Navigation – Plan du site

AccueilRubriquesEspace, Société, Territoire2021COVID-19 Death Rates and County S...

2021
965

COVID-19 Death Rates and County Subdivision Level Contextual Characteristics: A Connecticut Case Study

Taux de mortalité COVID-19 et caractéristiques contextuelles au niveau de la subdivision du comté: une étude de cas du Connecticut
Tasas de mortalidad de COVID-19 y características particulares a escala de condado: el estudio de caso de Connecticut
Yunliang Meng

Résumés

Le 15 juillet 2020, au moins 3 483 832 cas confirmés de COVID-19 et 136 938 décès ont été signalés respectivement aux États-Unis, posant des problèmes socio-économiques et sanitaires sans précédent pour le pays. Les preuves empiriques existantes examinant l'association spatiale entre les facteurs contextuels et les taux de mortalité liés au COVID-19 restent cependant rares et ambiguës. L'objectif de cette recherche est d'examiner la relation spatiale entre les taux de mortalité COVID-19 et les caractéristiques des comtés de l’Etat du Connecticut aux États-Unis. L'analyse montre que les variables explicatives telles que le revenu, la race, l'âge, ou le type de logement, ainsi que les indicateurs sanitaires sous-jacents, sont associés aux taux de mortalité liés au COVID-19 dans l'Etat. Plus encore, l'association entre les taux de mortalité COVID-19 et les variables explicatives de notre analyse varie considérablement dans l'espace, mettant en évidence la nécessité de programmes de prévention et d'intervention spécifiques au contexte local face au COVID-19.

Haut de page

Texte intégral

1. Introduction

1Coronavirus disease 2019 (or COVID-19), an infectious disease caused by the SARS-CoV-2 virus, was first identified by doctors in Wuhan, China at the end of 2019 before an outbreak of the disease was officially declared by the Chinese government in January 2020 (WHO, 2020a). The World Health Organization (WHO) officially declared a COVID-19 pandemic on March 11th, 2020 (WHO, 2020a). COVID-19 became a global health concern due to the rapid spread of the disease and the lack of vaccines and therapies (WHO, 2020b). As of July 15th, 2020, at least 136,938 deaths and 3,483,832 confirmed cases have been reported in the U.S.A. (CDC, 2020a), while more than 574,464 deaths and nearly 13,150,645 cases have been confirmed in more than 188 countries and territories around the world (WHO, 2020c). These numbers are continuously rising every day across the world. The United Nations has characterized the COVID-19 pandemic as a social, human, and economic crisis unlike any in its 75-year history (United Nations, 2020). Although the COVID-19 pandemic affects all segments of the population in almost every country in the world, early research evidence indicates that the most severe health impact – death caused by the virus is being borne disproportionately by disadvantaged people, senior people, and people with underlying medical conditions. For example, Latino and African American residents of the United States have been three times as likely to become infected as their white counterparts and death rates among Black and Hispanic people are much higher than for White people according to the CDC data (CDC, 2020b). In addition, Covid-19 death rates are higher in poorer and more urbanized counties in the U.S. (Finch and Finch, 2020). According to the early research, COVID-19 impacts different age groups differently (CDC, 2020c). 8 out of 10 COVID-19-related deaths reported in the United States have been among adults aged 65 years and older (CDC, 2020c). Among COVID-19 cases, the two most common underlying health conditions were cardiovascular disease (32%) and diabetes (30%) (Stokes et al., 2020). Deaths were 12 times higher among those with reported underlying conditions compared with those with none reported (Stokes et al., 2020). According to Kuderer et al.’s (2020) findings, cancer patients who developed COVID-19 were much more likely to die within a month than people without cancer who were infected with the disease. In their research, 13% of cancer patients who contracted COVID-19 were dead within a month.

2Spatial models are important tools to investigate the statistical relationship between explanatory variables and disease rates (Hipp and Chalise, 2015; Mollalo et al., 2020). However, a very limited number of spatial modeling case studies on COVID-19 have been conducted since the declaration of the pandemic in March 2020. Xiong et al. (2020) used spatial autocorrelation and Spearman’s rank correlation methods to examine the spatio-temporal patterns and influencing factors of the COVID-19 epidemic at multiple scales. Lakhani (2020) utilized GIS mapping techniques to identify COVID-19 health care priority locations pertaining to vulnerable populations, including elderly, palliative, and disabled patients in Melbourne, Australia. Mollalo et al. (2020), applied spatial regression models to explore the spatial association between COVID-19 case rates and explanatory variables (i.e., income inequality, median household income, the proportion of black females, the proportion of nurse practitioners etc.) at the county level in the U.S. Little research examined the difference between using the global and local regression models in modeling variances in COVID-19 death rates. Additionally, the early research on COVID-19 was focused on the case rate, not the death rate and it was conducted at a coarse scale (i.e. at a county or country level) due to the lack of COVID-19 data. To fill the gap, this study uses global and local regression models (i.e. OLS and Geographically Weighted Regression) to explain the spatial variations in COVID-19 death rates at the county subdivision level in Connecticut based on several demographic, socioeconomic, and health related explanatory variables. It should be noted that county subdivisions are better known as cities, towns, or municipalities. The county subdivision level is chosen in this study not only because it is a finer local scale, but because it is designed to have visible, permanent, easily described, and less arbitrary boundaries (Census Bureau, 2020).

3This article is organized as follows. Section 2 illustrates the methodology outlining a description of the study area, variables used in this study, and the local analysis method – Geographically Weighted Regression (GWR). Following this, Section 3 discusses the GWR results which illustrate the relationship between spatial variations in COVID-19 death rates and contextual variables in Connecticut. The final section presents concluding remarks.

2. Methodology

2.1 Study Area

4Connecticut is a state located in Northeastern of U.S.A. It is made up of a mix of 169 small cities and incorporated towns situated in rural areas (Figure 1). It is surrounded by Rhode Island to the east, Massachusetts to the north, New York to the west, and Long Island Sound to the south (Figure 1). Connecticut is a part of New England, although the southwest part of it is often grouped with New York and New Jersey as the tri-state area.

Figure 1. Study Area

Figure 1. Study Area

5Connecticut is slightly larger than the country of Montenegro with a land area of 12,559 km2 and a water area of 1,809 km2. The capital of Connecticut is Hartford and its most populous city is Bridgeport. These cities have populations of 122,591 and 144,900 respectively according to 2017 American Community Survey. Connecticut is the third smallest state by area, the 29th most populous, and the fourth most densely populated of the 50 states. However, Connecticut’s rural areas and small towns, located in the northeast and northwest corners of the state, contrast sharply with its industrial cities such as Stamford, Bridgeport, and New Haven, located along the coastal highways I-95 from the New York border to New London. The Interstate highways in the state are I-95 traveling southwest to northeast along the coast, I-84 traveling southwest to northeast in the center of the state, I-91 traveling north to south in the center of the state, and I-395 traveling north to south near the eastern border of the state (see Figure 1). Connecticut’s per capita personal income in 2019 was estimated at $79,087, the highest of any state in the U.S.A. (Bureau of Economic Analysis, 2020) However, there is a huge income gap throughout the state. After New York, Connecticut had the second largest gap nationwide between the average income of the top 1% and the average income of the bottom 99% (Sommeiller and Price, 2014).

