Austria did two representative COVID-19 prevalence studies that so far gained little international attention despite their unique strong points.
The first was the SORA study, carried out on behalf of the Austrian ministry of science by the SORA Institute with the intention to investigate the number of unrecognized cases slipping through official surveillance, testing and contact tracking
https://www.sora.at/nc/news-presse/news/news-einzelansicht/news/covid-19-praevalenz-1006.html
English Summary: https://www.sora.at/fileadmin/downloads/projekte/Austria_Spread_of_SARS-CoV-2_Study_Report.pdf
In the SORA study 1544 PCR tests (cobas SARS-CoV-2) were carried out in a representative sample of the Austrian population in the period April 1-6 with 6 "positive" results. The authors concluded that 0.33% (or 28500 individuals) of the Austrian population was COVID-19 positive by that time - with a fairly wide 95% CI of 0.12-0.77% .Compared to that on April 5th the "official" number of "active" cases was 8849 (JHU CSSE https://github.com/CSSEGISandData/COVID-19/
) so it appears the "real" number of cases was approximately 3 fold of
the "known" cases.
The authors did not attempt to calculate the mortality from this data but a very adhoc calculation shows that the IFR (mortality) could be somewhere around 1%.
The strengths of this study:
* prevalence was tested using "gold standard" PCR test
* it was carried out near (probably shortly after) the first (and hopefully last) peak of the pandemics in Austria when PCR tests allow good estimate of prevalence
* relatively homogeneous population in a small country, meaning fairly representative results.
Conclusion: the study shows that only a very small part of Austrian population (probably in the order of 0.44%) was ever infected with the SARS-CoV-2 virus until April 2020.
The second was performed by "Statistik Austria" between April 21th to 24th.
This time not only representative PCR testing was done, but a special focus was put on several known hotspots where SARS-CoV-2-Antibody tests were performed.
https://www.statistik.at/web_de/frageboegen/private_haushalte/covid19/index.html
http://www.statistik.at/web_de/presse/123051.html
The results in summary:
* in the time of April 21th to 24th less than 0.15% of Austrian residents were acutely infected with SARS-CoV-2, iow less than 10823 individuals.
* higher prevalence of active infections of up to 0.75% was found in known hotspots
* in the 27 most severely affected places approximately 4.71% of the population tested positive for SARS-CoV-2antibodies
Comment:
The results of the second study seem the confirm the data from the first study. Overall Austria is very far away from achieving "herd immunity" (if it exists).
Mortality (IFR) judged by the data from Austria seems to be around 1%.For every known case there were at least 2 undetected cases which may seem like a huge problem for epidemic control. Despite that Austria is currently very successful at limiting the epidemic.
These are based on PCR tests, and not seroprevalence antibody tests. Rest is mathematical extrapolations. Isn't that vastly inaccurate?
ReplyDeleteFirst study(SORA) was PCR test only, the second PCR for the representative sample and antibody/seroprevalence for selected hotspots. Both were pretty early,the first representative nationwide studies known to me and at a time when antibody studies were around 85% sensitivity and 90% specificity compared to PCR 96% sensitivity and 99.6% specificity.
DeletePCR tests are considered much more accurate than antibody tests, seroprevalence testing has the advantage that it can detect past infections (with a much lower sensitivity and specificity) whereas PCR only detects current infections.
The first study was done close to the peak of the first wave - its main significance is twofold:
* it allows an estimate of the real number of infected in comparison with the known cases. The authors gave a "wide" confidence interval for this, nevertheless the whole CI fits well way bellow 1% of total population
* it did confirm, that while there are many undetected cases, the number of undetected cases isn't astronomically high (but approximately 3-4 times in Austria). Prior to the SORA study it was rather unclear how many people had an asymptomatic infection
The second study was done long after the first peak of the epidemics so PCR testing was nearly useless but the antibody testing in selected hotspots confirmed that the total number of population that passed the infection was indeed relatively small.
Since than the prevalence study on the legendary Ischgl hotspot was published where iirc over 40% of the population were infected.
Yes, my mathematical extrapolations are inaccurate. At the time of writing known data suggested that infected patients would die within 2 weeks of infection on average. Today I think it is at least a week longer, which would mean a higher mortality than what I have calculated. Additionally there are uncertainties about the timelag to detect and report infections and to report deaths.