The dramatic (and welcome) shift in UK Gov policy has been accompanied by an admission that previous modelling they relied on was flawed. As a computational modeller, I have an academic interest in where the modelling went wrong .. 1/n
The latest modelling (and how it deviates from previous models) is described in the #Imperialcollege report here: imperial.ac.uk/media/imperial… .. 2/n
The critical section explaining the sudden shift in policy is on page 16, where we’re informed that the UK has now realised that a “mitigation” strategy will exceed our health care capacity at least 8 times over, but that (quote follows): ..3/n
“this conclusion has only been reached in the last few days, with the refinement of estimates of likely ICU demand due to COVID-19 based on experience in Italy and the UK (previous planning estimates assumed half the demand now estimated)”
How were they so far wide of the mark? The demand estimated in the latest model is 30%, from which we can infer that the previous estimate was 15%. That estimate was based on experience with viral pneumonia .. 5/n ft.com/content/249daf…
And given the novelty of #COVID19, what else could they do but estimate demand by drawing on their knowledge of the flu? Well, they could have based their estimate on data concerning the *actual disease* they were modelling. ..6/n
A paper describing “Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China” was published in the Lancet on *January 24* (yes -- almost 2 months ago)
thelancet.com/journals/lance… .. 7/n
thelancet.com/journals/lance… .. 7/n
It notes that 32% of hospitalised patients were admitted to an ICU and 15% died -- remarkably similar to the “refined” estimates in the new modelling. .. 8/n
Estimates of about 1/3 hospitalised patients requiring ICU care are also contained in the WHO report from mid-February who.int/docs/default-s… .. 9/n
This is the estimate that has been used in other countries when preparing for #COVID19. For example, this letter, published 9th March, reports Singapore's preparations: ccforum.biomedcentral.com/articles/10.11… .. 10/n
It’s hard to avoid concluding that the basic mistake modellers made (which motivated a strategy that by their own estimate would have led to quarter of a million deaths) was to ignore China's experience and treat a novel coronavirus as “just another flu” .. 11/n
