Urban green infrastructure and pandemic response: how urban parks were used to support the COVID-19 response and the relationship between parkland acreage and excess mortality in large cities in the USA

ABSTRACT

The coronavirus disease 2019 (COVID-19) pandemic was an unprecedented public health emergency, and relatively little research has investigated the relationship between urban green infrastructure, such as parks, and pandemic-related health outcomes. This study describes how urban parks were used to support the pandemic response and examines the associations between urban park acreage and excess mortality in 2020 in the 50 most populous cities in the USA. The descriptive portion of the study revealed that all cities temporarily closed at least some parks during the first year of the pandemic, and most reported collaborations between parks and public health authorities. Further, urban parks were used to support the pandemic response by serving as venues for meal distributions, diagnostic testing, education and childcare, housing shelters, vaccination clinics, personal protective equipment distributions and other services. A series of linear regression models were used to investigate the association between park acreage and excess mortality. Univariable regression revealed a negative, statistically significant relationship between total urban park acreage and excess mortality rates. However, the relationship weakened and was no longer significant when additional demographic and socioeconomic variables were added in multiple linear regressions. In combination, these results may inform efforts to optimize the design of urban parks and strengthen urban resilience against future infectious disease outbreaks, especially if they consider and address aspects such as park accessibility and equity.

INTRODUCTION

The emergence of the severe acute respiratory syndrome coronavirus 2 virus (SARS-CoV-2) in China in late 2019 resulted in an unprecedented infectious disease outbreak—the coronavirus disease 2019 (COVID-19) pandemic. The rapid, global spread of the virus resulted in the World Health Organization declaring a public health emergency of international concern in January 2020 and later characterizing the outbreak as a pandemic in March 2020 [1, 2]. This makes COVID-19 the first pandemic to confront humanity since humans became a predominantly urban species in the mid-2010s [3]. In response to this emergency, governments at all levels were required to implement outbreak responses. Among other actions, these responses included ‘lockdown’ measures at both the individual and community levels [4, 5]. The isolation of unwell persons and the quarantine of persons known or suspected of being exposed to disease comprised individual-level measures, while community-level measures involved the temporary closure of schools, businesses and certain recreational facilities, such as parks.

In urban environments, parks and other varieties of green infrastructure (i.e. gardens, trails, etc.) are widely recognized to provide and promote an array of health benefits; these include those related to physical and mental health, such as increased physical activity, lower obesity levels, lower stress levels, improved mental well-being and increased social cohesion [6–11]. Notably absent from this list are considerations related to infectious diseases. While the threats posed by infectious diseases are ever-present, they have largely been omitted from recent debates on shaping cities and urban green spaces for health and well-being [11, 12]. Instead, as suggested by the previous list, these debates have been primarily defined by non-communicable diseases and chronic conditions—a reality only made possible by the development and widespread use of antimicrobials, vaccines and other medical countermeasures geared toward controlling the spread of infectious diseases.

Still, the perception of parks as a means to improve public health predates the epidemiological transition that resulted from these technologies. The transition to a disease burden predominantly characterized by non-communicable diseases did not occur until the mid-20th century [13]. Prior to this time, urban parks were also widely regarded as being protective against infectious diseases. Indeed, while urban planning and public health represent two distinct disciplines today, much of urban planning and design have roots in 19th-century efforts to prevent the spread of infectious diseases [11, 14, 15].  For instance, in the USA, the development of many early urban parks was justified by their perceived potential to offer protection against various scourges such as cholera, malaria, tuberculosis and yellow fever [16, 17]. In the 1850s, for example, authorities, urban designers and landscape architects in New York City were concerned that the topographical and drainage features of Manhattan rendered the city vulnerable to infectious disease epidemics [18]. These concerns led to a proposal and design competition for an urban park that would improve water drainage and also serve as a recreational space—eventually leading to Frederick Law Olmsted and Calvert Vaux’s Greensward Plan for a space that is now known as Central Park [17, 18].

