search

epiforecasts / evaluate-delta-for-forecasting

Evaluating the impact of modelling strain dynamics on short-term COVID-19 forecast performance
Other
1 stars 0 forks source link

Early prior source #5

Open seabbs opened 3 years ago

seabbs commented 3 years ago

Plan is to use this work though will need to back out an effective size from Rt comparision as not stated in any of the reports as far as I can tell.

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/993104/S1253_Modelling_B.1.617.2_LSHTM.pdf

SAGE source: https://www.gov.uk/government/publications/lshtm-dynamics-of-delta-b16172-in-the-uk-from-importations-traveller-linked-and-non-traveller-linked-transmission-25-may-2021

Summary report from SAGE: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/993318/S1251_SPI-M-O_Consensus_Statement.pdf

seabbs commented 3 years ago

Backing out the estimates I get:

Growth rate of non-delta cases: 2.8% (95% CrI: 2.3-3.6%) Assumed generation time: LogNormal(log(5.4 days), 0.4) Non-Traveller Delta Rt: 1.9 (95% CrI: 1.7-2) Non-Traveller Delta Rt post 7th May: 1.2 (95% CrI: 1-1.5) Report links its estimated growth rate with the SPI-M Rt estimate from the 23rd of April (https://www.gov.uk/guidance/the-r-value-and-growth-rate#history): This gives an Rt of 0.9 (0.8, 1.0)

The estimates and method are fairly non-standard which makes deriving an effect size a little tricky. If we assume that Rt estimates are comparable to SPI-M estimates (which were used as a reference and hence this seems a reasonable approximation) then we can find a crude bound for the additional transmissibility using the upper and lower bounds and the central estimates.

Pre 7th May: 2.1 (1.7 - 2.5) Post 7th May: 1.3 (1 - 1.875)

Logged this gives:

Pre 7th May: 0.74 (0.53 - 0.92) Post 7th May: 0.26 (0 - 0.623)

Assuming this is approximately normal (which it is not) we get:

Pre 7th May: 0.74 SD: 0.1 Post 7th May: 0.26 SD: 0.16

It is entirely unclear from the report whether non-delta growth was rescaled after the 7th of May and so we choose to use the pre 7th of May estimate.

seabbs commented 3 years ago

Need to add this to paper draft as a reference and then can close.