Question

Findings

Meaning

Natural outdoor environment attributes, such as green spaces (i.e., forests or parks), blue spaces (i.e., visible bodies of water), and coastal proximity, have long-term effects on behavior and health (1–3). Researchers have provided preliminary evidence that proximity to the coast, an essential natural outdoor environment attribute, is positively related to both self-reported general and mental health, and the beneficial effect of coastal proximity was mainly mediated by improving behavioral pathways (such as physical activity, sleep, and diet), alleviating stress, and avoiding environmental pollution (4–6). However, most of these studies are limited by a small sample size, weak geographical representation, insufficient adjustment for confounders, and unclear definition of exposure. Besides, most of the outcomes of previous reports were self-reported and not relating to specific diseases. Hence, little is known about how the residential distance to the coast is associated with the incidence of myocardial infarction (MI) and which mechanisms may explain the association.

To deal with these limitations, we explored the association between distance to the coast and incident MI using data from UK Biobank, a prospective, population-based cohort study with unprecedented sample size and broad geographical coverage.

The baseline characteristics of the participants are summarized in Table 1. The mean value of residential distance to the coast for 377,340 participants was 45.7 ± 26.7 km. Briefly, participants who lived proximal to the coast (<1 km) were more likely to be white, retired, less social deprivation, less diabetes, more physically active, and exposed to less NO₂ air pollution and traffic intensity. Supplementary Table S1 shows the characteristics of participants by quintiles (Q1–Q5) of residential distance to the coast.

During a median of 8.0 years (3.0 million person-years) of follow-up, 4,059 cases of MI occurred. Table 2 shows the association between residential distance to the coast and incident MI. After adjusting potential confounders, no significant association with MI was observed for distance to coast as continuing variable (HR 0.98 per SD increase, 95% CI 0.95–1.02, p = 0.321). Using group (<1 km) as reference, group (20–50 km) was associated with a statistically significant lower risk of MI (HR 0.79, 95% CI 0.64–0.98, p = 0.033). When using the lowest quintile Q1 (<14.1 km) as reference, Q2 (14.1–40.2 km), and Q3 (40.2–56.4 km) were both associated with a statistically significant lower risk of MI (HR 0.86, 95% CI 0.78–0.95, p = 0.003; and HR 0.85, 95% CI 0.77–0.94, p = 0.001, respectively; p for trend = 0.586, Table 2).

We then applied the restricted cubic spline curve to explore potential non-linear patterns. Figure 1 shows a U-shaped relation between distance to coast and MI with a relatively lower risk interval between 32 and 64 km (p for non-linear = 0.0012). According to the curve, we divided the population into three categories: offshore region (<32 km), inland region (>64 km), and intermediate area (32–64 km). Using participants of the intermediate region (32–64 km) as a reference, participants of the offshore region (<32 km), and inland region (>64 km) were both associated with a higher risk of incident MI and HRs were 1.12 (95% CI 1.04–1.21, p = 0.004) and 1.09 (95% CI 1.01–1.18, p = 0.027) after adjusting for all confounders, respectively (Table 3).

Figure 2 shows the associations between participants living in the offshore region (<32 km) and incident MI in subgroups analyses. HR for the offshore area (<32 km) was higher in the subgroup with low total physical activity (<24 h/week) (HR 1.24, 95% CI 1.09–1.42, p = 0.001), compared with subgroups with moderate and high total physical activity (p for interaction = 0.043). Figure 3 shows the associations between participants living in the inland region (>64 km) and incident MI in subgroups stratified by potential effect modifiers. HR for the inland region (>64 km) was significantly larger in the subgroup of the urban area (HR 1.12, 95% CI 1.03–1.22, p = 0.007), compared with the subgroup of the rural area (p for interaction 0.065). Also, HR for the inland region (>64 km) was significantly larger in the subgroup of high NO₂ air pollution exposure (HR 1.29, 95% CI 1.11–1.50, p = 0.001), compared with middle and low NO₂ air pollution (p for interaction = 0.021).

The main finding of the current study was that residential distance to coast had a U-shaped relation with incident MI in over 370,000 individuals followed for over 3.0 million person-years and that both residents of the offshore region (<32 km) and that of the inland region (>64 km) had around 10% increase risk of MI compared with residents of the intermediate region (32–64 km) (Figure 4). Moreover, the associations of offshore region and inland region with incident MI were modified by various factors. Participants of offshore regions with low total physical activity had a higher risk of MI (increased by 24%), suggesting that this subgroup may benefit from increased physical activity. Meanwhile, participants of inland regions living in the urban area or exposed to high NO₂ air pollution had a higher risk of MI (increased by 12 and 29%, respectively), suggesting that urban environment improvement and air pollution control played a crucial role in these population. The results were mostly consistent in a series of sensitivity and subgroup analyses.

