhoods making public parks available for residents throughout their lives may help reduce the rate of mental impairment residents may be more likely to experience at a later stage in their lives (47). Parks in urban environments may act as a more stimulating space from a cognitive perspective compared to other types of green spaces, such as green spaces in non-urban settings (48). In addition, greenness in neighborhood areas can facilitate social relationships, develop social cohesion, and ultimately create a sense of solidarity among community members (49), which could also bene ts individuals' cognitive function. In a Taiwanese cohort study with 11 years of follow-up, it has been reported that social activities share an inverse relationship with a decline in scores on a Mental Status Questionnaire (50). Increased pursuit of physical, intellectual, and social activities in addition to increased cognitive capacity have been linked with a decreased risk of receiving a diagnosis of dementia (51). There are complex and heterogeneous mechanisms associated with how green spaces affect cognitive ability. The protective effect from greenness on physical, mental health, and social cohesion may result in improved cognitive health and possible prevention of AD.

Limitations and Strengths
Several limitations should be considered in our study. First, we limited the type of urban green spaces to urban parks, excluding natural elds and wooded areas, as we decided to determine the effect of arti cial green spaces. Second, due to the lack of data, the residential district codes of our participants did not exactly match to addresses of respondents in order to preserve anonymity. Thus, NHIS-NSC residential district codes are unlikely to correlate with the addresses gathered for green space data, including exact addresses and GIS maps. Nevertheless, there are a couple of studies that have used participants' district codes and environmental data, and we adopted the same methodological design (39,52). Third, we had to rely on prescription and diagnostic codes for medications, as we were unable to access dementiarelated medical chart documents. However, for patients to receive reimbursement from the NHIS for antidementia drugs, two requirements have to be met; rst, they must receive an Mini-Mental State Examination (MMSE) rating of 26 or less; and secondly, they must receive either a Clinical Dementia rating of 1 or higher or a Global Deterioration Scale rating of 3 or greater. 17 Even though the diagnostic reliability of dementia seems to increase, there is a need for researchers in the future to use medical records to achieve a more accurate ndings to complement the results of our study. Finally, cautious interpretation is needed, due to the fact that only a 5-year follow-up period was used to discover the association of green space prevalence and AD diagnoses. Despite limitations, our study also features a number of strong points. First, in contrast to most of previous studies indirectly or directly reported the impacts of physical activity, mental well being, or social cohesion on AD risk, we rst discovered the link between green space coverage, a potential candidate for environmental intervention, on the long-term prevalence of AD. Second, we utilized data from a nationwide database of more than 290,000 participants, aged 40-74 aged, from seven metropolitan cities. Third, through a sensitivity analysis, we consolidated the data to determine the effects of urban green spaces on reduced AD risk with a 5-year washout period.

Conclusions
This study indicates that living in urban areas that feature adequate green spaces may ultimately result in reduced AD risk among adults in middle age, as well as elderly adults. Our ndings bolster the position that urban planning policies should be created to utilize urban green spaces to assist in reducing AD. It may be bene cial to create new or additional green spaces in urban environments or to look into developing outdoor activities capable of properly utilizing existing resources to reduce AD risk. Future long-term follow-up studies investigating whether greater urban green spaces actually result in reduced AD risk should be carried out to validate our study's ndings.

Declarations Ethics approval and participation consent
This study has been approved by Seoul National University Hospital's institutional review board (IRB number: E-1803-045-928). They removed the request for informed approval, as the NHIS-NSC database is anonymized and observes rigorous con dentiality rules.

Consent for publication
Not applicable

Availability of data and materials
The supporting data for this study's ndings is available from the Korean National Health Insurance Service (KNHIS) and the Ministry of Land, Infrastructure, and Transport (MOLIT) but restrictions apply to the availability of this data, which was licensed for this study, and so it is not publicly available. However, data is available from the authors, if a reasonable request made, and with permission of KNHIS and MOLIT.

Competing interests
None of the authors reported any disclosures.    Abbreviation: aHR, adjusted hazard ratio; CI, con dential interval a Urban green space coverage was de ned as the area of parks and arti cially designed green space divided by the area of residential districts. b Hazard ratio calculated by Cox proportional hazards regression analysis after adjustments for age, sex, income, Charlson comorbidity index, and disability.