Background: New York City (NYC) was the epicenter of the COVID-19 pandemic, and is home to underserved populations with higher prevalence of chronic conditions that put them in danger of more serious infection. Little is known about how the presence of chronic risk factors correlates with mortality at the population level. Here we determine the relationship between these factors and COVD-19 mortality in NYC. Methods: A cross-sectional study of mortality data obtained from the NYC Coronavirus data repository (03/02/2020–07/06/2020) and the prevalence of neighborhood-level risk factors for COVID-19 severity was performed. A risk index was created based on the CDC criteria for risk of severe illness and complications from COVID-19, and stepwise linear regression was implemented to predict the COVID-19 mortality rate across NYC zip code tabulation areas (ZCTAs) utilizing the risk index, median age, socioeconomic status index, and the racial and Hispanic composition at the ZCTA-level as predictors. Results: The COVID-19 death rate per 100,000 persons significantly decreased with the increasing proportion of white residents (βadj = − 0.91, SE = 0.31, p = 0.0037), while the increasing proportion of Hispanic residents (βadj = 0.90, SE = 0.38, p = 0.0200), median age (βadj = 3.45, SE = 1.74, p = 0.0489), and COVID-19 severity risk index (βadj = 5.84, SE = 0.82, p < 0.001) were statistically significantly positively associated with death rates. Conclusions: Disparities in COVID-19 mortality exist across NYC and these vulnerable areas require increased attention, including repeated and widespread testing, to minimize the threat of serious illness and mortality.
Keywords: Mortality; Coronavirus; Comorbidities
Supplementary Information The online version contains supplementary material available at https://doi.org/10.1186/s12889-021-11498-x.
New York City was once considered the epicenter of the coronavirus disease (COVID-19) epidemic in the United States. From the first reported case on March 1, 2020, to April 2020, nearly 600 people were dying each day, new daily cases numbered in the thousands, and new hospitalizations were roughly 1500 per day [[
Clinical studies across the US and NYC have shown that people with underlying health conditions and risk factors, including diabetes, chronic lung disease, and cardiovascular disease, are at higher risk for severe outcomes and mortality from COVID-19 [[
Risk factors were identified based on the CDC criteria for risk of severe illness and complications from COVID-19 [[
Each risk factor was then divided into quartiles and assigned a score of 1 to 4, with the lowest quartile (score 1) representing areas with the lowest severe COVID-19 risk associated with a given variable, and the highest quartile (score 4) representing the highest risk. Risk quartiles for Hepatitis B, C, and alcohol related hospitalizations were summed to create one variable capturing liver disease. These scores were then summed across all risk factors into a single score for each ZCTA, with higher values corresponding to higher prevalence of risk factors. NYC ZCTAs were classified into quartiles of this risk score.
Zip Code Tabulation Area (ZCTA) level data on median household income in the past 12 months (Table B19013), median gross rent (B25064), percent living below 150% of the poverty line (Table C17002), education (B15002), percent working class (C24010), percent unemployed (B23025), and > 1 occupants per room (B25014) was downloaded from 2018 American Community Survey (ACS) 5-year estimates [[
The COVID-19 related death count and the COVID-19 death rate were downloaded from the NYC Coronavirus (COVID-19) data repository [[
The geographic unit of analysis was the ZCTA, where each ZCTA has a risk index score, SES index score, and mortality information. Wilcoxon rank-sum tests were performed to assess differences in COVID-19 death rate, risk index components, racial and ethnic composition, and median age according to risk index quartiles. Spearman correlations were performed to assess the association between the COVID-19 death rate per 100,000 residents, the risk index, median age, SES index, and the racial and Hispanic composition of the ZCTA. Stepwise linear regression was performed to predict the COVID-19 death rate per 100,000 residents through July 6th utilizing the risk index, median age, socioeconomic status index, and the racial and Hispanic composition at the ZCTA level as predictors. Variables that were significant at the α = 0.25 level were considered for entry into the multivariable model and those that were significant at the α = 0.15 level after adjustment remained in the final model. All analyses were performed in SAS v9.4.
