Some seasonality expected for resp viruses, hard to say how much. England had unique circumstance Feb 7-13th.
- week long cold snap across country of similar magnitude
- no change in restrictions
- no increase in daily mobility
1/
Due to variables being reasonably controlled we can estimate how temperature may affect general transmission dynamics.
Note Seasonality: viruses transmit at ALL times of year but climate affects biology and behaviour which impacts on how much transmission occurs.
2/
Cases by specimen date in England showed a marked deceleration in the rate of decline the week after the cold period. From more than -5% per day to less than -2%.
(daily change adjusted for day of the week effect in testing)
3/
Those people testing positive on a given day didn't all catch SARS-COV-2 at the same time
Mean time from exposure to symptoms is 5.5 days + time for getting test (2days). There's a distribution (log-normal best fit) with slightly longer tail. ~8 days from infection to test.
4/
Daily growth rate can then be changed into R.
There's a high degree of correlation with temperature.
Pearson correlation coefficient 0.79
2.2% [1.7-2.6] increase in R per 1C decrease in temp
As an aside: changes in R affect time from exposure to test specimen. As R<1 weight is behind the right of the probability distribution (old exposures slightly dominant and R marginally too high by spec date). Further decline negligibly changes this until equilibrium.
6/
Anyways correlation ≠ causation. Considering other factors.
In Lockdown mobility increases 2.5% avg week on week despite no change in restrictions instead change in behaviour/adherence.
However that week was low/negative between Feb 9-13. Only time in any lockdown.
7/
Result of snow/ice conditions etc and people staying indoors at home.
Usual caveat applies that mobility!= transmission risk
Therefore expectation would be fall in R considering mobility alone.
8/
Also vaccination removes people from the susceptible population and proportionally reduces effective R.
Using compartmentalised mixing of over 70's and vulnerable groups with conservative 65% efficacy and 3 week till protection we see baseline R should have declined.
9/
Physical attendance in school ~18% nationally is unlikely to have contributed to the rising R in week 6 followed by rapid decline week 7 (half term). Effect on case rate was uniform across ages in both weeks, with no delay to infer secondary transmission via children.
10/
IF change in R with temperature is as indicated by this ecological study we can expect seasonality of 29% [22-35] from peak to trough in England over the year. R0 +/- 16%
Slightly higher than Imperial model (20% peak to trough)
11/
However scope of behavioural change was limited in the study period by lock down restrictions. eg indoor socialising with other households rather than outdoor etc.
Which may underestimate impact.
12/
In the context of B117 in England with year average R0=4
This would mean R 4.6 in winter and R 3.4 in summer.
Potential for year round exponential growth and epidemic outbreaks without either pre-existing immunity (vax or prior infxn) or social distancing restrictions.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
But Ab tires don't increase until 2nd dose for Pfizer, especially in the older age cohort who are the priority vaccine targets. How much protection are we giving them?👇
The 86% is a pooled estimate of all ages with younger population dominating.