Objective: This mixed design synthesis aimed to estimate the infection fatality rate (IFR) of
Coronavirus Disease 2019 (COVID-19) in community-dwelling elderly populations and other age
groups from seroprevalence studies. Protocol: https://osf.io/47cgb.
Methods and analyses: Eligible were seroprevalence studies done in 2020 and identified by any of
four existing systematic reviews; with ≥1000 participants aged ≥70 years that presented
seroprevalence in elderly people; that aimed to generate samples reflecting the general population;
and whose location had available data on cumulative COVID-19 deaths in elderly (primary cutoff
≥70 years; ≥65 or ≥60 also eligible). We extracted the most fully adjusted (if unavailable,
unadjusted) seroprevalence estimates. We also extracted age- and residence-stratified cumulative
COVID-19 deaths (until 1 week after the seroprevalence sampling midpoint) from official reports,
and population statistics, to calculate IFRs corrected for unmeasured antibody types. Sample sizeweighted IFRs were estimated for countries with multiple estimates. Secondary analyses examined
data on younger age strata from the same studies.
Results: Twenty-five seroprevalence surveys representing 14 countries were included. Across all
countries, the median IFR in community-dwelling elderly and elderly overall was 2.9% (range 0.2%-
6.9%) and 4.9% (range 0.2%-16.8%) without accounting for seroreversion (2.4% and 4.0%,
respectively, accounting for 5% monthly seroreversion). Multiple sensitivity analyses yielded similar
results. IFR was higher with larger proportions of people >85 years. Younger age strata had low IFR
values (median 0.0013%, 0.0088%, 0.021%, 0.042%, 0.14%, and 0.65%, at 0-19, 20-29, 30-39, 40-
49, 50-59, and 60-69 years even without accounting for seroreversion).
Author(s):Cathrine Axfors, John P A Ioannidis
Publication Date: 23 Dec 2021, accessed 17 Jan 2022
Publication Site: medrXiV