Fig. 2: Serological assay values and population rates derived from prevalence.

a Distribution of quantitative values for the LuLISA N, LuLISA S and pseudo-neutralisation assays. Readings in relative light units (RLU in logarithmic scale) are presented for LuLISA N (LN), LuLISA S (LS) and pseudo-neutralisation (PN) assays on sera from pre-pandemic (pp) patients, confirmed cases of COVID-19 (cc), and sera sampled during three collection periods 15/3 (9–15 March 2020), 12/4 (6–12 April 2020) and 17/5 (11–17 May 2020). Positivity thresholds are indicated by horizontal dotted lines, values above the threshold indicate positivity for the LuLISA tests, whereas values below the threshold indicate positivity for the pseudo-neutralisation test. b Weighted correlation between estimated prevalence of SARS-CoV-2 antibodies and reported mortality rates by region. Mortality rates per 100,000 were obtained as region-specific number of deaths attributed to COVID-19 as of 29 May 2020 divided by population size. The date to account for deaths was calculated assuming that individuals with detectable antibodies at sampling time (midpoint of interval from 11 to 17 May 2020) could have been infected at minimum 15 days previously and were susceptible of dying from their infection up to 30 days post-infection. Pearson correlation coefficient (r) was weighted by standard error of seroprevalence estimates. Circle sizes reflect this weighting. Regions are coded as in Fig. 1a. c Infection fatality and infection hospitalisation rates by age. Rates are in logarithmic scale. Infection fatality rate (IFR) is estimated as the cumulative number of deaths per 100 estimated infections stratified by age. Based on available data both from French COVID-19 surveillance and published literature, we considered a lag of 20 days for both time between infection and death, and between infection and seropositivity. Infection hospitalisation rate (IHR) is calculated as the cumulative number of patients hospitalised for COVID-19 per 100 estimated infections stratified by age. We consider a time lag from infection to hospitalisation of 10 days (see “Methods”).