The findings in this memo are based on rapid searches in PubMed, LitCovid and MedRxiv, as well as manual searches on websites. Two researchers shared tasks related to study selection and synthesis of results. In the current situation, there is an urgent need for identifying the most important evidence quickly. Hence, we opted for this rapid approach despite an inherent risk of overlooking key evidence or making misguided judgements.
Estimates of fatality rates for COVID-19
We selected data from across a range of surveillance studies and webpages, to obtain and overview at this point of time during the pandemic. From four countries we managed to find estimates of age-stratified case fatality ratio (China, Italy, South Korea and USA). The age-stratified case fatality range is very similar for the younger age groups. For some additional countries we found the case fatality rate for the whole population. These numbers differ across sites, mainly because of the denominator populations in these datasets. Dividing the cumulative number of deaths by the cumulative reported case will underestimate the true fatality ratio, especially early in the growing epidemic. A recent study has correlated for censoring, demography and under-ascertainment to calculate an estimate case fatality rate of 1.38% (1.23-1.53) for China. They estimate age related case fatality rate for <60 years to be 0.318% (0.274-0.378), while for ages 60 years or older the CFR was 6.38% (5.70-7.17). Information from large scale testing and seroprevalence studies should soon give us a clearer picture of the true frequency of infections and thus more accurate assessments of the overall infection fatality rate. These data from USA, Santa Clara, Italy and Germany suggest that many infections are either asymptomatic or mildly symptomatic and thus do not come to medical attention and will not be accounted for in the case fatality rate and suggest infection fatality rates to around 0.5 %.
Factors that can explain the different data in fatality rates across countries
That the observed CFR is higher in some countries, e.g. Italy, than other may be related to three factors: population age, definition of COVID-19 related death and testing strategies. In addition, the delay between testing and mortality in the numbers of early epidemic is a crucial bias. In a pandemic like COVID-19, even advanced health-care systems be overwhelmed by the increased hospitalisation and will influence the case fatality rate in the near future. Information from large scale testing and seroprevalence studies should soon give us a more accurate assessments of the overall infection fatality rate