Thread by @nataliexdean

A headline in the Financial Times today:

"Coronavirus may have infected half of UK population — Oxford study"

Remember that media headlines don't lend themselves to nuanced interpretations. A few words of caution...

ft.com/content/5ff646…

The Oxford model is fit to cumulative numbers of reported #COVID19 deaths in the UK.

A key parameter in the model is ρ, which is the proportion of the population at risk of severe disease. Their prior distributions are set around 0.1% to 1%.

dropbox.com/s/oxmu2rwsnhi9…

There is an intuitive relationship between this parameter and the level of herd immunity in the population.

Fixing the number of observed deaths, if most cases are mild, then we have missed more infections. (Think -- bigger iceberg.)

This relationship is most readily observed in the following figure.

As ρ increases (darker reds), then a higher proportion of the population is still susceptible.

As ρ decreases (yellow), then over 50% of the population (!) has been infected.

A key sentence:

"Overall, these results underscore the dependence of the inferred epidemic curve on the assumed fraction of the population vulnerable to severe disease (ρ) showing significant population level immunity accruing by mid March in the UK as ρ is decreased."

Thus, the results are sensitive to the value of ρ. The authors conclude that what we urgently need is serological testing to better establish ρ.

So please be cautious when interpreting headlines, particularly where those headlines may inform policy decisions. END