Acknowledging this history, the COVID-19 pandemic presents a unique opportunity to reconsider this history and reexamine the relationship between urban parks and infectious diseases. This is especially true when considering that, because of its novelty, there were few existing medical countermeasures offering protection from morbidity or mortality for much of the first year following its emergence. Others have noted that while the relationship between urban design characteristics—such as land use, connectivity and polycentricity—and pandemic health metrics has spurred several debates, the relationship between green infrastructure and health outcomes remains poorly understood and should be a priority for research efforts [11, 19–21]. To date, however, much of the research on urban parks and the COVID-19 pandemic has focused on park access and visitation trends during the pandemic [22–24]. The limited research that has focused on urban green infrastructure and pandemic-related health metrics has examined mental health outcomes rather than morbidity or mortality [25, 26]. Research conducted at different spatial scales has suggested that there may be significant associations between various measures of ‘greenness’ and pandemic-related health measures. Klompmaker et al. found that county-level measures of normalized difference vegetation index data were inversely associated with COVID-19 incidence rates [27]; other work by Russette et al. also found that increases in county-level leaf area index were associated with lower levels of mortality [28]. Still, the results of county-level analyses do not always translate to finer spatial scales [29].

As such, a knowledge gap remains regarding how urban green infrastructure was used to support the pandemic response and the relationship between these spaces and local-level pandemic-related health outcomes. This study aims to close this gap and examine the relationship between urban parks and mortality rates during 2020, as well as describe how parks and recreational facilities were used to support the response to the COVID-19 pandemic. More specifically, it investigates associations between parkland acreage as a percent of a city’s area and excess mortality rates in the 50 most populous cities in the USA and hypothesizes that cities with more green space will have lower excess mortality rates during the first year of the pandemic. This hypothesis rests on the premise that cities with greater park acreage provided greater opportunities for residents to socialize and gather outdoors, ultimately resulting in lower levels of morbidity and mortality. Such information could inform urban planning and future efforts to optimize the design of parks as a means of bolstering urban resilience against infectious disease threats.

MATERIALS & METHODS

Study population

The population in this study included the 50 most populous cities in the USA according to the decennial census conducted in 2020. Cities were categorized according to their population size (i.e. ≥1 000 000, 750 000–999 999, 500 000–749 999 and ≤500 000) and region according to the US Census Bureau [30].

Descriptive analysis

For the descriptive analysis, data were gathered from the Trust for Public Land’s City Park Facts Survey [31]. More specifically, the survey distributed in 2020 included several questions inquiring about (i) whether park facilities had been closed during that year in response to the pandemic, (ii) whether park staff had worked with public health staff (i.e. meeting, coordinating, permitting, etc.), (iii) whether park facilities had been used to help address COVID-19 and (iv) what purposes they served if they had been used. The first three questions were posed to survey respondents in a binary format (i.e. yes/no), while the final question used a free-response format that allowed survey respondents to detail how parks and recreational facilities had supported alternate functions and services during the pandemic response.

Data analysis

Analysis used a grounded theory methodology to iteratively develop a binary coding scheme for the free-response questions [32]. The initial coding review identified all response services and activities supported by park facilities. Once this initial review was complete, an intermediate review was conducted to consolidate the services and activities into categories of alternate functions. A final review was then conducted to ensure internal consistency and completion before the results were aggregated at the city level.

Regression analysis

Dependent variable

The dependent variable was the excess mortality rate per 100 000 persons in 2020. Accurately tracking COVID-19 morbidity and mortality across different geographic areas has been challenging due to differences in testing practices, definitions and classifications, reporting procedures and quality control measures [33, 34]. In the USA, with regard to COVID-19 mortality, there is evidence suggesting that the attribution of deaths to COVID-19 may not be uniform across the country [35, 36]. Excess mortality—defined as the difference between the observed numbers of deaths and expected numbers of deaths in a specified population and period of time—is viewed as one metric for overcoming these challenges because it removes these biases. Previous work has produced local-level estimates for excess mortality in the USA in 2020, which were collected and used in this study [36].

Independent variable

The independent variable of interest for the regression models was parkland acreage as a percent of a city’s area. These data were collected from the Trust for Public Land’s City Park Facts Survey. Data collected in 2020 and published in 2021 were used in this study [31].

Control variables

Other data that may influence the relationship between park acreage and mortality were also collected and included as control variables in multiple linear regression models. These variables included urban design considerations, demographic considerations and socioeconomic considerations.

