Seasonal and periodic patterns in US COVID-19 mortality using the Variable Bandpass Periodic Block Bootstrap

Abstract

Since the emergence of the SARS-CoV-2 virus, research into the existence, extent, and pattern of seasonality has been of the highest importance for public health preparation. This study uses a novel bandpass bootstrap approach called the Variable Bandpass Periodic Block Bootstrap to investigate the periodically correlated components including seasonality within US COVID-19 mortality. Bootstrapping to produce confidence intervals for periodic characteristics such as the seasonal mean requires preservation of the periodically correlated component’s correlation structure during resampling. While other existing bootstrap methods can preserve the periodically correlated component correlation structure, filtration of that periodically correlated component’s frequency from interference is critical to bootstrap the periodically correlated component’s characteristics accurately and efficiently. The Variable Bandpass Periodic Block Bootstrap filters the periodically correlated time series to reduce interference from other components such as noise. This greatly reduces bootstrapped confidence interval size and outperforms the statistical power and accuracy of other methods when estimating the periodic mean sampling distribution. Variable Bandpass Periodic Block Bootstrap analysis of US COVID-19 mortality periodically correlated components is provided and compared against alternative bootstrapping methods. Results show that both methods find a significant seasonal component, but the Variable Bandpass Periodic Block Bootstrap produces smaller confidence intervals and only the Variable Bandpass Periodic Block Bootstrap found significant components at the second through the fifth harmonics of the seasonal component, as well as weekly component. This crucial evidence supporting the presence of a seasonal pattern and existence of additional periodically correlated components, their timing, and confidence intervals for their effect which will aid prediction and preparation for future COVID-19 responses.