The COVID-19 pandemic disproportionately impacted communities of color in New York City. Early evidence suggested that social disadvantage limited the capacity for social distancing, leading to disparities in infection rates and mortality. This study aimed to quantify the association between neighborhood-level social disadvantage and COVID-19 outcomes. The research question focused on whether socioeconomic factors were associated with infection rates, mortality, and the ability to socially distance. The study's importance stems from the need to understand and address health disparities exacerbated by the pandemic. Understanding these disparities is crucial for developing targeted public health interventions and achieving equitable outcomes. The context of the study lies in the early stages of the pandemic in NYC, before widespread vaccine availability or effective treatment options, making non-pharmaceutical interventions like social distancing critical.

Existing literature highlighted the disproportionate impact of COVID-19 on communities of color. Studies in Chicago and NYC showed higher mortality rates among Black and Hispanic/Latinx populations compared to white populations. While pre-existing conditions contributed to disease severity, these differences did not fully explain variations in infection rates. The social determinants of health literature suggested that structural inequalities in housing, employment, and access to healthcare created a scaffolding for higher infection rates in these communities. Residential segregation and structural disadvantages were identified as contributing factors to racial disparities in infectious diseases. Research also showed that social distancing was more challenging for communities of color due to factors such as essential work, dense housing, and limited access to resources. These studies underscored the need for a more comprehensive understanding of how social factors contributed to the unequal burden of COVID-19.

This population-level ecological study used socioeconomic data from the 2018 American Community Survey (ACS) to analyze differences in infection rates among NYC neighborhoods. The researchers created a ZIP code-level COVID-19 inequity index using Bayesian Weighted Quantile Sums (BWQS) regression. This index combined ten social variables (average household size, median income, proportion uninsured, unemployment, essential worker status by commute mode, population density, and grocery store access) trained on infection rates and adjusted for testing rates. The BWQS method weighted the variables based on their relative contribution to the outcome. The study also used MTA subway ridership data as a proxy for social distancing capacity. The spatial distribution of the index was compared to infection and mortality data. Cross-sectional analyses assessed the association between the index and cumulative infections and mortality. Longitudinal analyses using subway ridership data examined differences in social distancing capacity across neighborhoods with varying index values using a generalized Weibull model. To account for spatial autocorrelation, a spatial filtering approach was used in the mortality analysis. The analysis involved several steps: data cleaning and preparation using ACS data, NYC building footprints, and food access data. They combined this with publicly available data on SARS-CoV-2 testing and COVID-19 mortality from the NYC Department of Health and Mental Hygiene (DOHMH). They utilized the BWQS regression method to develop a weighted index for neighborhood-level social disadvantage. Subway ridership data from the MTA were used to assess social distancing capacity. The study used negative binomial regression models and spatial filtering to analyze the relationship between the COVID-19 inequity index and COVID-19 mortality, accounting for spatial autocorrelation. They performed sensitivity analyses using tract-level data to estimate ZCTA-level exposure metrics. They also compared BWQS results with those from negative binomial regression using a subset of variables and principal component regression. The study used several statistical methods, including Kendall's tau correlation, Bayesian R-squared, root mean squared error (RMSE), and WAIC.

The BWQS regression revealed a strong association between the composite index of neighborhood social disadvantage and cumulative COVID-19 infection rates. Each unit increase in the index was associated with an 8% increase in infections per capita. The proportion of uninsured people, average household size, and proportion of essential workers commuting by personal vehicle were the most influential variables in the index. The spatial distribution of the index mirrored the pattern of infections. Black and Hispanic/Latinx communities were overrepresented in high-index neighborhoods, while white communities were overrepresented in low-index neighborhoods. Subway ridership data indicated lower social distancing capacity in high-index areas. While the rate of decrease in subway ridership was similar across high and low-index areas after the NYS on PAUSE order, the baseline subway usage was notably higher in high-index areas. Finally, the COVID-19 inequity index was strongly associated with cumulative COVID-19 mortality, with each unit increase in the index associated with a 20% increased risk of mortality, even after accounting for spatial autocorrelation.

The findings strongly support the hypothesis that neighborhood social disadvantage is associated with increased COVID-19 infection rates, reduced capacity for social distancing, and higher mortality in NYC. The COVID-19 inequity index effectively captured the combined effect of multiple social factors contributing to these disparities. The disproportionate representation of Black and Hispanic/Latinx communities in high-index areas highlights the role of structural racism in shaping health disparities. The use of subway ridership as a proxy for social distancing capacity demonstrated the limitations faced by disadvantaged communities in complying with public health guidelines. The association between the index and mortality underscores the long-term consequences of social disadvantage on health outcomes. These results add to the growing body of evidence demonstrating the critical role of social determinants of health in shaping the impact of infectious disease outbreaks.

This study provides a novel approach to quantifying neighborhood-level social disadvantage and its association with COVID-19 disparities. The COVID-19 inequity index can be a valuable tool for targeting interventions and resources to vulnerable communities. Future research should explore the generalizability of the index to other geographic areas and populations, and investigate the underlying mechanisms linking social disadvantage to health outcomes. Further investigation into more nuanced measures of social isolation, such as multigenerational housing, and the direct incorporation of race and ethnicity in future models would add valuable insights.

The study used ZCTA-level data, which may mask intra-ZCTA heterogeneity. The absence of ZCTA-level data on multigenerational housing and chronic diseases limited the analysis. Early testing protocols may have confounded infection data. Pre-pandemic social data may not fully capture changes in residential mobility during the pandemic. The use of subway ridership as a proxy for social distancing capacity may not fully capture other modes of transportation. The BWQS method is relatively novel in this field. Finally, there is a risk of stigmatizing high-index neighborhoods.