Each component of the risk index, as well as the proportion of white and Hispanic residents, median age, SES index and COVID-19 death rate were analyzed according to quartiles of the risk index (Table 1). Across risk index quartiles representing increasing risk, the COVID-19 death rate per 100,000 (p < 0.0001), asthma prevalence (p < 0.0001), kidney disease prevalence (p < 0.0001), hypertension prevalence (p < 0.0001), heart disease prevalence (p < 0.0001) obesity prevalence (p < 0.0001), COPD prevalence (p < 0.0001), diabetes prevalence (p < 0.0001), Hepatitis C (p < 0.0001), Hepatitis B (p = 0.0055), proportion of residents greater than 65 years old (p = 0.0378), birth rate (p = 0.0325), alcohol hospitalizations (p = 0.0005), and the proportion of Hispanic residents (p = 0.0004) increased, while the proportion white residents (p < 0.0001) and the SES index (p < 0.0001) decreased. All-cancer incidence rates (p = 0.1333) and median age (p = 0.1326) were similar across risk sum quartiles.
Table 1 Description of COVID-19 death rate per 100,000 residents, risk index components, and socioeconomic status index according to quartiles of the risk index
Variable Quartile 1 (lowest risk score; 21.46–40.26) mean (SD) Quartile 2 (40.27–48.15) mean (SD) Quartile 3 (48.16–55.01) mean (SD) Quartile 4 (highest risk score; 55.02–73.84) mean (SD) COVID-19 death rate per 100,000 109.72 (68.83) 192.78 (72.51) 251.98 (97.16) 284.16 (112.68) < 0.0001 Median age (years) 35.4 (3.31) 37.64 (3.81) 36.71 (4.26) 36.73 (6.23) 0.1326 Proportion white residents (%) 66.68 (15.40) 45.89 (21.26) 36.78 (23.53) 32.21 (26.81) < 0.0001 Proportion Hispanic residents (%) 17.12 (12.400 29.13 (18.53) 25.67 (18.82) 34.52 (22.98) 0.0004 Obesity prevalence (%) 17.87 (5.01) 22.31 (4.24) 26.53 (4.69) 29.72 (5.98) < 0.0001 Kidney disease prevalence (%) 2.17 (0.49) 2.84 (0.27) 3.20 (0.26) 3.67 (0.48) < 0.0001 Hypertension prevalence (%) 20.36 (4.00) 26.43 (2.15) 30.37 (3.13) 33.60 (3.41) < 0.0001 Heart Disease prevalence (%) 3.75 (1.01) 5.25 (0.42) 5.76 (0.59) 6.49 (0.81) < 0.0001 Diabetes prevalence (%) 6.87 (2.27) 10.48 (1.58) 11.78 (1.40) 13.13 (2.17) < 0.0001 COPD prevalence (%) 3.78 (1.11) 5.17 (0.62) 6.04 (0.72) 6.75 (0.90) < 0.0001 Cancer (except skin) prevalence (%) 5.02 (1.14) 5.49 (1.03) 5.55 (0.94 5.79 (1.53) 0.1333 Asthma prevalence (%) 8.46 (0.74) 8.92 (0.95) 10.11 (1.34) 10.80 (1.26) < 0.0001 Alcohol Hospitalizations prevalence (%) 1.71 (1.02) 1.36 (0.56) 1.51 (0.77) 2.26 (1.20) 0.0005 Birth rate (%) 1.25 (0.37) 1.16 (0.32) 1.26 (0.32) 1.30 (0.23) 0.0325 Hepatitis C prevalence (%) 0.05 (0.02) 0.05 (0.01) 0.05 (0.01) 0.08 (0.03) < 0.0001 Hepatitis B prevalence (%) 0.06 (0.07) 0.09 (0.09) 0.08 (0.07) 0.07 (0.04) < 0.0001 Proportion ≥ 65 years old (%) 12.3 (4.7) 14.88 (4.09) 14.51 (3.72) 15.73 (5.95) 0.0378 Socioeconomic status index 33.87 (21.38) 12.03 (17.72) −1.20 (18.74) −16.37 (22.08) < 0.0001
The racial and Hispanic composition of each ZCTA were obtained from the 2018 American Community Survey 5-year estimates. Unless otherwise noted, all variables are reported per hundreds of residents. The socioeconomic status index was constructed from principal component analysis of 2018 American Community Survey estimates of median household income in the past 12 months, median gross rent, percent living below 150% of the poverty line, education, percent working class, percent unemployed, > 1 occupants per room. The education index was calculated on the population ≥ 25 years and is a weighted combination of the percent high school graduate, high school only, and more than high school, with a greater value indicating higher educational attainment
Correlations among the COVID-19 death rate, risk index, proportion White residents, proportion Hispanic residents, the median age, and the SES index (Table 2) indicated that the COVID-19 death rate was significantly positively correlated with the risk index (ρ = 0.67, p < 0.0001), increasing proportion of Hispanic residents (ρ = 0.42, p < 0.0001), and significantly inversely correlated with the proportion of white residents (ρ = − 0.56, p < 0.0001) and the SES index (ρ = − 0.63, p < 0.0001). The risk index was significantly positively correlated with the increasing proportion of Hispanic residents (ρ = 0.25, p = 0.0012), and significantly inversely correlated with the proportion white residents (ρ = − 0.48, p < 0.0001) and the SES index (ρ = − 0.67, p < 0.0001). The correlation between the risk index components and the SES index is reported in Supplementary Table 2.
