A few predictions

post by Tim Liptrot (rockthecasbah) · 2020-12-25T01:27:17.917Z · LW · GW · 0 comments

Contents

  Question
  Probablity
None
No comments
  1. Will the new, more infectious UK strain impact virus reproduction in the US?

A new Covid-19 strain is rapidly expanding in the UK and has increased reproduction. Several pieces of evidence all point toward the new strain increasing virus reprodcution by 50-80%. My favorite piece of evidence is the regression of new vs. old proportion on number of weekly cases, seen here

Among 40 local authorities in East and South East England with more than five VOC samples there is a significant trend of increasing reported cases with increasing frequency of N501Y (Figure 1, weighted linear regression p=10-6). A 10% difference in VOC frequency in mid-November corresponds to approximately 50 more weekly cases per 100 thousand in early December. Local authorities with few VOCsamples have similar reported cases as the rest of the UK (linear regression intercept = 137 cases per 100k versus UK median 130.4 per 100k).

Question

My question asks - will the new strain migrate to the US and produce an increase in reproduction? I will use the same regression method to answer. Specifically, will a regression between districts show a statistically and practically significant difference in cases associated with the new strain proportion. Statistical significance is p =.05. Practical significance means an association of a 1% change in proportion with an increase of 1% x mean cases per person. For example, if the model assigns all-old-strain districts one case per thousand, an all-new-strain district should have two per thousand. This is much smaller than the UK currently observes.

Null if no studies ever done.

Negative if media reports new strain still not in US on February 28th.

Probablity

The probability that the strain does not reach the US is very low, may .05%. If the strain gets here I expect a similar result to the UK. If anything the US has a weaker control mechanism. OTOH, model uncertainty is huge.

Conclusion: 80%

Note: This is just a calibration exercise for me.

0 comments

Comments sorted by top scores.