Spanish excess mortality statistics provide 1.2% to 2.1% range for COVID-19 infection fatality ratio

José Maria Martin-Olalla has provided an as yet non-peer reviewed preprint containing estimates of death rates from coronavirus in Spain earlier this year.

The base rate for deaths is estimated from country statistics since the beginning of the century on a weekly basis.

It is possible to compare death rates in excess of the normal expected value and estimate an upper bound for the mortality from coronavirus infection.

Reliable estimates for the extent of infection in Spain are also available from comprehensive antibody testing of the population.

This makes it possible to estimate an upper bound and the range for the case fatality ratio. In Spain it seems fatality rates are very high partly reflecting an ageing population.

The method used by José reveals a significant fraction of total deaths are being missed by the standard case identification methods in the health system.

Very approximately only half the coronavirus associated deaths are being detected from diagnosed causes of death.

Below is an abstract of the paper, and 2 key figures.

They illustrate how different the coronavirus outbreak is from an influenza epidemic in terms of numbers of deaths, even after stringent lockdown procedures to stop the outbreak before it is completed. They also illustrate the significant fraction of unaccounted deaths revealed by the excess deaths calculations.

Even as stringent lockdown procedures are implemented, deaths from the coronavirus pandemic are much worse than a bad influenza season

Careful use of expected baseline mortality estimates allows the numbers of unattributed deaths during a pandemic to be evaluated. These unattributed deaths are likely to be caused by coronavirus and allow an upper bound estimate for case fatality rate of 2.1% to be inferred. This compares with an estimate of 1.2% medically diagnosed coronavirus deaths in the Spanish population.