Incorporating temporal distribution of population-level viral load enables real-time estimation of COVID-19 transmissibility

doi:10.21203/rs.3.rs-841953/v1

Download PDF

Article

Incorporating temporal distribution of population-level viral load enables real-time estimation of COVID-19 transmissibility

https://doi.org/10.21203/rs.3.rs-841953/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 03 Mar, 2022

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

Many locations around the world have used real-time estimates of the time-varying effective reproductive number (\({R}_{t}\)) of COVID-19 to provide evidence of transmission intensity to inform control strategies. Estimates of \({R}_{t}\) are typically based on statistical models applied to case counts and typically suffer lags of more than a week because of the incubation period and reporting delays. Noting that viral loads tend to decline over time since illness onset, analysis of the distribution of viral loads among confirmed cases can provide insights into epidemic trajectory. Here, we analyzed viral load data on confirmed cases during two local epidemics in Hong Kong, identifying a strong correlation between temporal changes in the distribution of viral loads (measured by cycle threshold values) and estimates of \({R}_{t}\) based on case counts. We demonstrate that cycle threshold values could be used to improve real-time \({R}_{t}\) estimation, enabling more timely tracking of epidemic dynamics.

Immunology

Infectious Diseases

COVID-19

viral load

epidemic dynamics

Yes there is potential Competing Interest. BJC consults for AstraZeneca, GSK, Moderna, Roche and Sanofi Pasteur. The authors report no other potential conflicts of interest.

supplementary23aug.docx
FigS1.pdf
Supplementary Figure 1. Daily distribution of Ct values by sampling date over the study period. Dark grey vertical lines and red dots showed the interquartile and median of Ct values sampled per day, while light grey vertical lines represented either the minimum and maximum of all Ct values or the smallest and largest observed Ct values that are within the distance of 1.5 times the interquartile range of all Ct values on that day (i.e., whiskers). Orange lines and shaded areas represent the daily average and the 95% CI of Ct values estimated from a GAM as in Eq. 2 (same as Fig. 1).
FigS2.pdf
Supplementary Figure 2. Bi-weekly distribution of Ct values by date of sampling and of onset during the third (a) and fourth (b-c) wave of COVID-19 in Hong Kong. The Ct value shown by date of sampling was the actual value as sampled, whereas the Ct value shown by date of onset was the extrapolated value estimated from Eq. 3. The distribution of Ct values at sampling and at onset was illustrated in green and orange if during periods with increasing epidemic and in cyan and pink if during periods with decreasing epidemic. The skewness of Ct distribution at sampling (b_s) and at onset (b_o) over the biweekly period was shown at top-right of each panel with corresponding colors, while the exact time period was specified above each panel.
FigS3.pdf
Supplementary Figure 3. Consistent Ct-based R_t estimates before and after adjusting for the age distribution of sampled cases. a-b, Comparison of daily mean Ct (a) and Ct skewness (b) under various incidence-based R_t intervals during July 2020 - August 2020 (wave 3) and November 2020 - March 2021 (wave 4). Boxes show the interquartile and median of the corresponding Ct distribution derived from all samples (colored) and from samples of adult cases only (white) respectively. c) Comparison of Ct-based R_t estimates before and after adjustment of age. Pink lines and shaded areas show the point estimates and 95% prediction intervals of Ct-based R_t estimated from the main model (Eq. 6), while blue dots and vertical lines show the point estimates and 95% prediction intervals of Ct-based R_t estimated after adjusting for the mean age of daily sampled cases (as in Eq. 7).
FigS4.pdf
Supplementary Figure 4. Validation of the Ct-based model using an alternative training period. a-b, Comparison of incidence-based and Ct-based R_t over the alternative training (November-December 2020, a) and testing (July-August 2020, b) periods. Black lines and shaded areas indicate the mean and 95% CrIs for incidence-based R_t, while colored dots and vertical lines represent the point estimates and 95% prediction intervals of Ct-based R_t. c) Consistency between the incidence-based and Ct-based R_t over the alternative training (purple) and testing (pink) periods.
FigS5.pdf
Supplementary Figure 5. Performance of models fitted over various length and timing of training periods. a-b, Performance of models fitted over different training periods with various durations and starting dates, over a) July-September 2020 and b) November 2020-January 2021. Color lines show the performance of models (measured by adjusted R square) fitted over training periods with a duration of 30-60 days respectively, with the starting date of the training period specified in the x-axis. The grey dashed line in the background referred to adjusted R square being 0.7, as a reference for comparison. c-d, Time period covered by the model with the largest adjusted R square over c) July-September 2020 and d) November 2020-January 2021. Grey and orange bars represent the number of laboratory-confirmed cases (by date of reporting) and of sampled collections (by date of sampling) respectively. Time periods used for fitting the model with the largest adjusted R square among all models fitted over that certain length of training period was indicated by corresponding colored backgrounds (same color legend as in panel a).

Download PDF

Journal Publication

published 03 Mar, 2022

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

Incorporating temporal distribution of population-level viral load enables real-time estimation of COVID-19 transmissibility

Status:

Journal Publication

Version 1

Abstract

Figures

Full Text

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1