SARS-CoV-2 clade dynamics and their associations with hospitalisations during the first two years of the COVID-19 pandemic

Abstract

Background: The COVID-19 pandemic was characterised by rapid waves of disease, carried by the emergence of new and more infectious SARS-CoV-2 virus variants. How the pandemic unfolded in various locations during its first two years has yet to be sufficiently covered. To this end, here we are looking at the circulating SARS-CoV-2 variants, their diversity, and hospitalisation rates in Estonia in the period from March 2000 to March 2022.

Methods: We sequenced a total of 27,550 SARS-CoV-2 samples in Estonia between March 2020 and March 2022. High-quality sequences were genotyped and assigned to Nextstrain clades and Pango lineages. We used regression analysis to determine the dynamics of lineage diversity and the probability of clade-specific hospitalisation stratified by age and sex.

Results: We successfully sequenced a total of 25,375 SARS-CoV-2 genomes (or 92%), identifying 19 Nextstrain clades and 199 Pango lineages. In 2020 the most prevalent clades were 20B and 20A. The various subsequent waves of infection were driven by 20I (Alpha), 21J (Delta) and Omicron clades 21K and 21L. Lineage diversity via the Shannon index was at its highest during the Delta wave. About 3% of sequenced SARS-CoV-2 samples came from hospitalised individuals. Hospitalisation increased markedly with age in the over-forties, and was negligible in the under-forties. Vaccination decreased the odds of hospitalisation in over-forties. The effect of vaccination on hospitalisation rates was strongly dependent upon age but was clade-independent. People who were infected with Omicron clades had a lower hospitalisation likelihood in age groups of forty and over than was the case with pre-Omicron clades regardless of vaccination status.

Conclusions: COVID-19 disease waves in Estonia were driven by the Alpha, Delta, and Omicron clades. Omicron clades were associated with a substantially lower hospitalisation probability than pre-Omicron clades. The protective effect of vaccination in reducing hospitalisation likelihood was independent of the involved clade.

Fig 1. The workflow for the study population.

Epidemiological data availability means that, at least, the date of sampling and age or sex are present. The use of ‘ECDC’ refers to the European Centre for Disease Prevention and Control in Sweden.

Fig 2. Testing and sequencing SARS-CoV-2 in Estonia.

The bars show the number of tests which were carried out (left axis) and the line denotes the percentage of positive samples sequenced (right axis). Vertical lines denote the duration of waves, defined by the prevalence of Nextstrain clade(s). The wave start and end were defined as a week in which the lower or upper bound respectively of the 95% confidence interval of clade prevalence, as obtained through the Clopper-Pearson method, crossed the 50% threshold. SARS-CoV-2 test results were downloaded from the Estonian COVID-19 open-data portal (https://opendata.digilugu.ee; last accessed 5 March 2024).

Fig 3. Waves of SARS-CoV-2 in Estonia.

(A) the prevalence of Nextstrain clades; (B) the prevalence of VOCs. Full-length sequencing was not applied to samples collected between W47 to W52 in 2020.

Fig 4. The diversity of SARS-CoV-2 lineages in domestic and travel-related cases in Estonia.

(A) Shannon diversity index. Points denote individual weekly observations. The line denotes the autoregressive model which has been fitted to the data, and the shaded ribbon denotes a 95% credible interval, N = 14,341; (B) the effect-size of the Shannon index, imported cases compared to domestic cases. The line denotes the effect size as derived from the model fit which is shown in Panel A, and the shaded ribbon denotes a 95% credible interval. The model summary is presented in Table 3 in the S2 Appendix.

Fig 5. The hospitalisation of SARS-CoV-2-positive individuals in Estonia.

(A) the probability of hospitalisation is associated with vaccination status, age, and clade. Posterior summaries of the conditional effect of the clade, vaccination status, and age in logistic regression when adjusted for sex, the number of weekly cases, and population vaccination coverage, N = 23,456; (B) log the odds ratio of hospitalisation in vaccinated individuals when compared to unvaccinated SARS-CoV-2-positive individuals; (C) log the odds ratio in terms of the hospitalisation of unvaccinated SARS-CoV-2-positive individuals who have been infected with Omicron 21K (upper panels) or 21L (lower panels) when compared to individuals who have been infected with other common SARS-CoV-2 clades. The line denotes the model’s best estimate, and the ribbon denotes a 95% credible interval. The solid line denotes the span of ages which have been observed for each clade, while the dashed line denotes unobserved ages implied by the underlying model. The dotted line denotes log odds = 0. Panels B and C are based on the same model as for A, while the model summary is presented in Table 4 in the S2 Appendix.