Table 2 Correlations among predictors and COVID-19 death rate per 100,000 residents
Statistic Death Rate Up to July 6 Risk Index White Proportion (%) Hispanic Proportion (%) Median Age (years) SES Index Death Rate up to July 6 ρ – 0.67 −0.56 0.42 0.03 −0.63 p-value – <0.0001 <0.0001 <0.0001 0.6441 <0.0001 Risk Index ρ – – − 0.48 0.25 0.08 −0.67 p-value – – <0.0001 0.0012 0.27 <0.0001 White Proportion (%) ρ – – – −0.40 0.32 0.74 p-value – – – <0.0001 <0.0001 <0.0001 Hispanic Proportion (%) ρ – – – – −0.39 −0.69 p-value – – – – <0.0001 <0.0001 Median Age (years) ρ – – – – – 0.37 p-value – – – – – <0.0001
Note: Spearman correlations were performed The socioeconomic status index was constructed from principal component analysis of 2018 American Community Survey estimates of median household income in the past 12 months, median gross rent, percent living below 150% of the poverty line, education, percent working class, percent unemployed, > 1 occupants per room. The education index was calculated on the population ≥ 25 years and is a weighted combination of the percent high school graduate, high school only, and more than high school, with a greater value indicating higher educational attainment
There are 177 ZCTAs in NYC reporting COVID-19 death rate data and 18,851 cumulative deaths in NYC as of July 6, 2020. The distribution of the COVID-19 death rate were mapped according to each ZCTA, as well as the risk index, the median age, the SES index, and the racial and Hispanic proportion (Fig. 1). Death rates per 100,000 residents were highest in the most Northern portion of NYC encompassing the Bronx, and were lowest in Manhattan. There was heterogeneity among ZCTAs in other NYC boroughs, including in Staten Island, which has low death rates in the southwest but high death rates in the northeast portion of the borough. Likewise areas of central Queens and eastern Brooklyn have high death rates, while death rates decrease in areas closer to Manhattan in these boroughs. Areas with high death rates have noticeable overlap with the risk score and inverse correlations with socioeconomic index, such as the Bronx. However, areas of central Queens with high death rates have moderate risk and socioeconomic scores, illustrating the complex interplay between risk factors and mortality.
