Nowcasting the COVID-19 pandemic in Bavaria

Area of Science:

• Epidemiology
• Biostatistics
• Public Health Surveillance

Background:

• Accurate real-time epidemic surveillance requires timely data on new cases.
• Reporting delays between disease onset and case reporting hinder immediate understanding of epidemic dynamics.
• Nowcasting offers a method to adjust reported case counts for unreported events.

Purpose of the Study:

• To apply and adapt nowcasting techniques for real-time COVID-19 surveillance in Bavaria.
• To develop a hierarchical Bayesian model that accounts for time-varying reporting delays and reporting day-of-week effects.
• To estimate the time-varying effective case reproduction number ($$) using nowcast predictions.

Main Methods:

• A hierarchical Bayesian model was developed to nowcast daily COVID-19 cases.
• The model incorporates changes in reporting delay distributions over time and weekday effects.
• The effective reproduction number was estimated based on the nowcast predictions.

Main Results:

• The study presents a novel application of nowcasting to COVID-19 data in Bavaria.
• Methodological details, results from current pandemic data, and model evaluations using synthetic and retrospective data are provided.
• Nowcasting results are regularly reported to the Bavarian health authority and published online.

Conclusions:

• The developed nowcasting approach provides adjusted daily COVID-19 case counts, improving real-time situational awareness.
• The method effectively estimates the reproduction number, aiding epidemic dynamics assessment.
• The approach is adaptable to different datasets and is made publicly available.