Robust trend estimation for COVID-19 in Brazil.

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Estimating COVID-19 pandemic patterns is complex due to spatial-temporal heterogeneity.
• Health department data collection methods introduce measurement errors, affecting outcome analysis.

Purpose of the Study:

• To develop methods for estimating COVID-19 trends while accounting for measurement error.
• To compare proposed models with empirical moving average analyses.

Main Methods:

• Bayesian time series and spatio-temporal models for counting processes with latent components.
• Application of time series decomposition to COVID-19 deaths in Brazil, São Paulo, and Amazonas.
• Spatio-temporal analysis of COVID-19 deaths at the Brazil state level using global and regional components.

Main Results:

• The proposed Bayesian models effectively estimate underlying COVID-19 trends, controlling for data inaccuracies.
• Comparison with moving averages demonstrates the superiority of the proposed methods in capturing true patterns.
• Spatio-temporal analysis reveals distinct regional patterns in COVID-19 mortality across Brazilian states.

Conclusions:

• Bayesian time series and spatio-temporal models offer robust tools for analyzing epidemic data with inherent measurement errors.
• Accurate estimation of COVID-19 trends is crucial for effective public health interventions and pandemic response.
• The study highlights the importance of addressing data quality issues in epidemiological research.