6Connecticut is an important study area for contextual analysis of COVID-19 death rates due to the following reasons: 1) As of July 15th, 2020, Connecticut’s COVID-19 death rate is about 122.8 per 100,000 people, placing the state among the top 3 states in the U.S.A. in terms of COVID-19 death rates (CDC, 2020d); 2) Connecticut is densely populated and has a diverse population with a history of income polarization. These social and environmental settings might facilitate the spread of COVID-19; 3) There has been no scholarly research conducted in Connecticut on COVID-19 spatial patterns and their relationship to county subdivision level contextual characteristics.

2.2 Data Preparation

7This study is based on COVID-19 death data collected and managed by the Connecticut Department of Public Health or CDPH (https://portal.ct.gov/​Coronavirus/​COVID-19-Data-Tracker). The data consisted of 4,380 COVID-19 deaths reported by the 169 county subdivisions in Connecticut from the beginning of the pandemic to July 15th, 2020. The date of July 15th, 2020 was selected as the cut-off date for compiling Connecticut’s COVID-19 death data not only because of the data availability, but also because the date falls about 4 months after the first COVID-19 death was reported in Connecticut. In addition, July 15th, 2020 is the day 8 and 4 weeks after the execution of Phase 1 and 2 of Connecticut’s Reopen Plan, which began on May 20th, 2020 and June 17, 2020 respectively. Different from many other states in the U.S., the key metrics – COVID-19 cases and deaths in Connecticut – continued on a consistent downward trend 8 and 4 weeks after Phase 1 and 2 of reopening (see Figure 2 and 3 below). Therefore, July 15th, 2020 is chosen as a cut-off date to review and analyze the first wave of COVID-19 death data in Connecticut.

Figure 2. Daily New COVID-19 Cases in Connecticut, USA

Figure 2. Daily New COVID-19 Cases in Connecticut, USA

Figure 3. Daily New COVID-19 Deaths in Connecticut, USA

Figure 3. Daily New COVID-19 Deaths in Connecticut, USA

8COVID-19 death rates were calculated for each county subdivision in Connecticut as the number of COVID-19 deaths reported to the CDPH per 100,000 people. Thereafter, the COVID-19 death rates table was joined with a shapefile consisting of 169 county subdivisions (see Figure 1 for details) based on the unique 10-digit county subdivision FIPS code assigned by the U.S. Census Bureau using ArcMap 10.6.1 (Esri, 2018) so that the COVID-19 death rates could be later aligned with contextual variables for further analysis. The descriptive statistics for the dependent variable – COVID-19 death rates – are shown in Table 1 below.

Table 1. Descriptive Statistics for the Dependent Variable –COVID-19 Death Rates

Min

Max

Median

Interquartile Range

Standard Deviation

COVID-19 Death Rates: COVID-19 deaths per 100,000 people

0.0

465.0

39.0

140.0

96.9

9There are many factors that affect the health of individuals and the communities in which they live. The health of people can be determined by numerous interrelated factors including genetics, lifestyle, circumstances, environment, etc. (Lawrence, 2011). Some determinants (i.e. lifestyle, environment, and circumstances) can be influenced, while others are more difficult to control (i.e. genetics). Among those factors that have a significant influence on health are the environmental and social conditions in which people live (Lawrence, 2011). This study considers 11 contextual characteristics as the potential explanatory variables. The descriptive statistics for each explanatory variable are shown in Table 2. The 11 contextual variables, covering demographic (i.e. age and population density), socio-economic (i.e. income, poverty, and type of housing) and health aspects (i.e. health insurance coverage and underlying medical conditions), were chosen to reflect the key dimensions underlying the variation in the risk of COVID-19 death as suggested by existing empirical data and/or research (CDC, 2020; Kuderer et al., 2020; Finch and Finch, 2020; Stokes et al., 2020).

Table 2. Descriptive Statistics for the Explanatory Variables

Variables

Min

Max

Median

Interquartile Range

Standard Deviation

Age: % of people aged 65 and plus

5.9

41.4

14.6

5.9

5.4

Income: median household income

33,841.0

219,868.0

85,296.0

29,393.0

28102.9

Poverty: % of people living under the poverty line

1.1

30.1

10.4

13.0

4.4

Race: % Black population

0.1

56.2

1.3

2.3

7.4

Ethnicity: % of Hispanic Population

0.6

43.4

3.4

3.5

7.5

Population density: the number of people per square kilometer

11.4

2946.3

116.4

291.8

468.2

Type of housing: % of multi-family housing

0.0

80.4

16.4

20.1

16.5

Healthcare coverage: % of people without health insurance coverage

1.0

19.1

3.8

2.9

2.8

Underlying medical condition indicator 1 –Cardiovascular disease death rate: Cardiovascular disease deaths per 100,000 people

108.7

704.3

207.9

101.7

80.9

Underlying medical condition indicator 2 – Diabetes death rate: Diabetes deaths per 100,000 people

0.0

183.7

52.9

24.9

22.2

Underlying medical condition indicator 3 – Cancer rate: Cancer cases per 100,000 people

0.0

166.3

92.1

24.3

22.3

10In this research, the variable of health insurance coverage was also included for the following reason. In 2018, there were 27.8 million individuals in the U.S. who did not have some type of health insurance coverage (Berchick et al., 2019). In the U.S., people without health insurance coverage are less likely to visit doctors, have usual sources of care, undergo recommended tests or fill prescriptions, and receive medical care in an ICU. Consequently, they are at a greater risk of death caused by COVID-19.

