Exponential growth, high prevalence of SARS-CoV-2, and vaccine effectiveness associated with the Delta variant

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections were rising during early summer 2021 in many countries as a result of the Delta variant. We assessed reverse transcription polymerase chain reaction swab positivity in the Real-time Assessment of Community Transmission–1 (REACT-1) study in England. During June and July 2021, we observed sustained exponential growth with an average doubling time of 25 days, driven by complete replacement of the Alpha variant by Delta and by high prevalence at younger, less-vaccinated ages. Prevalence among unvaccinated people [1.21% (95% credible interval 1.03%, 1.41%)] was three times that among double-vaccinated people [0.40% (95% credible interval 0.34%, 0.48%)]. However, after adjusting for age and other variables, vaccine effectiveness for double-vaccinated people was estimated at between ~50% and ~60% during this period in England. Increased social mixing in the presence of Delta had the potential to generate sustained growth in infections, even at high levels of vaccination.

During 2021, SARS-CoV-2 variant replacement caused a rise in infections and raised concerns about vaccine effectiveness (VE) against infection.

Main and top left: Complete replacement of Alpha by the Delta variant from REACT-1 round 12 to round 13 and weighted prevalence of SARS-CoV-2 infection among a random sample of the population of England ages 5 years and above by self-reported vaccine status. Bottom right: VE adjusted for age, sex, index of multiple deprivation, region, and ethnicity.

Fig. 1.. Temporal trends in prevalence, proportion of positive cases determined to be the Delta variant, and vaccine coverage.

(A) Prevalence of national swab positivity for England estimated using a P-spline for all 13 rounds with central 50% (dark gray) and 95% (light gray) posterior credible intervals. From round 5 of the study onward, weighted observations (black dots) and 95% binomial confidence intervals (vertical lines) are also shown. Note that the period between rounds 7 and 8 (December) of the model is not included, as there were no data available to capture the late December peak of the epidemic. (B) Comparison of the exponential model fit to round 12 and 13 (blue) and the exponential model fit to round 13 only (red). Also shown is the P-spline model fit from (A). Shown here only for rounds 12 and 13 of the study with a log₁₀ y axis. (C) Proportion of Delta against Alpha over time. Points show raw data; error bars denote the 95% confidence interval. Shaded regions show best-fit Bayesian logistic regression models, fit to rounds 10 to 13 (green) and rounds 11 and 12 (orange), with 95% credible interval. (D) Proportion of individuals with known vaccine status who reported being vaccinated with one (light blue) or two (dark blue) doses. Error bars denote 95% binomial confidence intervals.

Fig. 2.. Distribution of N-gene Ct values, by vaccine status, for positive samples obtained from individuals aged 18 to 64 years inclusive.

(A) Distribution of all N-gene Ct values for those who are unvaccinated (red) and those who reported receiving two doses of a vaccine (blue). Also shown are two black dashed lines at N-gene Ct = 33 and 35; these show the threshold values for a sample to be classed as positive, used in sensitivity analyses. (B) Cumulative density of N-gene Ct values using all available data for unvaccinated individuals (red) and individuals who have had two doses of a vaccine (blue). (C) Cumulative density of N-gene Ct values using all data in which N-gene Ct is less than 35 for unvaccinated individuals (red) and individuals who have had two doses of a vaccine (blue). (D) Cumulative density of N-gene Ct values using all data in which N-gene Ct is less than 33 for unvaccinated individuals (red) and individuals who have had two doses of a vaccine (blue). In (B) to (D), red and blue vertical dashed lines show the median value for each distribution.

Fig. 3.. Comparison of daily deaths and hospitalizations to swab positivity as measured by REACT-1.

Daily swab positivity for all 13 rounds of the REACT-1 study (black points with 95% confidence intervals, left y axis) with P-spline estimates for swab positivity (solid black line; shaded area is 95% credible interval). (A) Daily deaths in England (red points, right y axis) and P-spline model estimates for expected daily deaths in England (solid red line, right y axis; shaded area is 95% credible interval). Daily deaths have been shifted by 26 (26, 26) days backward in time along the x axis. The two y axes have been scaled using the best-fit population adjusted scaling parameter 0.059 (0.058, 0.061). (B) Daily hospitalizations in England (blue points, right y axis) and P-spline model estimates for expected daily hospitalizations in England (solid blue line, right y axis; shaded area is 95% credible interval). Daily hospitalizations have been shifted by 20 (19, 20) days backward in time along the x axis. The two y axes have been scaled using the best-fit population adjusted scaling parameter 0.241 (0.236, 0.246).