Urban design considerations included park access and population density. Park access was included as the amount of urban park acreage is unlikely to impact mortality if it is not accessible to those living in a city. Further, theory suggests that population density could impact mortality rates because localities with denser populations may experience greater amounts of disease transmission and mortality [37]. Accordingly, data relating to park accessibility (proportion of a population that lives within a 10-minute walk to a park) and population density (thousands of persons per square mile) were collected to be considered for inclusion in multiple linear regression models.

The demographic considerations included population characteristics, such as the proportion of a population that identifies as a racial minority, as well as biological vulnerabilities, such as older age and obesity. Marginalized populations and racial minorities have historically born a disproportionately large burden of morbidity and mortality during large infectious disease outbreaks [38]. There is now abundant evidence that racial minorities in the USA were at greater risk for more severe disease and mortality during the COVID-19 pandemic when compared to white populations [19, 37, 39–43]. Additionally, older age and certain health conditions like obesity have been linked to more severe disease and worse COVID-19 outcomes [20, 42–44]. As such, data relating to the racial composition (percent of population identifying as white), the elderly population (percent of a population ≥65 years of age) and obesity (percent of adults that are obese) were collected for inclusion as control variables.

Finally, evidence suggests that socioeconomic considerations can influence observed pandemic-related mortality rates. For example, poverty, inequality, education and insurance status can all impact COVID-19 mortality [19, 20, 37, 42, 43, 45, 46]. To account for this, data relating to income (median household income in 2020), income inequality (the ratio of household income at the 80th percentile to the 20th percentile), education (percent of population ≥25 years with a bachelor’s degree or higher) and insurance (percent of population <65 years of age without health insurance) were collected for inclusion as control variables.

These data were retrieved from sources maintained by the Trust for Public Land, the US Census Bureau and the US Centers for Disease Control and Prevention. For a full list of cities included in the study, population sizes, dependent, independent and control variables and other key data, see the dataset included as an appendix to this study (Supplementary Data).

Data analysis

Individual cities were the unit of analysis for the data analyses. The dependent, independent and control variables were investigated and summarized prior to conducting data analyses. Assumptions of linear regression were tested; in order to satisfy linear regression assumptions, excess mortality rates were transformed by taking the natural log. The relatively small study population also increases the risk of overfitting in regression analyses. To account for this, control variables were considered for inclusion in multiple linear regression models using a univariable screening process with a pre-determined alpha value set at 0.25 (P < 0.25). A univariable linear regression model was used to investigate the association between city park acreage and the natural log of excess mortality rate before multiple linear regression models were constructed by progressively adding blocks of control variables.

These analyses excluded Indianapolis (IN) and Las Vegas (NV) because there were no data for city park acreage; New York City (NY) and El Paso (TX) were also excluded because they represented outliers for excess mortality rate. The regression analysis was conducted using Stata BE/17.0 (College Station, TX); statistical significance was defined as an alpha value of ≤0.05 (P ≤ 0.05).

RESULTS

Descriptive analysis

A total of 361 respondents completed the survey. This included individuals from 50 city park & recreation agencies, 130 other public agencies and 181 private or non-profit park conservancies. On average, there were 7.2 respondents per city, with a range of 1–44 respondents per city.

All 50 cities reported that at least some parks were temporarily closed at some point during 2020. Forty-six (92.0%) of the cities reported that their parks and recreational spaces were utilized for alternate purposes during the first year of the pandemic, and 48 (96.0%) reported collaborations between park and public health authorities.

The iterative coding process resulted in the identification of seven common alternate functions supported by urban parks and recreational facilities. More specifically, 39 cities (78.0%) reported using facilities to support meal distribution services, 23 cities (46.0%) reported using facilities to support COVID-19 diagnostic testing services, 23 cities (46.0%) reported using facilities to support education and childcare services, 10 cities (20.0%) reported using facilities as temporary housing and shelter for homeless populations, 8 cities (16.0%) reported using facilities to support vaccination campaigns, 5 cities (10.0%) reported using facilities to support the distribution of personal protective equipment (PPE; e.g. facemasks) and 7 cities (14.0%) reported using facilities for other uses (Tables 1 and 2). Of the eight cities that reported using parks and recreational facilities to support vaccination campaigns, seven reported that they were used for COVID-19 vaccination efforts, while one reported that they were used to for seasonal influenza vaccination clinics.