11The demographic and socio-economic variables such as residential population, age, income, poverty, race, ethnicity, type of housing, and health insurance coverage were taken from the most recent and complete American Community Survey (ACS) five-year estimates (2014-2018) released on the US Census Bureau’s website (data.census.gov). The age variable was calculated as the percentage of population aged 65 and plus. The income variable was determined by the median household income. The poverty variable was measured as the percentage of people living under the poverty line. The race variable was quantified by the percentage of Black people in the residential population. The ethnicity variable was calculated by the percentage of Hispanic people in the residential population. Population density was determined by the number of people per square kilometer. Type of housing was measured as the percentage of multi-family housing (i.e. condo and apartment). The healthcare coverage variable was quantified by the percentage of people without health insurance coverage. Three underlying medical condition indicators – cardiovascular disease and diabetes death rates and cancer case rates – were compiled at the county subdivision level and acquired through the CDPH (https://portal.ct.gov/​DPH). It should be noted that the best indicators to describe people’s underlying conditions in different county subdivisions are cardiovascular disease, diabetes, and cancer case rates. However, no governmental or non-governmental organization collects and/or releases cardiovascular disease and diabetes cases at the county subdivision level in the State of Connecticut. In this study, cardiovascular disease and diabetes death rates are used because they are correlated to cardiovascular disease and diabetes case rates respectively (Gregg et al., 2012; Roth, 2018) and, therefore, can provide approximate estimates for the two underlying medical conditions in different county subdivisions across Connecticut. The cardiovascular disease and diabetes deaths and cancer cases used were collected between 2010 and 2014 by the CDPH. This is the most recent available cardiovascular disease and diabetes death and cancer case data in the state, since CDPH collects such data over a 5-year period and releases it once in every 5 years. The most recent data was released in the year of 2017. Then, the demographic, socioeconomic and health datasets were joined with the Connecticut county subdivisions shapefile and stored for further analysis using ArcMap 10.6.1 (ESRI, 2018).

2.3 GWR Model Building

12Many early contextual studies on the spatial distribution of disease are criticized for their use of global regression modeling techniques, such as Ordinary Least Squares (or OLS) regression (Moore and Carpenter, 1999), because the regression method violates some basic assumptions (e.g., independence of observations and spatial stationarity of the relationship between independent and dependent variables) when spatial data is used. GWR (Brunsdon et al., 1996; Fotheringham et al., 2002) relaxes these assumptions and enables the analysis of spatially clustered data. GWR is often considered as an extension of OLS regression, since it allows local parameters instead of global parameters to be estimated, hence making it possible to model spatial variations within the data (Fotheringham et al., 2002). Unlike OLS regression, which produces a single global model across space, GWR simultaneously conducts multiple regressions for different data units so that there is one regression model per spatial data unit (e.g. a county subdivision boundary) (Hipp and Chalise, 2015). In a GWR model, observations near a particular data unit will have more influence in the estimation than observations distance away (Hipp and Chalise, 2015). Given the weaknesses of OLS regression and strengths of GWR, this research uses GWR for analyzing the spatial non-stationarity relationship between the dependent variable and explanatory variables in the State of Connecticut.

13The first step is to examine the dependent variable – COVID-19 death rates – and explore its spatial heterogeneity. If the dependent variable is not spatially clustered, there is no need to build a spatially explicit model. The Moran’s I Index (Anselin, 1995) provided by ArcMap 10.6.1 (ESRI, 2018) was used to identify the clustering of COVID-19 death rates across county subdivisions in the State of Connecticut. Moran’s I ranges from −1.0, perfectly dispersed (e.g., a checkerboard pattern), to a +1.0, perfectly clustered. In this research, Moran’s I scores (0.425) and p value (0.000) were generated, indicating that COVID-19 death rates are spatially clustered and the results are statistically significant. Local Moran’s I cluster analyses of the dependent variable was conducted, and the results are shown in Figure 4 below.

Figure 4. Local Moran’s I Cluster Analyses of COVID-19 Death Rates in Connecticut

Figure 4. Local Moran’s I Cluster Analyses of COVID-19 Death Rates in Connecticut

14Five different types of spatial clustering are demonstrated in Figure 4: (1) high-high, for county subdivisions with high COVID-19 death rates that are in close proximity to county subdivisions with high COVID-19 death rates; (2) low-low, for county subdivisions with low COVID-19 death rates that are in close proximity to county subdivisions with low rates; (3) high-low (known as spatial outliers), for county subdivisions which have high COVID-19 death rates, but are proximate to county subdivisions with low rates; (4) low-high (also known as spatial outliers), for county subdivisions which have low COVID-19 death rates, yet are in close proximity to county subdivisions with high rates; (5) not significant, for county subdivisions where there is no significant spatial clustering. As shown in Figure 4, the low-low spatial clusters were mainly found in the east and northeast of Connecticut, except for a few low-low spatial clusters were found in county subdivisions located in the west and northwest. The high-high spatial clusters were mainly found in county subdivisions located in and around Hartford and New Haven metropolitan areas in the center and southwest of Connecticut. However, a few high-high spatial clusters found in and around county subdivisions in the southwest end of the state, right next to New York State.

15The OLS multivariate model (Aiken and West, 1991) in the Statistical Package for the Social Sciences (SPSS) 25 was then used to conduct initial data exploration and model specification. Two factors motivated the decision to first specify the OLS model: 1) there was a need to identify explanatory variables significantly correlated with the dependent variable (COVID-19 death rates) before specifying the GWR model; and 2) the GWR software used for spatial analysis does not provide a variance inflation factor (VIF) to measure multicollinearity. If the standard regression equation in the investigation of the dependent variable is given by:

where Yi is the COVID-19 death rate at county subdivision i, β0 is a constant term (i.e., the intercept), βk measures the relationship between the independent variable xk and Y for the set of i county subdivisions, and εi is the error associated with county subdivision i. It should be noted that i Є C = {1,2,…,n} which is the index set of locations of n observations (i.e. all county subdivisions in Connecticut).

16It should be noted that the above equation “results in one parameter estimate for each variable included” (Cahill and Mulligan, 2007). The summary of the OLS analysis results is presented in Table 3.

Table 3. Results from Ordinary Least Square Model of COVID-19 Death Rates at County Subdivision Level in Connecticut

Dependent Variable

Independent Variables

β

SE

p value

VIF

COVID-19 Death Rates

Intercept

-160.441

44.055

< 0.01

-----

Income: median household income

0.001

0.000

< 0.01

1.824

Age: % of people aged 65 and plus

9.866

2.328

< 0.01

2.911

Race: % of Black population

5.598

1.038

< 0.01

1.598

Type of housing: % of multi-family housing

3.371

0.652

< 0.01

3.139

Underlying medical condition indicator 1 – Cardiovascular disease death rate: Cardiovascular disease deaths per 100,000 people

0.277

0.079

< 0.01

1.123

17In the OLS regression, only variables significantly correlated with the dependent variable – COVID-19 death rates – were included. The OLS model is statistically significant (F = 18.4, p < 0.01). The adjusted R2 value is 0.361, meaning that the OLS model explained 36.1% of the variance in county subdivision level COVID-19 death rates in Connecticut. The VIF values for all variables were less than 5.0, a commonly used cutoff point (Becker et al., 2014; Ringle et al., 2015), suggesting no severe multicollinearity issue was detected among the explanatory variables. In fact, most VIF values are less than 3.0, indicating that the correlations among explanatory variables are low. As shown in Table 3, there is a positive and significant relationship between COVID-19 death rates and explanatory variables including income, age, race, type of housing, and underlying medical condition indicator 1 – Cardiovascular disease death rate. In other words, the higher the median household income, the percent of people aged 65 and plus, the percent of black population, the percent of multi-family housing, or the cardiovascular disease death rate in a county subdivision; the higher the COVID-19 death rate in that county subdivision. The rest of the explanatory variables are not significantly related to the dependent variable in this study. The residuals of the OLS model were spatially auto-correlated (Moran’s I = 0.156, p < 0.05), indicating the OLS model overestimated COVID-19 death rates for some county subdivisions, while it underestimated the outcomes for some others.

