Spatiotemporal contact density explains the disparity of COVID-19 spread in urban neighborhoods

Rajat Verma, Takahiro Yabe, Satish V. UkkusuriScientific 2021 · DOI

TL;DR

This paper introduces a novel "contact density index" (CDI) using mobile phone mobility data to quantify interpersonal contact in urban neighborhoods. They found that higher contact density, particularly in lower-income areas, significantly correlates with increased COVID-19 case growth, suggesting a causal link between income, contact exposure, and disease spread.

What Problem Does This Paper Address?

The COVID-19 pandemic highlighted stark disparities in its spread and impact across different socioeconomic groups and geographic areas. While many studies have looked at broad mobility patterns and social distancing measures, there was a gap in understanding the specific nature of interpersonal contact within urban neighborhoods and how it contributes to these disparities. Traditional mobility metrics like traffic flow or distance traveled don't capture the nuances of close proximity and duration of contact, which are crucial for disease transmission. This paper aims to bridge this gap by developing a more granular measure of contact density and linking it to COVID-19 case growth, particularly in relation to socioeconomic factors.

Methods

The core methodological contribution is the development of a Spatiotemporal Contact Density Index (CDI). This index quantifies the strength of interpersonal contact by considering three key factors:

  • Number of visitors: How many people visit a specific location (Point of Interest - POI).
  • Spatial distribution: How spread out visitors are within the POI (approximated by floor area).
  • Duration of stay: How long visitors remain at the POI.

The CDI is calculated at the POI level and then aggregated to the neighborhood (ZCTA - Zip Code Tabulation Area) level. To understand the causal relationships, the researchers employed Structural Equations Modeling (SEM). This statistical technique allowed them to model the hypothesized causal pathways: from socioeconomic factors (like income) to latent exposure (measured by CDI and other social distancing metrics), and then to the daily growth of COVID-19 cases. They also analyzed the contribution of different industry types (POIs) to the overall CDI.

Datasets

  • Mobility Data: Anonymized origin-destination foot traffic movement patterns from mobile phone GPS records, aggregated by SafeGraph Inc. This data provided information on POI visits, including their industry (NAICS code), floor area, visitor counts, dwell times, and origin CBGs (Census Block Groups).
  • Epidemiological Data: Daily new positive COVID-19 cases at the ZCTA level for New York City and Chicago. Data for Chicago was available from April 18, 2020, and for NYC from April 3, 2020.
  • Socioeconomic Data: 2017 American Community Survey (ACS) data for socioeconomic variables (e.g., mean household income, education level, poverty rate) at the CBG level, aggregated to ZCTA.

Key Insights and Ideas

  • Contact Density as a Key Driver: The study strongly suggests that the density of interpersonal contact, not just overall mobility, is a critical factor in COVID-19 spread. This moves beyond simple "stay-at-home" metrics to understand where and how people are interacting.
  • Socioeconomic Disparity Amplified by Contact: Lower-income neighborhoods consistently showed higher contact density and, consequently, higher COVID-19 case growth. This points to a causal pathway where income influences exposure to contact, which in turn drives disease spread.
  • Industry-Specific Impact: Certain industries, particularly schools and full-service restaurants, were identified as high-contact density locations. Targeted restrictions on these types of POIs were found to be most effective in reducing overall exposure.
  • Limitations of Traditional Metrics: The paper highlights how simple measures like trip counts can be misleading, as they don't account for the actual risk of transmission during interactions at destinations.

Key Results

  • The CDI was consistently positive and higher in lower-income neighborhoods, indicating greater interpersonal contact.
  • SEM analysis revealed a significant causal path: Income → Latent Exposure (CDI) → COVID-19 Cases.
  • The effect of income on exposure was consistently negative, meaning lower income was associated with higher exposure.
  • The effect of exposure (CDI) on new cases was consistently positive, meaning higher exposure led to more cases.
  • Schools and restaurants were identified as major contributors to contact density. Their closure or restriction significantly reduced overall CDI.
  • While overall mobility declined during lockdowns, the nature of contact within remaining activities became more critical.

Limitations and Future Work

  • Data Limitations: The study acknowledges the lack of ZCTA-level epidemiological data before April 2020, which limits the ability to establish causality for the earliest phase of the pandemic. The mobility data also has limitations regarding representative coverage and potential privacy concerns.
  • Contact with Infected Individuals: The CDI measures general contact density, not specifically contact with infected individuals. Future work could incorporate epidemiological models to account for this.
  • Assumptions in CDI Calculation: The CDI relies on assumptions about visitor distribution and arrival times within POIs, which may not always hold true.
  • Future Directions: The authors suggest extending the study to include epidemiological models and further refining the CDI measure. They also emphasize the utility of this approach for monitoring the effectiveness of mobility restriction policies.

Summary updated

Abstract

The rapid early spread of COVID-19 in the US was experienced very differently by different socioeconomic groups and business industries. In this study, we study aggregate mobility patterns of New York City and Chicago to identify the relationship between the amount of interpersonal contact between people in urban neighborhoods and the disparity in the growth of positive cases among these groups. We introduce an aggregate spatiotemporal contact density index (CDI) to measure the strength of this interpersonal contact using mobility data collected from mobile phones, and combine it with social distancing metrics to show its effect on positive case growth. With the help of structural equations modeling, we find that the effect of CDI on case growth was consistently positive and that it remained consistently higher in lower-income neighborhoods, suggesting a causal path of income on case growth via CDI. Using the CDI, schools and restaurants are identified as high contact density industries, and the estimation suggests that implementing specific mobility restrictions on these point-of-interest categories is most effective. This analysis can be useful in providing insights for government officials targeting specific population groups and businesses to reduce infection spread as reopening efforts continue to expand across the nation.
BibTeX
@inproceedings{verma2021spatiotemporal,
  author = {Rajat Verma and Takahiro Yabe and Satish V. Ukkusuri},
  title = {{Spatiotemporal contact density explains the disparity of COVID-19 spread in urban neighborhoods}},
  booktitle = {Scientific},
  year = {2021},
  doi = {10.1038/s41598-021-90483-1},
}
Figure