Other, less common, alternate uses included supporting hospital surge capacity (n = 4; 8.0%), wellness checks (n = 2; 4.0%), blood drives (n = 2; 4.0%), cooling centers (n = 2; 4.0%), voter registration and polling efforts (n = 2; 4.0%) and census registration efforts (n = 1; 2.0%).

Regression analysis

On average, cities reported 129.09 excess mortalities per 100 000 persons in 2020, and 10.97% of a city’s land area was parkland acreage. Summary statistics for all variables considered for included in analyses are presented in Table 3.

Of the 10 control variables that were considered for inclusion in regression models, five met the prespecified threshold (Table 4). These included the elderly population, obesity, income, income inequality and education. The univariable linear regression showed a negative, statistically significant relationship between excess mortality and parkland acreage (Std. β = −0.303; P = 0.041) (Table 5), suggesting that an increase of 1 SD of parkland acreage as a percent of a city’s area resulted in a decrease of 0.303 SD in the excess mortality rate. However, the results lost significance when controlling for demographic considerations (Std. β = −0.072; P = 0.629) and socioeconomic considerations (Std. β = 0.030; P = 0.795). The adjusted R-squared value of the full model suggests that the incorporated variables explain ~60% of the observed variance in excess mortality rates.

DISCUSSION

Understanding the relationship between urban design characteristics, such as park acreage as a percentage of city area, and key COVID-19 metrics is important for improving preparedness for future infectious disease outbreaks. While some questions at the nexus of public health and urban design have received much attention—such as how the COVID-19 pandemic impacted park visitation trends [22–24]—other questions have received comparatively less attention. This study presents the first analysis of how urban parks were used to support the pandemic response efforts, as well as the relationship between urban park acreage and mortality during the pandemic. More specifically, it utilized data previously collected by the Trust for Public Land to explore the alternate functions supported by urban parks during the pandemic response and investigated the associations between total parkland acreage as a percentage of a city’s area and local-level excess mortality in 2020 while controlling for other key variables that could act as effect modifiers or confounders. These results provide insights into how certain urban design features could improve resilience against infectious disease threats, as well as how urban parks could be proactively designed to better support response efforts to future epidemics and pandemics.

A negative, statistically significantly association existed between urban park acreage and 2020 excess mortality rates in the univariable regression model. However, this association was no longer statistically significant when accounting for demographic and socioeconomic considerations that may also influence mortality. These results differ from those produced by similar analyses at the county level [27, 28]. Results from the full model suggest that socioeconomic considerations—such as income, income inequality and education—may have had the greatest influence on excess mortality. In combination, these results indicate that as total park acreage increased, excess mortality rates decreased but that total park acreage was relatively less important when compared to other considerations. To the best of the author’s knowledge, this is the first time that associations between park acreage and pandemic-related mortality outcomes have been investigated in cities in the USA. Future research should seek to explore why these results differed from those produced by previous analyses and to validate these results by investigating the relationship between these variables in other contexts (e.g. smaller cities and/or other countries) and for other outcomes and metrics.

Results also showed that parks in many cities were closed, at least temporarily, during the first year of the pandemic. This was in part due to the novel nature of COVID-19 and a rapidly evolving understanding of the SARS-Cov-2 virus and its transmission between humans. Some features of parks rendered them the focus of much policy attention at the beginning of the pandemic because policymakers worried that these spaces could serve as transmission hotspots during the outbreak [24, 47, 48]. Still, during a pandemic, a dearth of open public spaces can create differential experiences for those residing in urban environments who lack access to private green spaces [47, 49, 50]. Recognizing this, and acknowledging the health benefits afforded by parks, others have advocated for keeping urban parks open during public health emergencies as a means of ensuring the physical and mental health of individuals [22, 47].

This recommendation seems especially pertinent when also considering the alternate functions urban parks supported during the pandemic response and how they promoted urban resilience. To date, much of the discussion surrounding urban resilience has focused on climate change, but the COVID-19 pandemic has made it clear that this conceptualization must expand to be more encompassing of other topics, including public health [14]. Within the resilience literature, there are several conceptualizations of resilience; these include absorptive capacity—the ability of a system to remain stable and continue functioning under stress—and adaptive capacity—the ability of a system to respond and change under stress [51–53]. When considering the alternate functions supported by urban parks, they promoted absorptive resilience capacities by supporting housing, meal distribution and childcare and education services, as well as adaptive resilience capacities by directly supporting the pandemic response through serving as venues for diagnostic testing, vaccination campaigns and PPE distribution.