18The same set of variables was then used to specify a GWR model using the GWR4 software (https://gwrtools.github.io/​). GWR is a modeling technique used to explore spatial non-stationarity (Brunsdon et al., 1996). The “main characteristic of GWR is that it allows regression coefficients to vary across space, and so the values of the parameters can vary between locations” (Mateu, 2010, p. 453). In other words, instead of estimating a single parameter for each variable, GWR estimates local parameters. By estimating a parameter for each data location (i.e. a county subdivision) in the State of Connecticut, the GWR equation would only alter the OLS equation as follows:

where Yi is the COVID-19 death rate at county subdivision i, β0i is the constant term at county subdivision i, xki is the explanatory variable at county subdivision i, βki is the value of the parameter for the corresponding explanatory variable at county subdivision i, and εi is the error term at county subdivision i. It should be noted that i Є C = {1,2,…,n} which is the index set of locations of n observations (i.e. all county subdivisions in the State of Connecticut).

19GWR becomes useful when “a single global model cannot explain the relationship between some sets of variables” (Brunsdon et al., 1996, p. 281). In the GWR model, a continuous surface of parameter values is estimated under the assumption that locations closer to i will have more influence on the estimation of the parameter β̂i for that location. Consequently, GWR allows researchers to explore “spatial non-stationarity by calibrating a multiple regression model which allows different relationships to exist at different geographical locations” (Leung et al., 2000). The GWR model was used to explore the macro-level spatial non-stationarity of the statistical relationship between COVID-19 death rates and the explanatory variables such as age, population density, income, race/ethnicity, etc.

20While conducting GWR, the adaptive kernel was used, which was produced using the bi-square weighting function. The adaptive kernel uses varying spatial areas, but a fixed number of observations for each estimation. It is the most appropriate technique when the distribution of observations varies across space. In this case, county subdivisions are smaller and closer together in the center of the state than they are at the edge. Finally, a process that minimizes the Akaike Information Criteria (AIC) was used to determine the best kernel bandwidth. The parameter estimates and t values produced by the software were exported and mapped using ArcMap 10.6.1 (ESRI, 2018).

3. Results and Discussion

21A local Moran’s I cluster analysis (Anselin, 1995) was conducted for the residuals of the GWR model as a diagnostic for the collinearity in GWR residuals. There were no violations of residual independence. The GWR model of COVID-19 death rates generated β coefficients (see Table 4 and Figure 5), t values (see Figure 6), and R2 values (see Figure 5F) for each county subdivisions. The direction of the relationships among the dependent variable and the explanatory variables (see Table 4, Figure 5) was inconsistent in some county subdivisions included in the study. In addition, the COVID-19 death rates at the county subdivision level were significantly clustered (Moran’s I value: 0.425 and p value: 0.000). This clustering and the inconsistent direction of the relationships suggest that local contextual variables are associated with COVID-19 death rates and that the amplitude of such contexts varies across the State of Connecticut.

Table 4. Results from GWR Model of COVID-19 Death Rates in Connecticut

Dependent Variable

Independent Variables

β coefficients

Percentage of County Subdivisions by 95% of t Statistics

Min

Max

t≤ 1.96

1.96 < t≤ 2.58

t> 2.58

COVID-19 Death Rates

Intercept

-612.699

-102.288

-----

-----

-----

Income: median household income

0.001

0.004

32.0%

23.0%

45.0%

Age: % of people aged 65 and plus

-0.308

21.058

42.0%

18.9%

39.1%

Race: % of Black population

1.666

7.364

52.7%

3.5%

43.8%

Type of housing: % of multi-family housing

0.902

8.368

18.3%

4.8%

76.9%

Underlying condition indicator 1 – Cardiovascular disease death rate: Cardiovascular disease deaths per 100,000 people

-0.066

1.073

37.3%

13.6%

49.1%

* 1.96 and 2.58 are the cut-off values for t-test. When │t│ > 1.96, the β coefficient estimate for a variable is significant at a significance level of 0.05.When │t│ > 2.58, the β coefficient estimate for a variable is significant at a significance level of 0.01.

Figure 5. Spatial Variations in β Coefficient Estimates in the State of Connecticut for Explanatory Variables: Income (Map A), Age (Map B), Race (Maps C), Type of Housing (Map D), Underlying Condition Indicator 1 – Cardiovascular Disease Death Rate (Map E), and Local R2 Value (Map F) from the GWR model of COVID-19 Death Rates.

Figure 5. Spatial Variations in β Coefficient Estimates in the State of Connecticut for Explanatory Variables: Income (Map A), Age (Map B), Race (Maps C), Type of Housing (Map D), Underlying Condition Indicator 1 – Cardiovascular Disease Death Rate (Map E), and Local R2 Value (Map F) from the GWR model of COVID-19 Death Rates.

Figure 6. Spatial Variations in t Values for Explanatory Variables in Modeling COVID-19 Death Rates in Connecticut: Income (Map A), Age (Map B), Race (Maps C), Type of Housing (Map D), and Cardiovascular Disease Death Rate (Map E). Note: 1.96 and 2.58 are the cut-off values for t-test. When │t│ > 1.96, the β coefficient estimate for a variable is significant at a significance level of 0.05. When │t│ > 2.58, the β coefficient estimate for a variable is significant at a significance level of 0.01.

Figure 6. Spatial Variations in t Values for Explanatory Variables in Modeling COVID-19 Death Rates in Connecticut: Income (Map A), Age (Map B), Race (Maps C), Type of Housing (Map D), and Cardiovascular Disease Death Rate (Map E). Note: 1.96 and 2.58 are the cut-off values for t-test. When │t│ > 1.96, the β coefficient estimate for a variable is significant at a significance level of 0.05. When │t│ > 2.58, the β coefficient estimate for a variable is significant at a significance level of 0.01.

