Thread by @trvrb

There is now enough genomic data from the US and Germany to repeat this approach to estimating Omicron-specific rate of epidemic spread. Here, we observe similar initial rapid spread in the US and Germany. 1/10

https://twitter.com/trvrb/status/1469157770716872708

As before, we partition case counts from @OurWorldInData using sequences from @GISAID into estimated Omicron, Delta and other cases, and we use this partitioning to infer variant-specific Rt and epidemic growth rate r (methods and code here github.com/blab/rt-from-f…). 2/10

We find that logistic growth of Omicron sequence fraction looks similar between the UK, the US and Germany with roughly 1% of sequenced cases in all three countries being Omicron on Dec 1. 3/10

Combining this with case count data in @marlinfiggins model framework estimates Omicron Rt at 3.1 for Germany, 3.0 for the US and 4.8 for the UK but with wide 80% credible intervals. 4/10

To avoid assumptions of generation interval, we also estimate epidemic growth rate r across countries, with r = 0.19 in Germany, r = 0.19 in the US and r = 0.27 in the UK. This corresponds to doubling times on Dec 1 of 3.7 days, 3.7 days and 2.5 days respectively. 5/10

These numbers differ somewhat from what I presented on Thursday primarily due to exact model choice. On Thursday I presented estimates for South Africa and the UK from the "free Rt" model, whereas today's estimates are from the "fixed growth" model. 6/10

Basically, the "free Rt" model treats each variant separately and tries to estimate an independent Rt trajectory for each, while the "fixed growth" model estimates a single Rt trajectory and inflates variant-specific Rt by a fixed value. 7/10

We've observed that the "fixed growth" model is more robust during initial spread while variant is still rare and so we've switched to this model for primary analysis. Full details on model differences are available in the technical report at bedford.io/papers/figgins…. 8/10

I've also pulled back data to 10 days before present rather than the previous 7 days to better correspond to lags in sequence deposition from the US and Germany. @marlinfiggins and I plan to keep updating figures at github.com/blab/rt-from-f… every day as new data arrives. 9/10

It's interesting that so far Rt values appear quite similar across countries. However, as time accrues and more countries collect sequence data, we may be able to make some systematic comparisons in rate of spread across countries. 10/10