(R)egional variation and case numbers
May 17, 2020 • Reading time 3 minutes
Published 18 May 2020
Since the Government’s announcement last week, there have been several attempts to get under the skin of the critical variables underpinning the decision making in their strategy – R and the number of people infected. These are both valuable bits of information for people to “stay alert”.
Both values need to be nowcasted as the available data are limited. There is a lack of community testing and a persistent inability to report critical variables like the number of deaths accurately. But what has become clear is that there is significant regional variation in both R and the number of cases. The map below shows our estimate of the number of cases (1 in X people) along with estimates of R calculated by the London School of Hygiene and Tropical Medicine. *
This map shows that in London R and the number of people infected has fallen substantially over the past few weeks – some estimates suggest R in London is as low as 0.4 – but remain relatively high. While Cornwall has very few cases spread over a large area, the estimate of R is estimated to be close to 1.
A lot sits behind a proper understanding of these number. For example, is London’s low R due to some herd immunity from the high infections it incurred in March and April?
As seen in countries like Singapore, the risk of a surge in new cases can happen even when the virus is under control. These surges, like the increase in cases in the North West of England, tend to be localised. If they are detected early, they can more easily be brought under some control. This requires a flexible regional response as well as aggressive test and trace.
The Government’s “Plan To Rebuild” sets out the possibility for regional variations (although this is currently ruled out in practice):
“Restrictions may be adjusted by the devolved administrations at a different pace in Scotland, Wales and Northern Ireland because the level of infection – and therefore the risk – will differ. Similarly, in England, the Government may adjust restrictions in some regions before others: a greater risk in Cornwall should not lead to disproportionate restrictions in Newcastle if the risk is lower”.
Aside from the regional variations, cities and mass gatherings are likely to remain more vulnerable than rural areas to surges in new cases when lockdown measures are eased. This is because Covid-19 is a disease that spreads fastest in more densely populated areas.
* Note on methodology for calculating the number of cases
Broadly steps include:
- To estimate how many people may be infected with Covid-19 we follow these steps:
- Estimate the number of deaths that have taken place in each region between the 28th of April and the 12th of May
- Based on IFR rates consistent with the study by Cambridge and estimated case growth from our own study, we estimate the number of people likely to be infected per region
- These estimates of infection numbers are then allocated to Upper Tier Local Authorities (UTLA) based on their share of reported cases between the 28th of April and 12th of May
- The number of infected people (1 in X) is calculated based on reported population numbers
Points to note when interpreting this methodology and findings from this study:
- Deaths are those recorded in hospitals only (23,380 as of the 8th of May) – so deaths in care homes are not included. This may mean total infections are under-recorded, although regional variations should be less affected. (Non-hospitals deaths are not given a location, so currently these cannot be allocated.)
- Total infections are based on IFR of 0.3 in London and 0.64 in other regions. This is an attempt to reflect different age profiles. The growth rate in cases is -12% in London and -10% in the other regions. This reduction in the number of cases is comparable with other studies, such as the one by Cambridge.
- Variation in testing, which results in a UTLA within a region having a higher or lower share of total regional cases may distort estimates of total cases.