Estimating COVID-19 Virus Prevalence from Records of Testing Rate and Test Positivity

Abstract

Introduction: PCR testing for COVID-19 is not done at random but selectively on suspected cases. This paper presents a method to estimate a “genuine virus prevalence” by quantifying and removing the bias related to selective testing. Methods: The data used was from nine (9) neighbouring countries in Western Europe that recorded similar epidemic trends despite differences in testing rate. Regression analysis was used to establish a relationship of declining test positivity with increased testing rate. By extrapolating this trend to an “infinitely complete” testing rate, an unbiased test positivity or “genuine virus prevalence” was computed. Via pairing of “genuine virus prevalence” with excess-deaths, a “genuine infection fatality rate” (IFR) was also derived. Results: Peak levels of “genuine virus prevalence” were around 0.5 to 2% during the 1st epidemic “wave” (week 10 to week 20) and are approaching similar levels in the ongoing 2nd “wave” (week 34 onward). “Genuine virus prevalence” estimates are relatively close to reported seroprevalence in the studied countries with a correlation coefficient of 0.54. “Genuine” IFR is found comparable to closed-community model IFR. Finally, results of community mass-testing in Slovakia are within the estimated range of “genuine virus prevalence”. Conclusions: Estimates of “genuine virus prevalence” benchmark favourably to other indications of virus prevalence suggesting the estimation method is robust and potentially deployable beyond this initial dataset of countries. “Genuine virus prevalence” curves suggest that during the 1st epidemic “wave”, curve flattening and waning happened at very modest levels of infection spread, either naturally or facilitated by government measures.