22As shown in Table 4 and Figure 5A, income, defined by median household income, is positively associated with the county subdivision level COVID-19 death rates across Connecticut. This finding is inconsistent with previous research which suggests that the high death rates due to Covid-19 are associated with poorer population (Finch and Finch, 2020). The inconsistency in this finding is caused by the spread of COVID-19 cases and death into southwestern Connecticut, also known as Gold Coast. The region is the most affluent area in the state, but also the most affected by COVID-19 pandemic in Connecticut (see high-high cluster in Figure 4). The southwestern Connecticut area is about 80 kilometers northeast of New York City which was one of the epi-centers during the first wave of COVID-19 pandemic in the U.S. About 43,000 people commuted daily to jobs in New York City from this area (Braun and Westin, 2020), because they can enjoy relatively low taxes and spacious homes in Connecticut, while having convenient access to the Metro-North commuter line and Interstate highway 95 (see Figure 1) connecting the area with New York City. Additionally, between March and June, 2020, approximately 10,000 people reported a mailing address change from New York City to Connecticut to U.S. Postal Services – an eight-fold increase from 2019 (Freedman, 2020), indicating a large population movement from New York City to Connecticut. The unique geographic location of the Gold Coast region, the frequent cross-state movement of people brought by the convenient transportation from and to New York City, and the large outbound migration from the city to Connecticut during the peak of the COVID-19 pandemic, together led to the positive spatial correlation between income and COVID-19 death rates across Connecticut.

23As illustrated in Table 4 and Figure 5B, age, defined by the percent of people aged 65 and plus, is largely positively associated with the county subdivision level COVID-19 death rates across Connecticut. This finding supports previous research which shows that senior residents are more likely to have a higher mortality rate due to their weaker immune systems (Liu et al., 2020). The fact that senior residents are at a greater risk of COVID-19 infection and death provides an important indication for those who are in close contact with people who have COVID-19, including family members and health care workers who care for people who are infected. If people are 65 and older and live in an area where a great number of COVID-19 cases and deaths have been reported, they must take precautions to reduce their exposure, such as stopping or restricting visits from children, friends, or caregivers and/or limiting or avoiding routine trips to their doctors’ offices.

24As demonstrated in Table 4 and Figure 5C, race, defined by the percent of Black people, is positively associated with the county subdivision level COVID-19 death rates across Connecticut. This finding is consistent with previous research which suggests racial minorities (i.e. Black people) are more likely to be infected by COVID-19 and have a higher mortality rate. As suggested by the CDC, the race gaps in COVID-19 death rates are likely caused by long-standing systemic health and social inequities in the U.S. Racial and ethnic minority groups in the U.S. are at increased risk of getting COVID-19 and experiencing severe illness or death, regardless of age (CDC, 2020b). Among Black people, evidence points to higher rates of hospitalization or death from COVID-19 than among White people, due to a higher prevalence of underlying conditions such as hypertension, diabetes, and obesity (Hamidi et al., 2020). They also may have less recourse to workplace policies that enable social distancing (Hamidi et al., 2020).

25As shown in Table 4 and Figure 5D, type of housing, defined by the percent of multi-family housing, is positively associated with the county subdivision level COVID-19 death rates across the state. This finding supports previous research which shows that a higher COVID-19 death rate is likely associated with more urbanized areas in the U.S.A. (Finch and Finch, 2020). As illustrated in Figures 5D and 6D, the coefficient for the type of housing explanatory variable is highest and most significant in and around the city of Hartford. According to the New York Times, Hartford is among the top 10 cities with the highest population density growth from 2010 to 2016 in the U.S. (Kolko, 2017). It is also the capital city and second largest city in the State of Connecticut. In addition, 80.4% of all Hartford housing units are multi-family housing units, the highest percentage in the State of Connecticut. So far, the most effective way to slow the spread of and death caused by the highly contagious disease – COVID-19 – is to minimize human-to- human contact through different measures of social distancing. Social distancing can be implemented by governments at the local and state levels through one or a combination of interventions, including the closure of schools, bars, restaurants, and any social or sporting events; sick leave; work-from-home policies; splitting shifts to reduce workplace interactions; sheltering in place; and travel/trade bans (Chu et al., 2017). However, multi-family housing residents may have challenges with social distancing to prevent the spread of COVID-19 because they often gather closely for social and recreational activities, shared dining, laundry facilities, and stairwells or elevators. To effectively reduce the transmission risk of COVID-19 in multi-family housing located in and around the city of Hartford, some targeted strategies are needed, including but not limited to limiting the number of people in common areas, installing touch-free door openers, disinfecting common areas regularly, restricting visitors, canceling scheduled events, and requiring the use of Personal Protective Equipment (or PPE) inside of the properties.

26As demonstrated in Table 4 and Figure 5E, the underlying medical condition indicator, cardiovascular disease death rate, is mostly positively associated with the county subdivision level COVID-19 death rates across the state. This finding is consistent with previous research which suggests the COVID-19 death rate is higher among populations with underlying conditions (Stokes et al., 2020). In 2018, 30.3 million U.S. adults (12.1%) were diagnosed with cardiovascular disease. Cardiovascular disease is not only the leading cause of death for people in the United States – one person dies every 37 seconds in the United States because of the disease (CDC, 2020e) – but also a reported underlying condition that causes a much higher death rate for COVID-19 patients (Stokes et al., 2020). During the pandemic, people with cardiovascular disease are at risk for severe illness or death as a result of both COVID-19 and lack of adequate cardiovascular disease monitoring or diagnosing due to the shutdown of clinics or reduced routine doctor appointments.

27The local Moran I and GWR results are potentially useful for the CDPH in targeting priority county subdivisions for COVID-19 prevention and intervention, since preventive strategies should be informed by an understanding of spatial distribution of COVID-19 death rates (see Figure 4) and contextual factors (see Figure 5 and 6). In particular, these factors should be examined locally and different strategies aimed at reducing COVID-19 death rates should be applied in different county subdivisions in Connecticut. For example, a strategy designed to help Black people practice social distancing may be sufficient to reduce COVID-19 death rates in county subdivisions located in northern Connecticut, since the variable of race has a greater effect in the area (Figure 6C). However, the same approach is unlikely to be effective in the county subdivisions located in southwestern Connecticut. In those county subdivisions, COVID-19 prevention/intervention programs aimed at helping, for example, senior residents or residents in multi-family housing practice social distance should be considered as tools for reducing COVID-19 death rates, since the variables of age and type of housing have a greater effect in the area (Figure 6B and 6D).