Notably, these alternate functions were not confined to cities of a certain population size or region, indicating that urban parks were successfully used to support resilience and pandemic response in diverse urban contexts across the USA. The value of these services should not be discounted. One reason that the COVID-19 pandemic was so deadly is that many communities were disconnected from health care and government services [54]. This may be particularly true in cities that bore a large burden of disease, such as New York City (NY) or El Paso (TX). While these cities were excluded from this study due to methodological considerations, they may represent important contexts for conducting case studies and studying aspects related to the pandemic response.

That urban parks were used to address these vulnerabilities and offer services could also hold significance for urban planning, health policy and health equity in the future. Others have noted that in the wake of disasters, there is a unique opportunity to solve structural problems and prevent future suffering by working to increase resilience [48, 50, 55]. Renowned urban designer Galen Cranz has discussed how urban parks in the USA have generally been designed to solve specific problems—such as promoting mental health and recreation or helping cities become more ecologically sustainable—and predicts that the management of large infectious disease outbreaks could influence the shape and use of urban parks in the future [17]. Cities that adopt a more proactive posture toward reconceptualizing their public spaces will likely be more resilient to future pandemics [56].

Parks have a potential role to play in these efforts. In other contexts, urban parks can fulfill multiple functions—not all focused on leisure and recreation [50]. Perhaps the most notable example of this can be seen in Japan. In response to recurring earthquake disasters, a formal system was established for the planning and construction of disaster refuge parks, which are urban parks that are intentionally designed to aid in the response to disasters and other emergencies [57]. These disaster parks could serve as a model for intentionally designing parks and other green infrastructure to provide routine functions and services while also supporting the response to epidemics and pandemics (e.g. by supporting meal distribution services, diagnostic testing, etc.). Of importance, however, is that these parks are designed to support and provide these services in times of need, and they are not merely used in an extemporaneous fashion, which could compromise the quality of services offered. For instance, people experiencing homelessness deserve to be treated with dignity and respect, but ad hoc shelters stood up in urban parks during the responses to infectious disease outbreaks that are not designed or operated in a way that supports safe habitation could result in undignified housing for these individuals and pose additional risks to public health and safety.

There are also important implications for equity. It is widely accepted that ethnic minorities and people of lower socioeconomic status in cities are often disadvantaged in accessing high-quality parks and parks more broadly [58–60]; this same population has borne a disproportionately large burden of morbidity and mortality during the COVID-19 pandemic [38–41, 43, 61]. Accordingly, if urban parks are to be reimagined as a form of infrastructure that supports resilience against infectious disease outbreaks, their distribution throughout a city and standards of management and maintenance also represent critical considerations for ensuring that outbreak response services are provided in an equitable fashion. Indeed, there have already been calls for continued work to expand public park infrastructure and increase the accessibility of existing parks so that the multifaceted benefits afforded by these spaces are more equitable [23, 62].

Limitations

This analysis has several limitations that should be noted. First, inferences from the regression analysis may be constrained by variations in temporal and spatial local-level measurement scales across the various data sources. For example, some local-level metrics were reported at the city level, while others were at the county level. Such inconsistencies are a recognized challenge for city-level analyses [63], but conversion to a standardized geographical unit encompassing an identical sample population was not possible given the different data sources and reporting scales. Additionally, as noted previously, two cities [Indianapolis (IN) and Las Vegas (NV)] were omitted from the analysis because of incomplete data. These considerations, therefore, evidence a clear need for data harmonization efforts to provide more consistent and complete data that would facilitate city-level analyses across a variety of geographies and contexts.

