Analysis of SARS-CoV-2 growth illustration presented at Vallance-Whitty Downing Street Briefing on 21 Sep 2020

-

ANALYSIS OF SARS-CoV-2 GROWTH ILLUSTRATION
PRESENTED AT VALLANCE-WHITTY
DOWNING STREET BRIEFING ON 21 SEP 2020


[26 Sep 2020v1]

Implied CFR appears to be >50% lower than IFR used by ICL in March, suggesting new cases in red area would lead to ~2,000 deaths (~5% of spring total).



The red area would contain a total of ~500,000 new cases.

How many deaths would be expected to occur as a result?

PV commented that, by mid October, there might be 50,000 new cases per day & that this would be expected to lead, a month or so later, to 200 plus deaths per day. 

This implies a Case Fatality Rate of approx:

CFR = 200 / 50,000 = ~ 0.4%.

Since not all infections are detected, the Infection Fatality Rate (IFR) will be lower still.

This ~0.4% CFR is, however, in itself 50%+ lower than the 0.9% IFR used in ICL’s March paper & modelling. 

Total deaths resulting from the new cases that occurred during the red period would be:

~500,000 x 0.4% = ~2,000 deaths.

This is equivalent to 5% of the spring total. 


SUMMARY:

1) Implied CFR of ~0.4% appears to be at least 50% lower than IFR used in ICL March 2020 paper & modelling.

2) Total deaths caused by new cases occurring during the red period would be significantly lower than the spring total (~2,000, about 5% of the spring total).

 


@RuminatorDan


Popular posts from this blog

Czech Republic & other European Higher Fatality Countries

-
Image
CZECH REPUBLIC & OTHER EUROPEAN HIGHER FATALITY COUNTRIES Signs of Czech Rep deaths slowing? I wish them well. Interesting, yet again, that the cumulative total is so similar to other hard hit locations, which are all around 10x lower than generally used predictions... The point is not to use these data for some ugly competition. Rather, these data and their differences - or similarities - can reveal very important information about #SARSCoV2 . For example, UK deaths were predicted to reach ~7500 per Million (=~500k people, without measures). However, if the hypotheses that led to such predictions were true, & such totals have only been averted by lockdown, then one would expect large variations between countries (as exact details & timing of lockdown and other measures would be critical). High sensitivity to timing would be the case if "high lethality+all susceptible+lockdowns work" hypotheses were true. Yet all the high fatality locations have quite similar tota...
Read more

England & Wales Autumn Covid deaths show regional heterogeneity. London, in particular, is interesting

-
Image
England & Wales Autumn Covid deaths show regional heterogeneity. London, in particular, is interesting [27Nov2020_1] London's more severe & earlier spring growth may well explain why it has, to date, seen less of an autumn rise. [Charts normalised by year to date non-Covid deaths.] The generally held belief is that only a small percentage of people have so far been infected by SARS-CoV-2 & that the vast majority remain susceptible to infection - hence all the measures taken. There are, however, reasons & evidence suggesting that this is not the case. If the vast majority of people were still susceptible, one would expect growth everywhere. However, autumn rises have varied greatly across regions. London especially, given its nature (crowding etc) would be expected to show aggressive growth, as it did in spring.  It has not. The past provides a clue as to the reason. London's spring rise in Covid deaths was the most aggressive: not only faster & larger than o...
Read more

False Positives mean R can never reach zero and will remain around 1 indefinitely

-
Image
False Positives mean R can never reach zero and will remain around 1 indefinitely If prevalence was low enough that most positives were false, cases would appear to stop falling & settle at a low level (+noise). R would rise from below 1 to around 1 & jump around at that level. Indefinitely. E.g. below. % of positive tests declines; then settles at a low but non zero level. If R is determined by the changes in this number (slope of graph) it would be <1 during decline, then rise & settle around 1 but with significant variation due to noise etc being larger at low levels. @RuminatorDan
Read more