COVID-19: Nowcasting Reproduction Factors Using Biased Case Testing Data

TL;DR

This paper presents a method to estimate the COVID-19 reproduction number $R$ using biased case testing data, providing earlier insights into epidemic dynamics crucial for timely policy decisions.

Contribution

It introduces a novel estimator for $R$ based on testing data, enabling earlier and potentially more accurate epidemic assessment compared to traditional methods.

Findings

Estimated $R$ for Scotland on 20 May 2020 as 0.74

95% confidence interval for $R$ is [0.48, 0.86]

Method demonstrates utility with real testing data

Abstract

Timely estimation of the current value for COVID-19 reproduction factor $R$ has become a key aim of efforts to inform management strategies. $R$ is an important metric used by policy-makers in setting mitigation levels and is also important for accurate modelling of epidemic progression. This brief paper introduces a method for estimating $R$ from biased case testing data. Using testing data, rather than hospitalisation or death data, provides a much earlier metric along the symptomatic progression scale. This can be hugely important when fighting the exponential nature of an epidemic. We develop a practical estimator and apply it to Scottish case testing data to infer a current (20 May 2020) $R$ value of $0.74$ with $95\%$ confidence interval $[0.48 - 0.86]$.