A second limitation relates to the format of the City Park Facts Survey used for the descriptive analysis. This survey asked respondents to use text to report what alternate purposes park facilities served during the COVID-19 pandemic. While the free-response question format provided survey respondents with greater flexibility and was particularly well suited for exploratory research, this portion of the survey did not present respondents with a predetermined set of alternate functions and asked survey respondents to affirm that parks did not serve a purpose. Because of this, the results reported in this analysis should not be interpreted as indicating that parks did not serve a particular function, but rather that parks were reported to have been used in these ways. As a result, the results from the descriptive analysis may represent an underestimate of the ways in which urban parks supported the pandemic response because respondents may have erroneously omitted alternate functions that were supported by parks.

Finally, this study adhered to a cross-sectional study design, and the results only offer insights for the first year of the pandemic (i.e. 2020). The results from both the descriptive and regression analyses are likely to differ when considering different time frames. For instance, when the survey was circulated, COVID-19 vaccines were not yet widely available; in reality, many more cities may have used parks to support vaccination campaigns in the following months and years. Similar to the previous limitation, this may have resulted in an underestimate of the true magnitude in which parks supported the COVID-19 response.

CONCLUSIONS

This study described how urban parks were used to support the COVID-19 pandemic response and explored the relationship between total urban park acreage and excess mortality rates in large American cities.

Results demonstrated the potential of urban parks to support pandemic response activities in times of emergency while also supporting leisure and recreation in peacetime. More specifically, cities of different population sizes and in different regions of the country used parks to directly support the pandemic response by repurposing them as venues for diagnostic testing, vaccination campaigns and PPE distribution. This type of adaptability validates assertions that urban parks are integral to urban resilience, which demands that venues and infrastructure in cities can be repurposed, stretched and modified during times of stress [64].

Results also suggest that urban park acreage was not meaningfully associated with lower levels of excess mortality during the first year of the pandemic. Univariable regression modeling showed that higher total urban park acreage was associated with lower excess mortality rates, but this relationship weakened and was no longer statistically significant when accounting for other demographic and socioeconomic variables that may have confounded or modified the relationship.

Looking ahead, much work remains regarding our understanding of the value of urban green infrastructure for pandemic response and resilience. Although this study found that total park acreage in American cities was not meaningfully associated with excess mortality, future work may wish to investigate this relationship in different contexts, such as cities in different countries or with smaller populations, or using different metrics, such as measures of equity. Indeed, some have already hypothesized that increasing the accessibility and equitability of green infrastructure could represent one method for improving pandemic preparedness and response [23]. Other work may also wish to investigate how certain park features (e.g. park quality, public perceptions, physical complexity of the built environment, etc.) impacted the utilization of pandemic response services as well as the quality of the alternate services provided in parks during the pandemic. This work would be especially informative should the COVID-19 pandemic change the design and planning of urban public spaces, as has been suggested by some scholars [17, 64]. Answering these questions could improve our understanding of the true value added by urban green infrastructure and optimize the design of these spaces to maximize societal benefits.

SUPPLEMENTARY DATA

Supplementary data are available at Oxford Open Infrastructure and Health online.

STUDY FUNDING

This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

CONFLICT OF INTEREST

I confirm that this work is original, has never been published and is not under review elsewhere. Additionally, I do not have any conflict of interest or relationship, financial or otherwise, to disclose.

AUTHORS’ CONTRIBUTIONS

M.B.: Conceptualization, data curation, formal analysis, investigation, methodology, validation, writing—original draft and writing—review & editing.

DATA AVAILABILITY

Most of the data underlying this article are available in the article and in its online supplementary material. However, the data underlying the descriptive analysis contained in this manuscript were provided by the Trust for Public Land and are not publicly available. Access to these data must be requested from the Trust for Public Land because the author does not maintain the authority to share them.

ACKNOWLEDGEMENTS

Thanks to Jenn Dunn, Dominic Regester, Isabelle Weber (Salzburg Global Seminar) and Lydia Gaby (World Urban Parks) for their work organizing and supporting the Salzburg Global Seminar’s Emerging Urban Leaders Program, which helped support the completion of this project. Many thanks to Helen Beck (University of Washington) for her thoughtful and continuous engagement on this project and her comments on early drafts of the manuscript. Thanks also to Tess Stevens and Hailey Robertson (Center for Global Health Science & Security) for their reviews and comments on early drafts of this manuscript. Finally, thanks to Will Klein (Trust for Public Land) for his assistance in sourcing the data used in this research and for his willingness to answer my many questions about it.

References