COVID-ARG.COM          

By Matt Fletcher 

Matt Fletcher

COVID-19 Actuaries Response Group – Learn. Share. Educate. Influence.

Summary

As we move through the pandemic, more SARS-CoV-2 variants continue to emerge. Most have no features of particular interest, but some emerge that are deemed “of concern” because of

• their speed of spread
• the severity of illness they cause
• the likelihood that they may “escape” the vaccines and / or infect those who have been previously infected.

The first variant of concern (VoC) that came to the attention of the wider public in the UK was B.1.1.7 (“Kent”) which we discussed in our December 2020 bulletin “A tale of two variants”. This spread more quickly than existing strains, and is widely believed to been a key driver behind the second wave of infection and deaths seen in the UK as well as elsewhere in the world.

As the UK takes steps out of lockdown, we are seeing news about different VoCs, particularly B.1.617.2 which first emerged in India. In this note, we focus on B.1.617.2 but we recognise that there are currently five VoCs, and a further eight variants under investigation.

We conclude that B.1.617.2 does appear to be worthy of additional concern although it is still possible that additional data will show that the variant is not spreading significantly more quickly than existing variants.

We also note that, whilst this variant’s eventual impact in the UK may be muted due to the high level of vaccination in this country, countries with lower levels of vaccination will need to be vigilant about the importation and spread of this and similar variants.

The data

Relatively little is currently known about B.1.617.2, but evidence is being collected as to its likely features compared to previous variants.

Public Health England (PHE) have produced various publications covering variants of concern in recent days, including a Technical briefing, a Risk assessment and a detailed study on vaccine effectiveness against the B.1.617.2 variant.

PHE clearly have significant concern over the new variant, as set out in the risk assessment:

• Transmissibility appears greater than wild-type SARS-CoV-2.
  • It is likely that it is more transmissible than B.1.1.7 – it’s estimated that B.1.617.2 infections are increasing twice as fast as B.1.1.7.
  • Despite this, the magnitude of the transmissibility advantage is uncertain, particularly because no adjustment has yet been made for vaccination status. If those infected with B.1.617.2 and their contacts are less likely to be vaccinated, the observed transmissibility advantage may be overstated.
  • Reduced vaccine effectiveness, covered next, might also be driving the faster spread.
• There is evidence of reduced effectiveness of vaccines, with a current estimate of a 20% reduction after a single dose.
  • The technical briefing sets out that overall effectiveness (combined across all vaccines) against B.1.1.7 is estimated to be around 51% after one dose, and 87% after two doses, compared to 33% after one dose and 81% after two doses against B.1.617.2.
  • The detailed study suggests that effectiveness after a single dose is similar for the two main vaccines used in the UK (Pfizer/BioNTech (“Pfizer”) and Oxford/AstraZeneca (“AZ”)) but that currently AZ is less effective than Pfizer after two doses (60% compared to 88%). It’s important to note that this analysis is based on very few cases. It may also be explained by limited follow-up after second doses of the AZ vaccine, assuming that maximum effectiveness is reached more than two weeks after vaccination.
• The reduced effectiveness after one dose is particularly relevant today, as around 30% of the adult population have had their first but are yet to have their second dose. This is likely to be behind the move to accelerate second doses for the remainder of Groups 1 to 9 (over 50s).

There are some points on the (tentatively) positive side that are also set out in PHE’s report:

• No evidence of increased severity of infection, though there has been insufficient follow-up time to be certain of this.
• Monitoring continues on reinfection risk; a national healthcare worker cohort study (most of whom have been vaccinated) shows no increase.
• Vaccines are still expected to remain effective against severe disease from B.1.617.2.

The modelling

It is sensible to ask what the features of a variant would need to look like before the roadmap out of lockdown is slowed or even reversed. Previous decisions having been made based on the numbers of deaths expected and (in particular) whether the NHS might be overwhelmed. Whilst we still don’t have enough information to definitively determine whether B.1.617.2 is likely to be significantly more impactful than existing variants, it is possible to model what may happen if we change certain features.

Modelling of different variant characteristics (e.g. higher speed of spread, lower vaccine efficacy, higher risk of vaccine escape or reinfection) was carried out by various academic groups and presented to SAGE in early May. We focus here on the Imperial paper as (in our view) this provides the clearest set of scenarios relating to VoCs. The paper was based on key sources of data (such as past infection rates, vaccination coverage, vaccination effectiveness) as at 1 May 2021.

These scenarios are specifically not intended to be predictions as to the exact characteristics of B.1.617.2, but models of what the impact could be if a VoC does have these characteristics.

Scenarios

The central assumption is that a VoC:

• Is as transmissible as B.1.1.7
• Can be caught after B.1.1.7 (prior infection having similar efficacy to one dose of Pfizer vaccine)

It is assumed that efficacy of mRNA vaccines (Pfizer and Moderna) is high against both disease and infection, but that efficacy of vector vaccines (AZ) is low.

Scenarios are considered with no VoC, as well as scenarios with lower transmissibility and higher or lower immune escape.

The scenarios considered are labelled as Central (i.e. no VoC emerges that has features that are materially worse than existing variants) and six different possibilities for VoCs with characteristics as follows:

Impacts on deaths, infections, hospitalisations and maximum hospital occupancy are estimated, both up to step 3 of the lockdown (no further lifting after 17 May) and up to step 4 (full lift on 21 June).

The first chart shows the number of deaths from COVID-19 that might be expected between 1 May 2021 and 9 August 2021 if no further steps are taken towards reopening after Step 3. The extent of the blue areas reflects the uncertainty; the white line shows the central result.

None of the scenarios results in very large numbers of additional deaths over the coming months (relative to the over 100,000 COVID-19 deaths recorded in the UK in the last year). However, an outcome such as VoC2 would mean a notable increase in deaths.

Typically, a scenario where a variant is spreading much more quickly can be picked up by looking at the number of infections and the number of hospitalisations.

Here, VoC 1 and VoC2 are producing noticeably more infections than the central assumption.

The next chart shows the numbers of deaths expected over the period 1 May 2021 to 1 June 2022 in each scenario, assuming we move to step 4 on 21 June 2021 as currently planned.

Here, the scenarios VoC1, VoC2, VoC4, and VoC5 give much higher numbers of deaths than the Central scenario. As an example, whilst the Central scenario gives 1,500 deaths under step 3 and 9,000 after step 4, VoC2 increases from 5,300 deaths under step 3 to 225,400 with step 4. It should be stressed that the evidence now emerging suggests a much more limited vaccine escape than used under VoC2.

VoC3 and VoC6 are in contrast much less concerning. The paper notes that rapid roll-out of vaccines and high vaccine effectiveness means that moving to step 4 with the current variant is likely to be achievable with a small ‘exit wave’ of deaths (as illustrated by the Central scenario). However, they note that it is important to prevent the importation of VoCs with high immune escape properties, which (as set out in the last chart) could lead to future waves with even more deaths than those already experienced.

It is worth noting that this analysis is specifically based on the situation in England (and can be reliably extrapolated to the rest of the United Kingdom) which has high levels of vaccination; clearly this is a global pandemic, and countries with lower levels of vaccination could be very badly hit by a VoC which spreads more quickly than other variants.

We have not sought to audit the model produced by Imperial although we note that other models produce similar results given similar inputs. However, we are also aware that other assumptions in relation to both current variants and VoCs, as well as on the population’s behaviour throughout and following the lifting of lockdown, would produce different results.

27 May 2021