A large Istat survey (64k) of the Italian population finds 2.5% seropositivity rate nationwide. Care homes residents not sampled and overall IFR based on clinically confirmed deaths (35k) comes to 2.3%. A few thoughts 1/n

Test kit characteristics: The survey was done over a two month period with the Abbott N-protein IgG kit. Issues with test kits primed to the NP-protein is discussed here and suggest poor ability to detect previous SARS-COV-2 infection >200 days POS. https://www.medrxiv.org/content/10.1101/2020.07.16.20155663v2.full.pdf">https://www.medrxiv.org/content/1... 2/n

A sero kit targeting the spike protein esp. the spike trimer assays being more stable and providing much better sensitivity. Several studies have shown very stable (>95%) antibodies in longitudinal studies (>3 mths) when the spike trimer is used https://www.medrxiv.org/content/10.1101/2020.08.01.20166553v1">https://www.medrxiv.org/content/1... 3/n

However, 41.7% of familial contacts of SARS-COV-2 confirmed cases were seropositive providing lower bounds for test kit sensitivity (~50%). True serop+ve rate is probably around 4-5% nationwide. This implies a crude 1.15-1.5% IFR. 4/n

The crude IFR estimates (not accounting for excess deaths) do not seem improbable given the unique epidemic situation in Italy in Late January to March. @DiseaseEcology gives a good summary of the data 5/n https://twitter.com/DiseaseEcology/status/1290754158438387713">https://twitter.com/DiseaseEc...

However, as pointed out earlier in the thread and by others, they are several issues with the survey overall. For instance participation rate may be skewing the data substantially https://twitter.com/WesPegden/status/1290757276547125248">https://twitter.com/WesPegden... 6/n

Overall, a more detailed report on the seropositive samples probably broken down by space-time should provide more data for deeper insights but it would be best if the repeat studies employ the use of a more sensitive kit to allow for more precise estimations. 7/n