Time-series modeling of epidemics in complex populations: Detecting changes in incidence volatility over time

Area of Science:

• Epidemiology
• Statistical modeling
• Public health

Background:

• Infectious disease incidence trends are crucial for understanding epidemic dynamics and control strategies.
• Temporal volatility in incidence data, often overlooked, can reveal underlying epidemic drivers like transmission heterogeneity and reporting fluctuations.
• Current analyses often rely on moving averages, potentially missing critical shifts in epidemic behavior.

Purpose of the Study:

• To develop a statistical framework for quantifying temporal changes in incidence dispersion.
• To detect rapid shifts in the dispersion parameter that may signal new epidemic phases.
• To apply this framework to COVID-19 incidence data in US counties to identify changes in epidemic dynamics.

Main Methods:

• Developed a statistical framework to quantify temporal changes in incidence dispersion.
• Applied the method to COVID-19 incidence data from 144 United States counties (January 2020 - March 2023).
• Analyzed temporal trends in dispersion, comparing them with incidence data and identifying shifts.

Main Results:

• Revealed pronounced temporal trends in dispersion not fully explained by incidence alone.
• Observed increased dispersion around the major 2022 COVID-19 case surge.
• Found that highly overdispersed patterns became more frequent later in the time series, replicated across counties.

Conclusions:

• Dispersion shifts can indicate changes in transmission heterogeneity, susceptibility, or reporting.
• The developed method offers a tool for public health officials to anticipate and manage shifts in epidemic regimes.
• Quantifying temporal dispersion provides deeper insights into epidemic drivers beyond simple incidence trends.