Road networks and socio-demographic factors to explore COVID-19 infection during its different waves

Shahadat Uddin, Arif Khan, Haohui Lu, Fangyu Zhou, Shakir Karim, Farshid Hajati, Mohammad Ali MoniScientific 2024 · DOI

Abstract

The COVID-19 pandemic triggered an unprecedented level of restrictive measures globally. While lockdowns and social distancing generally slowed case growth, concerns persist regarding the socio-economic impact on disadvantaged segments of society. This study proposes a road-network-based approach to model mobility patterns between localities during lockdown stages. Utilizing a panel regression method, we analyze the effects of mobility on COVID-19 transmission in Sydney, Australia, alongside socio-demographic variables like age, income, and education. We model how local road networks between neighboring suburbs (the "neighbourhood measure") and current infection counts affect case growth across the Delta and Omicron waves. Results show strong model performance in predicting infection rates based on road connectivity during the Delta wave. However, apart from age in the Delta context, income and education levels did not significantly moderate the relationship between infection rates and road networks. These findings suggest that socio-economically disadvantaged suburbs do not necessarily contribute more to community transmission, offering valuable insights for future public health decision-making.

TL;DR

This study uses suburban road networks as a structural proxy for human mobility to model COVID-19 transmission across 100 Sydney suburbs. The researchers found that while road connectivity strongly predicted infection rates during the Delta wave (under lockdown), it became a less reliable predictor during the highly transmissible Omicron wave, and surprisingly, socio-economic disadvantages like lower income or education did not inherently drive higher community transmission.

What Problem Does This Paper Address?

Predicting how a virus spreads through a city often relies on "black box" mobility data from third-party providers like Google or Facebook. However, this data can suffer from sampling bias and only reflects past behavior.

The authors address a critical gap: can we use the physical "skeleton" of a city—its road network—to predict disease spread? Furthermore, they tackle the controversial question of whether socio-demographic factors (age, income, and education) moderate how mobility affects infection rates. This is particularly relevant given the public debate over whether disadvantaged communities were unfairly blamed for community transmission during lockdowns.

Methods

The researchers proposed a "neighbourhood measure" ($RNInf_{t-1}$) to quantify the risk of infection influx. Instead of just looking at a suburb's own cases, they calculated a weighted sum of infections in neighboring suburbs, where the weights were determined by the number of direct road connections between them.

Example of road network construction for the Burwood suburb and its neighbors.

Key technical components include:

  • Network Construction: Suburbs were treated as nodes, and roads as edges. Edge weights represented the actual count of roads connecting two suburbs (derived from Google Maps).
  • Panel Regression: A fixed-effect panel regression model was used to analyze time-series data across four-week windows for both the Delta and Omicron variants.
  • Moderation Analysis: The team tested if age, income, and education levels changed the relationship between road connectivity and infection rates by creating composite interaction variables.

Datasets

The study utilized several integrated datasets centered on the Greater Sydney area in Australia:

  • Infection Data: Weekly COVID-19 case counts for 100 suburbs from NSW Health, covering the Delta wave (Aug–Sept 2021) and the Omicron wave (Nov–Dec 2021).
  • Road Network Data: Manual extraction of road connection counts between suburbs via Google Maps.
  • Socio-demographic Data: Median age, median weekly household income, and tertiary education percentages from the 2016 Australian Bureau of Statistics (ABS) Census.
Spatial distribution of the suburban network across Greater Sydney.

Key Insights and Ideas

  • Infrastructure as Destiny (During Lockdowns): During the Delta wave, when strict 5km travel limits were in place, the road network was an excellent predictor of spread ($R^2 = 0.8566$). This suggests that when movement is restricted, the physical "ease of access" between neighbors becomes the primary driver of transmission.
  • The Omicron Shift: The model was significantly less effective for the Omicron variant ($R^2 = 0.5267$). The authors attribute this to Omicron’s higher transmissibility and the lack of lockdowns, which allowed for more erratic, long-distance movement that bypassed simple suburban road-link logic.
  • Debunking Socio-Economic Blame: One of the most important findings was that income and education levels did not significantly moderate the relationship between road networks and infection rates. This suggests that suburbs with lower socio-economic status were not inherently "riskier" or more prone to spreading the virus through their local networks than wealthier areas.
  • Age Matters: Age was the only socio-demographic factor that significantly moderated the Delta wave. Suburbs with higher median ages saw a stronger correlation between road connectivity and case growth, likely due to the higher clinical vulnerability of older populations.

Key Results

  • Model Accuracy: The fixed-effect model explained 85.7% of the variance in the Delta wave but only 52.7% in the Omicron wave.
  • Neighbourhood Impact: In the Delta wave, the "neighbourhood measure" had a significant positive coefficient (0.2642, $p < 0.01$), meaning more road connections to infected neighbors directly led to more local cases.
  • Counter-intuitive Omicron Result: For Omicron, the neighbourhood measure actually showed a negative impact (-0.1016), suggesting that local road connectivity was no longer the dominant factor in how the virus moved through the city.

Limitations and Future Work

The authors acknowledge two primary limitations:

  1. Road Capacity: The study counted the number of roads but did not distinguish between a small two-lane street and a multi-lane highway, which have vastly different transport capacities.
  2. Model Assumptions: While panel regression is robust, it relies on specific statistical assumptions. The authors suggest that future work could employ Bayesian structural time-series models to better capture complex temporal dynamics.

The study concludes that while road networks are a powerful tool for tailoring health decisions during restricted-movement phases (like lockdowns), they must be combined with other data as variants become more transmissible and society reopens.

Summary updated

Abstract

The COVID-19 pandemic triggered an unprecedented level of restrictive measures globally. Most countries resorted to lockdowns at some point to buy the much-needed time for flattening the curve and scaling up vaccination and treatment capacity. Although lockdowns, social distancing and business closures generally slowed the case growth, there is a growing concern about these restrictions' social, economic and psychological impact, especially on the disadvantaged and poorer segments of society. While we are all in this together, these segments often take the heavier toll of the pandemic and face harsher restrictions or get blamed for community transmission. This study proposes a road-network-based networked approach to model mobility patterns between localities during lockdown stages. It utilises a panel regression method to analyse the effects of mobility in transmitting COVID-19 in an Australian context, together with a close look at a suburban population's characteristics like their age, income and education. Firstly, we attempt to model how the local road networks between the neighbouring suburbs (i.e., neighbourhood measure) and current infection count affect the case growth and how they differ between delta and omicron variants. We use a geographic information system, population and infection data to measure road connections, mobility and transmission probability across the suburbs. We then looked at three socio-demographic variables: age, education and income and explored how they moderate independent and dependent variables (infection rates and neighbourhood measures). The result shows strong model performance to predict infection rate based on neighbourhood road connection. However, apart from age in the delta variant context, the other variables (income and education level) do not seem to moderate the relationship between infection rate and neighbourhood measure. The results indicate that suburbs with a more socio-economically disadvantaged population do not necessarily contribute to more community transmission. The study findings could be potentially helpful for stakeholders in tailoring any health decision for future pandemics.
BibTeX
@inproceedings{uddin2024road,
  author = {Shahadat Uddin and Arif Khan and Haohui Lu and Fangyu Zhou and Shakir Karim and Farshid Hajati and Mohammad Ali Moni},
  title = {{Road networks and socio-demographic factors to explore COVID-19 infection during its different waves}},
  booktitle = {Scientific},
  year = {2024},
  doi = {10.1038/s41598-024-51610-w},
}
Figure