Refining reproduction number estimates to account for unobserved generations of infection in emerging epidemics (preprint)

BackgroundEstimating the transmissibility of infectious diseases is key to inform situational awareness and for response planning. Several methods tend to overestimate the basic (R0) and effective (Rt) reproduction numbers during the initial phases of an epidemic. The reasons driving the observed bias are unknown. In this work we explore the impact of incomplete observations and underreporting of the first generations of infections during the initial epidemic phase. MethodsWe propose a debiasing procedure which utilises a linear exponential growth model to infer unobserved initial generations of infections and apply it to EpiEstim. We assess the performance of our adjustment using simulated data, considering different levels of transmissibility and reporting rates. We also apply the proposed correction to SARS-CoV-2 incidence data reported in Italy, Sweden, the United Kingdom and the United States of America. ResultsIn all simulation scenarios, our adjustment outperforms the original EpiEstim method. The proposed correction reduces the systematic bias and the quantification of uncertainty is more precise, as better coverage of the true R0 values is achieved with tighter credible intervals. When applied to real world data, the proposed adjustment produces basic reproduction number estimates which closely match the estimates obtained in other studies while making use of a minimal amount of data. ConclusionsThe proposed adjustment refines the reproduction number estimates obtained with the current EpiEstim implementation by producing improved, more precise estimates earlier than with the original method. This has relevant public health implications. SummaryWe propose a back-imputation procedure tackling the issue of unobserved initial generations of infections to reduce the bias observed in the early R0 and Rt estimates and apply it to EpiEstim using simulated and reported COVID-19 data to evaluate it.

Age-specific mortality and immunity patterns of SARS-CoV-2 infection in 45 countries (preprint)

The number of COVID-19 deaths is often used as a key indicator of SARS-CoV-2 epidemic size. However, heterogeneous burdens in nursing homes and variable reporting of deaths in elderly individuals can hamper comparisons of deaths and the number of infections associated with them across countries. Using age-specific death data from 45 countries, we find that relative differences in the number of deaths by age amongst individuals aged <65 years old are highly consistent across locations. Combining these data with data from 15 seroprevalence surveys we demonstrate how age-specific infection fatality ratios (IFRs) can be used to reconstruct infected population proportions. We find notable heterogeneity in overall IFR estimates as suggested by individual serological studies and observe that for most European countries the reported number of deaths amongst [≥]65s are significantly greater than expected, consistent with high infection attack rates experienced by nursing home populations in Europe. Age-specific COVID-19 death data in younger individuals can provide a robust indicator of population immunity.

A comparative analysis of statistical methods to estimate the reproduction number in emerging epidemics with implications for the current COVID-19 pandemic (preprint)

As the SARS-CoV-2 pandemic continues its rapid global spread, quantification of local transmission patterns has been, and will continue to be, critical for guiding pandemic response. Understanding the accuracy and limitations of statistical methods to estimate the reproduction number, R0, in the context of emerging epidemics is therefore vital to ensure appropriate interpretation of results and the subsequent implications for control efforts. Using simulated epidemic data we assess the performance of 6 commonly-used statistical methods to estimate R0 as they would be applied in a real-time outbreak analysis scenario - fitting to an increasing number of data points over time and with varying levels of random noise in the data. Method comparison was also conducted on empirical outbreak data, using Zika surveillance data from the 2015-2016 epidemic in Latin America and the Caribbean. We find that all methods considered here frequently over-estimate R0 in the early stages of epidemic growth on simulated data, the magnitude of which decreases when fitted to an increasing number of time points. This trend of decreasing bias over time can easily lead to incorrect conclusions about the course of the epidemic or the need for control efforts. We show that true changes in pathogen transmissibility can be difficult to disentangle from changes in methodological accuracy and precision, particularly for data with significant over-dispersion. As localised epidemics of SARS-CoV-2 take hold around the globe, awareness of this trend will be important for appropriately cautious interpretation of results and subsequent guidance for control efforts. Significance StatementIn line with a real-time outbreak analysis we use simulated epidemic data to assess the performance of 6 commonly-used statistical methods to estimate the reproduction number, R0, at different time points during the epidemic growth phase. We find that estimates of R0 are frequently overestimated by these methods in the early stages of epidemic growth, with decreasing bias when fitting to an increasing number of time points. Reductions in R0 estimates obtained at sequential time points during early epidemic growth may reflect increased methodological accuracy rather than reductions in pathogen transmissibility or effectiveness of interventions. As SARS-CoV-2 continues its geographic spread, awareness of this bias will be important for appropriate interpretation of results and subsequent guidance for control efforts.