28As shown in Table 4 and Figure 5 - 6, the change in magnitude and direction of the coefficients suggests spatial non-stationarity of the relationship between the dependent variable (COVID-19 death rates) and the explanatory variables. The variation in parameter estimates from GWR suggests the necessity of applying this spatial statistical tool to future COVID-19 modeling studies that would be restricted by using global OLS models, since GWR provides insights on how a particular explanatory variable influences the COVID-19 death rates across the study area. The importance of using spatial statistical tools such as GWR in the future study of COVID-19 death rates can also be confirmed by the local R2 value (see Figure 5F). The adjusted R2 for the GWR model of COVID-19 death rates ranged from 0.418 to 0.742, with an average of 0.617, while the adjusted R2 in the OLS model was 0.361. The OLS R2 of 0.361 masks a wide distribution of local associations between the explanatory variables and COVID-19 death rates. In other words, without GWR, it would be impossible to estimate the variance of local associations. For example, as shown in Figure 5F, in the northwest regions of Connecticut the GWR model explained up to 74.2% of the variance in COVID-19 death rates. However, among county subdivisions clustered in the east of the state, the model only explains about 41.8% to 50.0% of the variance, a spatial variation that would have been neglected with the OLS model alone.

29This study is not without limitations. The first group of limitations is associated with geographic boundary effects, such as the Modifiable Areal Unit Problem (MAUP) and edge effect. It should be noted that the statistical relationships drawn from areal data must be carefully interpreted. Robinson (1950) long ago suggested the data scale/boundary problem and clearly explained that inferring individual level relationships from macro-level correlations is inappropriate. In this study, county subdivision boundaries in Connecticut were used, so the relationships between COVID-19 death rates and contextual characteristics at the county subdivision level cannot be interpreted as and/or applied to individual level relationships. Additionally, the GWR model is restricted by the edge effect, whereby county subdivisions located on the edges of Connecticut do not have the 360° influence of county subdivisions in the state’s interior.

30The second group of limitations is related to COVID-19 and explanatory variable data. It should be noted that the dataset prepared by the CDPH may underestimate the COVID-19 death rates to some extent due to the following reasons: 1) the reporting data can lag by a few days to one to two weeks; 2) in February and early March, many people could not take a COVID-19 test when they felt they had been infected by the virus because of the test kit shortage (Baird, 2020), so deaths due to Covid-19 may have been misclassified as pneumonia deaths in the absence of positive test results. Therefore, some COVID-19 deaths are missing from the reported data prepared by the CDPH. Furthermore, the COVID-19 death data in this research was collected on or before July 15th, 2020 and the pandemic will not be over anytime soon, so the analysis results need to be interpreted and applied with caution. For example, in this study, affluent areas in Connecticut are associated with higher COVID-19 death rates. This result might have changed after more COVID-19 death data collected after July 15, 2020 was added into the analysis. For example, 166 out of 169 county subdivisions in Connecticut were classified as red alert zones by CDPH on January 14th, 2021, since their average daily infection rate was higher than 15 cases per 100,000 people (Connecticut COVID-19 Response, 2021). The result implies that the COVID-19 deaths might be more widely and evenly distributed across the State of Connecticut after the second wave of COVID-19 hits the State. Therefore, it is worthwhile to conduct the GWR analysis again after the second wave COVID-19 death data are collected and released. In addition, a time alignment problem is encountered. COVID-19 death data was collected in the year of 2020. However, the most recent and complete demographic and socio-economic data is based on American Community Survey data collected between the years of 2014 and 2018. The most recent and complete underlying condition indicator data – cancer case rate and cardiovascular disease and diabetes death rates – were collected by CDPH between the years of 2010 and 2014. Moreover, the local R2 values accounted for 41.8% to 74.2% of the COVID-19 death rates, which means that other risk factors associated with the COVID-19 death rates need to be added into the GWR models. For example, masks are not 100% effective, but mask wearing does decrease the risk of COVID-19 viral spread and lower death rate as a result. According to Chu et al.’s research (2020) supported by WHO, the chance of infection or transmission was 3% with a mask compared with 17% without a mask. Although wearing masks is required for any person in a public place in Connecticut effective at 8:00 p.m. on Monday, April 20, 2020, no organizations or government agencies have collected and/or released data regarding how many people actually wore a mask in public places before and/or after April 20, 2020. The spatial distribution of available ICU beds and doctors is another important variable that needs to be considered, since research shows that patients admitted to hospitals with fewer ICU beds had a higher risk of death (Gupta et al., 2020). However, there is no organization or government agency in Connecticut that has collected and/or released such data.

4. Conclusions

31This research analyzed the spatial distribution and correlations of COVID-19 death rates in Connecticut, U.S.A. Specifically, this study incorporated demographic, socio-economic and health related contextual variables with the COVID-19 death rates. The spatial relationships between the COVID-19 death rates and explanatory variables are still under investigation, and little research has been done to investigate the spatial heterogeneity of the relationships. This study fills the gap by illustrating that 1) there is a significant spatial correlation between the COVID-19 death rates and the explanatory contextual variables (i.e. income, age, race, type of housing and an underlying medical condition indicator – cardiovascular disease death rate) and 2) this relationship has a spatial but nonstationary association which highlights the need for local and context-specific COVID-19 prevention and intervention programs. In other words, by using GWR, health researchers and practitioners in Connecticut can obtain an understanding of COVID-19 death rates across the state and respond to the notion that “all health is local” (Gebreab and Diez-Roux, 2012). For example, generally speaking, the COVID-19 death rates are higher among communities where the income is lower in the U.S. (Wadhera et al., 2020). However, the death rates are largely positively correlated with income across the county subdivision in Connecticut. In other words, the COVID-19 death rates in the first wave of the pandemic are more likely to be higher in wealthy county subdivisions in the State of Connecticut, which contradicts previous reports. In addition, the results of this study can also be used by the CDPH to tailor unique COVID-19 prevention and intervention strategies to different targeted county subdivisions in Connecticut. This study presents an initial and exploratory step towards better understanding of COVID-19 death rates in Connecticut, U.S.A., but much more in-depth work remains before health researchers and practitioners understand why these spatial variations exist and why explanatory factors, such as income, age, race, type of housing, and cardiovascular disease death rates have relatively low explanatory effects in some county subdivisions, but explain up to 74.2% of the COVID-19 death rates in others.

Haut de page

Bibliographie

Aiken L.S., and West S.G., 1991, Multiple Regression: Testing and Interpreting Interactions, Newbury Park: Sage.

Anselin L., 1995, "Local Indicators of Spatial Association – LISA", Geographical Analysis, Vol.27, No.2, 93-115.

Baird R.P., 2020, "Why Widespread Coronavirus Testing Isn't Coming Anytime Soon", Retrieved from: https://www.newyorker.com/news/news-desk/why-widespread-coronavirus-testing-isnt-coming-anytime-soon (last visited on August 6, 2020).

