Background: The coronavirus disease (COVID-19) pandemic is slowing down, and countries are discussing whether preventive measures have remained effective or not. This study aimed to investigate a particular property of the trend of COVID-19 that existed and if its variants of concern were cointegrated, determining its possible transformation into an endemic.
Methods: Biweekly expected new cases by variants of COVID-19 for 48 countries from 02 May 2020 to 29 August 2022 were acquired from the GISAID database. While the case series was tested for homoscedasticity with the Breusch-Pagan test, seasonal decomposition was used to obtain a trend component of the biweekly global new case series. The percentage change of trend was then tested for zero-mean symmetry with the one-sample Wilcoxon signed rank test and zero-mean stationarity with the augmented Dickey-Fuller test to confirm a random COVID trend globally. Vector error correction models with the same seasonal adjustment were regressed to obtain a variant-cointegrated series for each country. They were tested by the augmented Dickey-Fuller test for stationarity to confirm a constant long-term stochastic intervariant interaction within the country.
Results: The trend series of seasonality-adjusted global COVID-19 new cases was found to be heteroscedastic (p = 0.002), while its rate of change was indeterministic (p = 0.052) and stationary (p = 0.024). Seasonal cointegration relationships between expected new case series by variants were found in 37 out of 48 countries (p < 0.05), reflecting a constant long-term stochastic trend in new case numbers contributed from different variants of concern within most countries.
Conclusion: Our results indicated that the new case long-term trends were random on a global scale and stable within most countries; therefore, the virus was unlikely to be eliminated but containable. Policymakers are currently in the process of adapting to the transformation of the pandemic into an endemic.
Keywords: COVID-19; global; mutation; strategy; time-series; variant of concern (VOC).
Copyright © 2023 Chu, Szeto, Wong and Chung.