The changing epidemiology of SARS-CoV-2

Abstract

We have come a long way since the start of the COVID-19 pandemic-from hoarding toilet paper and wiping down groceries to sending our children back to school and vaccinating billions. Over this period, the global community of epidemiologists and evolutionary biologists has also come a long way in understanding the complex and changing dynamics of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus that causes COVID-19. In this Review, we retrace our steps through the questions that this community faced as the pandemic unfolded. We focus on the key roles that mathematical modeling and quantitative analyses of empirical data have played in allowing us to address these questions and ultimately to better understand and control the pandemic.

Fig. 1.. Timeline of changing epidemiological questions asked as the SARS-CoV-2 pandemic unfolded.

Confirmed infections are shown by continent as well as on a worldwide scale. Gray bars indicate the time periods discussed in this Review (the emerging pandemic, flattening the curve, riding out the waves, and vaccines and variants).

Fig. 2.. Estimation of the basic reproduction number R₀ from case data and detailed outbreak data.

(A) Exponential growth models are fit to observed case data to estimate the speed at which a virus spreads through a population. For SARS-CoV-2 spread in Wuhan, early estimates of the exponential growth rate r fell in the range of 0.10 to 0.20 per day (3), yielding epidemic doubling times of ~3.5 to 7 days. (B) Outbreak data are used to estimate the viral serial interval, defined as the time between symptom onset of an index case and symptom onset of the index case’s contacts. This serial interval is often used as an approximation for the viral generation interval. For SARS-CoV-2, an early Wuhan estimate of the mean serial interval was 7.5 days (86) (distribution in red). An example of a generation interval distribution with a smaller mean is shown in light orange. (C) The basic reproduction number R₀ can be calculated from the exponential growth rate r and the distribution of the generation interval using an equation derived from demographic analyses (87). Red and light orange curves, corresponding to the distributions in (B), show how the R₀ estimates depend on the generation interval. The transmission chain at the top illustrates an outbreak with an R₀ = 3 pathogen.

Fig. 3.. Coronavirus sequence data have informed epidemiological understanding of SARS-CoV-2.

(A) The frequencies of SARS-CoV-2 variants of concern over time. The first indication that SAR-COV-2 was adapting to humans was the population-level replacement of the 614D allele with the 614G allele in early 2020. Variant frequencies on the y axis are calculated on the basis of SARS-CoV-2 sequence data deposited in GISAID (Global Initiative for Sharing Avian Influenza Data). Only major, globally circulating variant lineages are shown. (B) An example of viral sequencing that has provided evidence of reinfection with SARS-CoV-2. Reproduced with permission from (48). Samples from a patient in the context of circulating viral genetic variation indicate that the observed secondary infection is not a reactivation of a latent infection but is instead a reinfection. Schematic on the right shows the substitutions present in the primary and secondary infection viral samples. ORF, open reading frame. (C) A phylogeny of seasonal human coronavirus OC43 (lineage A). Phylogenetic analysis points toward antigenic evolution in this viral population. Reproduced with permission from (50). (D) A phylogeny inferred from SARS-CoV-2 sequence data showing the evolutionary relationships between the variant lineages included in (A).

Fig. 4.. Vaccine, host, and viral factors that affect vaccine efficacy.

Each of the factors shown were investigated as potential moderators of vaccine efficacy in clinical trials and postintroduction observational studies. Scenario modeling projected the population-level effects of vaccination by incorporating data-driven assumptions on how vaccination affects susceptibility to infection, disease, and severe outcomes like hospitalization and death and how these effects differed by host factors, such as age. A key determinant of population impact of vaccination is the effect on transmission, which is a combination of infection-blocking and transmission-reducing effects of vaccines. As data emerged, mainly from observational studies, assumptions about the infectiousness of vaccine breakthrough cases were incorporated into models. With the evolution of Delta and then Omicron, variant-specific vaccine efficacy and the potential for viral immune escape became critical quantities to understand. With this array of vaccine, host, and viral data on the effects of vaccination on vaccine recipients, epidemiological models could project both the direct and indirect effects of vaccination on the population-level spread of SARS-CoV-2.