Becker J.M., Ringle C.M., Sarstedt M. and Völckner M., 2014, "How Collinearity Affects Mixture Regression Results", Marketing Letters, Vol.26, No.4, 643-659.

Berchick E.R., Barnett J.C. and Upton R.D., 2019, Health Insurance Coverage in the United States: 2018, Census Bureau: Washington D.C.

Braun M., and Westin D., 2020, Connecticut Governor Says Days of Commuting to NYC May End, Retrieved from: https://www.bloomberg.com/news/articles/2020-05-13/connecticut-governor-says-days-of-commuting-to-nyc-may-be-over (last visited on August 6, 2020).

Brunsdon C., Fotheringham A.S., and Charlton M.E., 1996, "Geographically Weighted Regression: A Method for Exploring Spatial Non-stationarity", Geographical Analysis, Vol.28, No.4, 281-298.

Bureau of Economic Analysis, 2020, State Annual Personal Income, 2019 (Preliminary) and State Quarterly Personal Income, 4th Quarter 2019, Retrieved from: https://www.bea.gov/sites/default/files/2020-03/spi0320_0_0_0_0_0_0_0.pdf (last visited on August 6, 2020).

Cahill M. and Mulligan G., 2007, "Using Geographically Weighted Regression to Explore Local Crime Patterns", Social Science Computer Review, Vol.25, No.2, 174-193.

Census Bureau, 2020, County Subdivisions, Retrieved from: https://www2.census.gov/geo/pdfs/reference/GARM/Ch8GARM.pdf (last visited on August 6, 2020).

Centers for Disease Control and Prevention, 2020a, Cases in the U.S., Retrieved from: https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/cases-in-us.html (last visited on August 6, 2020).

Centers for Disease Control and Prevention, 2020b, Health Equity Considerations and Racial and Ethnic Minority Groups, Retrieved from: https://www.cdc.gov/coronavirus/2019-ncov/community/health-equity/race-ethnicity.html?CDC_AA_refVal=https%3A%2F%2Fwww.cdc.gov%2Fcoronavirus%2F2019-ncov%2Fneed-extra-precautions%2Fracial-ethnic-minorities.html (last visited on August 6, 2020).

Centers for Disease Control and Prevention, 2020c, Older Adults, Retrieved from: https://www.cdc.gov/coronavirus/2019-ncov/need-extra-precautions/older-adults.html (last visited on August 6, 2020).

Centers for Disease Control and Prevention, 2020d, United States COVID-19 Cases and Deaths by State, Retrieved from: https://www.cdc.gov/covid-data-tracker/#cases (last visited on August 6, 2020).

Centers for Disease Control and Prevention, 2020e, Heart Disease Facts, Retrieved from: https://www.cdc.gov/heartdisease/facts.htm#:~:text=One%20person%20dies%20every%2037,1%20in%20every%204%20deaths (last visited on August 6, 2020).

Chu S., Swarup S., Chen J., and Marathe A., 2017, "A Comparison of Targeted Layered Containment Strategies for a Flu Pandemic in Three US Cities", in Proceedings of the 16th Conference on Autonomous Agents and MultiAgent Systems (pp. 1499-1501), International Foundation for Autonomous Agents and Multiagent Systems.

Chu, D.K., Akl E.A., Duda S., Solo K., Yaacoub S., and Schünemann H.J., 2020, "Physical Distancing, Face Masks, and Eye Protection to Prevent Person-to-person Transmission of SARS-CoV-2 and COVID-19: A Systematic Review and Meta-analysis", The Lancet, VOL.395, No.10242, 1973-1987.

Connecticut COVID-19 Response, 2021, Average Daily Rate of COVID-19 Cases Among Persons Living in Community Settings per 100,000 Population By Town, Retrieved from: https://portal.ct.gov/coronavirus/covid-19-data-tracker (last visited on January 21st, 2021).

Environmental System Research Institute (ESRI), 2018, ArcGIS Desktop: Release 10.6.1, Redlands, CA.

Finch W.H., and Finch M.E.H. 2020, "Poverty and Covid-19: Rates of Incidence and Deaths in the United States During the First 10 Weeks of the Pandemic", Frontiers in Sociology, Vol.5, Article 47.

Fotheringham A.S., Brunsdon C., and Charlton M., 2002, Geographically Weighted Regression: The Analysis of Spatially Varying Relationships, Chichester: Wiley.

Freedman J., 2020, New York is losing its young people. Is Connecticut gaining them? Retrieved from: https://ctmirror.org/category/ct-viewpoints/new-york-is-losing-its-young-people-is-connecticut-gaining-them-jessica-freedman/ (last visited on August 6, 2020).

Gebreab S.Y. and Diez-Roux A.V., 2012, "Exploring Racial Disparities in CHD Mortality between Blacks and Whites across the United States: A Geographically Weighted Regression Approach", Health Place, Vol.18, No.5, 1006-1014.

Gregg E.W., Cheng Y.J., Saydah S., Cowie C, Garfield S., Geiss L., and Barker L., 2012, "Trends in Death Rates Among U.S. Adults With and Without Diabetes Between 1997 and 2006, Findings from the National Health Interview Survey", Diabetes Care, Vol.35, No.6, 1252-1257.

Gupta S., Hayek S., and Wang W., 2020, "Factors Associated With Death in Critically Ill Patients With Coronavirus Disease 2019 in the US", JAMA Internal Medicine, 180(11), 1436-1446

Hamidi S., Sabouri S., and Ewing R., 2020, "Does Density Aggravate the COVID-19 Pandemic? Early Findings and Lessons for Planners", Journal of the American Planning Association, https://doi.org/10.1080/01944363.2020.1777891.

Hipp, J.A., and Chalise, N., 2015, "Spatial Analysis and Correlates of County-level Diabetes Prevalence, 2009-2010", Preventing Chronic Disease, 12:140404. DOI: http://dx.doi.org/10.5888/pcd12.140404.

Kolko J., 2017, Seattle Climbs but Austin Sprawls: The Myth of the Return to Cities, Retrieved from: https://www.nytimes.com/2017/05/22/upshot/seattle-climbs-but-austin-sprawls-the-myth-of-the-return-to-cities.html (last visited on August 6, 2020).

Lakhani A., 2020, "Which Melbourne Metropolitan Areas Are Vulnerable to COVID-19 Based on Age, Disability and Access to Health Services? Using Spatial Analysis to Identify Service Gaps and Inform Delivery", Journal of Pain and Symptom Management, Vol.60, No.1, 41-44.

Lawrence R.J., 2011, "Understanding Environmental Quality Through Quality of Life (QOL) Studies." In Nriagu J.O. (eds.) Encyclopedia of Environmental Health, 518-525.