Unlike previous research that commonly reported linear associations, we investigated the non-linear association of coastal proximity with disease outcome. Our findings highlight the complex and diverse associations between residential distance to the coast and incident MI. Although modifying the residential environment tends to be problematic in the short term, our findings suggest that targeting based on coastal proximity and various effect modifiers (physical activity, population density, and air pollution) could help identify high-risk individuals and provide personalized interventions for diverse populations. However, the classification of the offshore, inland, and intermediate regions was constructed for illustrative purposes; hence the cut-off values of distance to coast identified in this study cannot be simply generalized. This classification is more about raising public awareness of the residential location than about being implemented as a real-world tool. The harmful effect of residential distance to coast on incident MI may vary with the coastline's distance and is regulated by various factors. Therefore, when possible, advice on the living environment and health should be personalized.

Scientific and public interest in the role of the natural outdoor environment in preventing CVDs is growing. A tremendous body of studies has provided evidence on the associations between various outdoor environment attributes, behavioral pathways, CV risk factors, and mortality (18–21). A meta-analysis of longitudinal studies found inverse relationships between neighborhood walkability and risk factors, such as obesity, hypertension, and type 2 diabetes mellitus (20). Similarly, another meta-analysis found that more residential greenspace was associated with reduced CVD mortality (22). The links between air and noise pollution, stress, and CVDs have also been recognized (23). The previous research suggested that living near and regularly visiting the coast or other large waterbodies was associated with better general and mental health. However, most of the outcomes used were self-reported and not disease-specific (24–26). However, the association of residential distance to the coast with CVD incidence is relatively less well studied, and high-quality epidemiological evidence was too scant to draw a conclusion. To investigate this issue at a broad level of disease specificity, we set out to address the question: does the incidence of MI increase or decrease with proximity to the coast?

The results of the study were somewhat unexpected, particularly the finding that MI risk increased in coastal areas, which contradicts most of the previous studies that have shown the benefits of living near the sea. However, our study was based on sufficient confounding adjustment and subgroup analyses, and the results were reliable and consistent. One possible explanation is that the same environment attribute might have diverse effects on cardiometabolic risk factors and different population subgroups. For example, whereas high population density might help improve the weight status through better availability of physical activity destinations and healthy food options, high population density might also have adverse effects on airway and CVD through more exposure to air pollution. Similarly, a previous study showed that individuals living in the <1 km coastal category had an average a 4 nmol/l higher vitamin D status compared to those living inland through increased solar irradiance, which can provide benefits in terms of vitamin D status but may also pose a risk due to higher skin cancer rates (27). Examining single environmental attributes with multiple behaviors and risk factors in single studies can provide insights into the differential effects of the natural environment on CVD. Further interdisciplinary research initiatives involving cardiology and urban design researchers must disentangle the complex relationships between the residential coastal distance and CVD.

Strengths of this study include its prospective design, a large sample size with harmonized exposure, health, and covariate data. We could adjust for a wide range of health, demographic, behavioral, and environmental confounders. The possibility of confounding was dealt with through statistical adjustment for a wide range of covariates, such as health, demographic, behavioral, and ecological confounders, and a series of sensitivity analyses. This study has several limitations. First, we were not able to consider the impact on the exposure of residential changes during follow-up, which will contribute to misclassification of long-term exposure relevant to the development of MI. These misclassifications are believed to be non-differential for cases, and non-cases likely bias the risk estimates toward the null. Second, as is the case for any observational study, residual confounding is always possible, and associations may not imply causation. We cannot rule out the possibility of residual confounding by other unaccounted factors, such as coastal climate, humidity, and sun exposure. Third, the UK Biobank represents the general population for age, sex, ethnicity, and deprivation within the age range recruited but is not representative in other regards, which may indicate a healthy volunteer selection bias. While this limits the ability to generalize prevalence rates, it should be possible to generalize the estimates of the associations' magnitude. Forth, the offshore, inland, and intermediate regions were constructed for illustrative purposes rather than as a tool ready for implementation. The cut-off values of distance to coast (32 km, 64 km) have not been validated.

The study found a U-shaped association between residential distance to the coast and incident MI. Moreover, the association of offshore region with incident MI was modified by total physical activity. The association of inland region with incident MI was limited by urban/rural area or NO₂ air pollution. Our findings highlight the complex and diverse associations between residential distance to the coast and incident MI, and residential advice should be personalized.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

ZX-d performed statistical analysis. LL-z handled funding and supervision. LX-x acquired the data. ZS-z conceived and designed the research. HX drafted the manuscript and made critical revision of the manuscript for critical intellectual content. All authors contributed to the article and approved the submitted version.

The UK Biobank was supported by the Welcome Trust, Medical Research Council, Department of Health, Scottish Government, and the Northwest Regional Development Agency. It has also had funding from the Welsh Assembly Government and the British Heart Foundation. The research was designed, conducted, analyzed, and interpreted by the authors entirely independently of the funding sources.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.