A demographic scaling model for estimating the total number of COVID-19 infections.

Christina Bohk-Ewald,Mikko Myrskylä,Christian Dudel

doi:10.1093/ije/dyaa198

Abstract

BackgroundUnderstanding how widely COVID-19 has spread is critical information for monitoring the pandemic. The actual number of infections potentially exceeds the number of confirmed cases.DevelopmentWe develop a demographic scaling model to estimate COVID-19 infections, based on minimal data requirements: COVID-19-related deaths, infection fatality rates (IFRs), and life tables. As many countries lack IFR estimates, we scale them from a reference country based on remaining lifetime to better match the context in a target population with respect to age structure, health conditions and medical services. We introduce formulas to account for bias in input data and provide a heuristic to assess whether local seroprevalence estimates are representative for the total population.ApplicationAcross 10 countries with most reported COVID-19 deaths as of 23 July 2020, the number of infections is estimated to be three [95% prediction interval: 2–8] times the number of confirmed cases. Cross-country variation is high. The estimated number of infections is 5.3 million for the USA, 1.8 million for the UK, 1.4 million for France, and 0.4 million for Peru, or more than one, six, seven and more than one times the number of confirmed cases, respectively. Our central prevalence estimates for entire countries are markedly lower than most others based on local seroprevalence studies.ConclusionsThe national infection estimates indicate that the pandemic is far more widespread than the numbers of confirmed cases suggest. Some local seroprevalence estimates largely deviate from their corresponding national mean and are unlikely to be representative for the total population.

Highlights

The total number of COVID-19 infections is a key indicator for understanding the spread of the pandemic
We introduce a demographic scaling model to estimate COVID-19 infections using an broadly applicable approach that is based on minimal data requirements: COVID-19 related deaths, infection fatality rates (IFRs), and life tables
This indirect approach can be applied in many contexts, as it requires only a small amount of input data: namely, the number of COVID-19-related deaths for the population of interest; and the age-specific infection fatality rates (IFR; deaths over infections) from a reference population, scaled to match the target population based on life tables

Summary

Introduction

The total number of COVID-19 infections is a key indicator for understanding the spread of the pandemic. We use the scaling approach (1) to estimate the total number and prevalence of COVID-19 infections for the 10 countries that have reported the most deaths caused by COVID-19 as of May 13, 2020 and (2) to assess the validity of recent local seroprevalence studies for the U.S, Italy, and Germany. We disaggregate COVID-19 total deaths into the same 10-year age groups using the global average pattern over age that we estimated based on the data provided by Dudel and colleagues.[8] Details about the model and additional findings based on, e.g., unadjusted IFRs, adjusted IFRs from a European reference country as reported by Salje and colleagues,[21] and average time to death from COVID-19 are given in SI appendix 1 through 6

Results

Discussion

16. Epidemiology Group of the New Coronavirus Pneumonia Emergency Response