Graph: Fig. 1 Distribution of the COVID-19 death rate per 100,000 residents from March 2nd to July 6th (top-left), the risk index (top-right), the median age (middle-left), the proportion white residents (middle-right), the socioeconomic status index (bottom-left) and the Hispanic composition (bottom-right) across New York City Zip Code Tabulation Areas. The risk index is a sum score of quartiles of asthma prevalence + cancer prevalence (excluding skin cancer) + COPD prevalence + diabetes prevalence + heart disease prevalence + hypertension prevalence + kidney disease prevalence + obesity prevalence + Heart attack prevalence + proportion of the population aged ≥65 years + birthrate prevalence and liver disease (measured by summing quartiles of Hepatitis B prevalence, Hepatitis C prevalence and alcohol hospitalizations prevalence). The socioeconomic status index was constructed from principal component analysis of 2018 American Community Survey estimates of median household income in the past 12 months, median gross rent, percent living below 150% of the poverty line, education, percent working class, percent unemployed, > 1 occupants per room. The education index was calculated on the population ≥ 25 years and is a weighted combination of the percent high school graduate, high school only, and more than high school, with a greater value indicating higher educational attainment. This figure was created using our licensed copy of ArcGIS (version 10.7.1; ESRI, Redlands, CA)
In the final multivariable model, the COVID-19 death rate per 100,000 persons significantly decreased with the increasing proportion of white residents in the ZCTA (β
Table 3 Predictors of cumulative COVID-19 death rate per 100,000 residents
βadj (SE) Intercept − 176.12 (61.35) 0.005 White residents (%) −0.91 (0.31) 0.004 Hispanic Composition (%) 0.90 (0.38) 0.02 Median Age (years) 3.45 (1.74) 0.049 Risk Index 5.84 (0.82) < 0.0001
This is the first analysis to incorporate CDC-defined risk factors for severe COVID-19 illness, NYC mortality data, racial and ethnic composition, and SES to predict COVID-19 death rates across NYC at the neighborhood level. Here we identify that areas with fewer white residents, more Hispanic residents, older residents, and with higher prevalence of a greater number of risk factors had increased COVID-19 mortality. These findings expand on existing literature that has partially addressed the potential risk of severe COVID-19 illness deriving from the complex intersection of chronic risk factors, SES and race [[
It has been noted that blanket public health recommendations in regards to social distancing for COVID-19 do not take local contexts and populations into account [[
One limitation of this analysis is that the data used are aggregate ZCTA-level data, which limits our ability to draw individual-level conclusions. For instance, we cannot comment on the interaction between being non-white and having chronic risk factors on the likelihood of mortality, information that could help identify the most vulnerable populations. This study was ecological in nature and there are more individual level factors that may influence COVID-19 mortality that could not be taken into account here, although an accepted and comprehensive risk factor measure from the CDC was utilized. Another limitation is the cross-sectional nature of the analysis. Future studies should investigate these relationships with mortality using a larger study period, representing a larger number of COVID-19 related deaths, to validate these findings.
This analysis identifies that there are behavioral and racial/ethnic disparities in COVID-19 mortality, and highlights the most vulnerable areas in NYC that require increased focus from COVID-19 response policies. This analysis represents the first to our knowledge to investigate risk factors for COVID-19 mortality at the population-level in New York City, and supports the need for repeated and widespread testing in these communities with many risk factors for severe illness and mortality from COVID-19. Additionally, this analysis shines light on vulnerable communities that require equitable resources to rebound from COVID-19 related hardships and mortality in these communities. Future analyses could explore how the presence of comorbidities influences the immune response following COVID-19 infection, as this could ultimately be a major determinant of mortality outcomes [[
Not applicable.
ET conceptualized the study. NA provided statistical expertise. WL-C conducted the literature search. RF provided context for results within the COVID-19 landscape. ET, NA, and WL-C performed data analyses. All authors contributed to data interpretation and manuscript preparation. The corresponding author takes full responsibility for the content of the manuscript, including the data and analysis. All authors have read and approved the final manuscript.
Wil Lieberman-Cribbin: wil.lieberman-cribbin@icahn.mssm.edu
Naomi Alpert: naomi.alpert@mountsinai.org
Raja Flores: raja.flores@mountsinai.org
Emanuela Taioli: emanuela.taioli@mountsinai.org
No funding was received for this work.
Data used in this analysis is available for open public access at https://github.com/nychealth/coronavirus-data. Individual data was not used.
Not applicable.
Not applicable.
The authors declare that they have no competing interests.
Graph: Additional file 1: Supplementary Table 1. Description of Risk Factor Measures. Supplementary Table 2. Correlations among socioeconomic status index and risk index components. Note: Spearman correlations were performed
• ACS
- American Community Survey
• CDC
- Centers for Disease Control and Prevention
• COPD
- chronic obstructive pulmonary disease
- COVID-19
- coronavirus disease
• NYC
- New York City
• NYCDOH
- New York City Department of Health
• PCA
- principal component analysis
• SES
- Socioeconomic status
• UHF
- United Hospital Fund
• US
- United States
• ZCTA
- Zip code tabulation area
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
By Wil Lieberman-Cribbin; Naomi Alpert; Raja Flores and Emanuela Taioli
Reported by Author; Author; Author; Author