Leung Y., Mei C. and Zhuang W., 2000, "Statistical Tests for Spatial Nonstationarity Based on the Geographically Weighted Regression Model", Environment and Planning A, Vol.32, No.1, 9-32.

Liu P., Blet A., Smyth D., and Li H., 2020, "The Science Underlying COVID-19: Implications for the Cardiovascular System", Circulation, Vol.142, No.1, 68-78.

Kuderer N.M., Choueiri T.K., Shah D.P., Shyr Y., Rubinstein S.M., and Rivera D.R., 2020, "Clinical Impact of COVID-19 on Patients with Cancer (CCC19): A Cohort Study", The Lancet, Vol.395, No.10241, 1907-1918.

Mateu J., 2010, "Comments On: A Gernal Science-Based Framework for Dynamical Spatio-Temporal Model", Test, Vol.19, No.3, 452-455.

Mollalo A., Vahedi B, and Rivera K.M., 2020, "GIS-based Spatial Modeling of COVID-19 Incidence Rate in the Continental United States", Science of The Total Environment, Vol.728, 138884. https://doi.org/10.1016/j.scitotenv.2020.138884.

Moore, D.A. and Carpenter, T.E., 1999, "Spatial Analytical Methods and Geographic Information Systems: Use in Health Research and Epidemiology", Epidemiologic Reviews, Vol.21, No.2, 143-161.

Ringle C.M., Wende S. and Becker J.M., 2015, SmartPLS 3, Bönningstedt: SmartPLS.

Robinson W.S., 1950, "Ecological Correlations and the Behavior of Individuals", American Sociological Review, Vol.15, No.3, 351-357.

Roth G., 2018, "The Burden of Cardiovascular Diseases Among US States, 1990-2016", JAMA Cardiology, Vol.3, No.5, 375-389.

Stokes E., Zambrano E.D., Anderson K.N., Marder E.P., Raz K.M., Felix S.B., Tie Y., and Fullerton K.E., 2020, "Coronavirus Disease 2019 Case Surveillance – United States, January 22–May 30, 2020", Morbidity and Mortality Weekly Report, Vol.69, 759-765.

Sommeiller E. and Price M., 2014, The Increasingly Unequal States of America: Income Inequality by State, 1917 to 2011, The Economic Policy Institute: Washington D.C.

United Nations, 2020, The Social Impact of COVID-19, Retrieved from https://www.un.org/development/desa/dspd/2020/04/social-impact-of-covid-19/ (last visited on August 6, 2020).

Wadhera R.K., Wadhera P.P., and Gaba P., 2020, "Variation in COVID-19 Hospitalizations and Deaths Across New York City Boroughs", Research Letter, Vol.323, No.21, 2192-2195.

World Health Organization, 2020a, Rolling Updates on Coronavirus Disease (COVID-19), Retrieved from: https://www.who.int/emergencies/diseases/novel-coronavirus-2019/events-as-they-happen (last visited on August 6, 2020).

World Health Organization, 2020b, Report of the WHO-China Joint Mission on Coronavirus Disease 2019 (COVID-19), Retrieved from https://www.who.int/docs/default-source/coronaviruse/who-china-joint-mission-on-covid-19-final-report.pdf (last visited on August 6, 2020).

World Health Organization, 2020c, Coronavirus Disease (COVID-2019) Situation Reports, Retrieved from https://www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reports (last visited on August 6, 2020).

Xiong Y., Wang Y., Chen F., and Zhu M., 2020, "Spatial Statistics and Influencing Factors of the COVID-19 Epidemic at Both Prefecture and County Levels in Hubei Province, China", International Journal of Environmental Research and Public Health, Vol.17, No.11, 1-26.

Haut de page

Table des illustrations

Titre Figure 1. Study Area
URL http://journals.openedition.org/cybergeo/docannexe/image/36057/img-1.png
Fichier image/png, 299k
Titre Figure 2. Daily New COVID-19 Cases in Connecticut, USA
URL http://journals.openedition.org/cybergeo/docannexe/image/36057/img-2.png
Fichier image/png, 59k
Titre Figure 3. Daily New COVID-19 Deaths in Connecticut, USA
URL http://journals.openedition.org/cybergeo/docannexe/image/36057/img-3.png
Fichier image/png, 49k
Titre Figure 4. Local Moran’s I Cluster Analyses of COVID-19 Death Rates in Connecticut
URL http://journals.openedition.org/cybergeo/docannexe/image/36057/img-4.jpg
Fichier image/jpeg, 140k
URL http://journals.openedition.org/cybergeo/docannexe/image/36057/img-5.png
Fichier image/png, 3,4k
URL http://journals.openedition.org/cybergeo/docannexe/image/36057/img-6.png
Fichier image/png, 3,5k
Titre Figure 5. Spatial Variations in β Coefficient Estimates in the State of Connecticut for Explanatory Variables: Income (Map A), Age (Map B), Race (Maps C), Type of Housing (Map D), Underlying Condition Indicator 1 – Cardiovascular Disease Death Rate (Map E), and Local R2 Value (Map F) from the GWR model of COVID-19 Death Rates.
URL http://journals.openedition.org/cybergeo/docannexe/image/36057/img-7.jpg
Fichier image/jpeg, 656k
Titre Figure 6. Spatial Variations in t Values for Explanatory Variables in Modeling COVID-19 Death Rates in Connecticut: Income (Map A), Age (Map B), Race (Maps C), Type of Housing (Map D), and Cardiovascular Disease Death Rate (Map E). Note: 1.96 and 2.58 are the cut-off values for t-test. When │t│ > 1.96, the β coefficient estimate for a variable is significant at a significance level of 0.05. When │t│ > 2.58, the β coefficient estimate for a variable is significant at a significance level of 0.01.
URL http://journals.openedition.org/cybergeo/docannexe/image/36057/img-8.jpg
Fichier image/jpeg, 452k
Haut de page

Pour citer cet article

Référence électronique

Yunliang Meng, « COVID-19 Death Rates and County Subdivision Level Contextual Characteristics: A Connecticut Case Study », Cybergeo: European Journal of Geography [En ligne], Espace, Société, Territoire, document 965, mis en ligne le 10 février 2021, consulté le 29 mars 2024. URL : http://journals.openedition.org/cybergeo/36057 ; DOI : https://doi.org/10.4000/cybergeo.36057

Haut de page

Auteur

Yunliang Meng

Dr. Yunliang Meng, Associate Professor, Department of Geography, Central Connecticut State University, New Britain, Connecticut, 06053, United States of America. Email: mengy@ccsu.edu

Articles du même auteur

Haut de page

Droits d’auteur

CC-BY-4.0

Le texte seul est utilisable sous licence CC BY 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search