Bayesian back-calculation and nowcasting for line list data during the COVID-19 pandemic

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Pandemic surveillance relies on timely data, but aggregated case reports have delays causing misleading inferences.
• Line list data offers individual case details crucial for modeling reporting delays, unlike aggregated data.
• Existing methods struggle with line list data due to frequently missing symptom onset dates.

Purpose of the Study:

• To develop a novel Bayesian approach for analyzing line list data in pandemics.
• To accurately estimate epidemic curves and reproduction numbers despite missing symptom onset data.
• To provide a robust method for infectious disease surveillance and response.

Main Methods:

• Developed a Bayesian approach integrating imputation and estimation for line list data.
• Dynamically models reporting delays, accounting for missing symptom onset dates.
• Assessed robustness to changes in reporting delay distributions and maximum reporting delay.

Main Results:

• Accurately estimated epidemic curves and instantaneous reproduction numbers, even with extensive missing symptom onset dates.
• Demonstrated robustness to deviations from model assumptions.
• Applied to COVID-19 data, yielding reproduction number estimates better aligned with control measures than traditional methods.

Conclusions:

• The proposed Bayesian approach enhances the accuracy of epidemiological modeling using line list data.
• This method offers a more reliable tool for real-time pandemic surveillance and informing public health interventions.
• Improved estimation of reproduction numbers provides better insights into disease transmission dynamics and intervention effectiveness.