Blog

The UK is gradually emerging from its third lockdown during the pandemic. As restrictions are lifting, we invited Trystan Leng, a mathematical epidemiologist based at the University of Warwick to discuss his analysis. This was carried out during the first UK lockdown to model a way that would safely allow individuals to increase their physical social interactions beyond their household, while limiting the additional risk of infection.

Read the interview with Trystan to find out more about social bubbles, a strategy based on social clustering that relies on public adherence to be effective, and the relevant role it and similar strategies will play in our lives as we transition out of lockdown.

What is your research focus?

I’m a postdoctoral research assistant with the JUNIPER consortium, a consortium of modelling groups from different universities working on epidemic modelling. My research focuses on the importance of including contact network structure in epidemiological models, and constructing models that incorporate this structure to answer questions for public health.

What is the ‘bubble’ strategy?

In a lockdown situation, people’s social contacts are restricted to others within their own household. The social bubble strategy is the idea that households could choose one other household to form an exclusive ‘bubble’ with. Within this bubble, people are allowed to mix freely. Through the exclusivity of the bubble, it is hoped that such a strategy will allow an increase in people’s social contacts while minimising the risk resulting from such an increase in contacts.

Why was this a helpful approach to adopt as part of the initial lockdown exit strategy?  

We know that epidemics spread slower in clustered populations. The idea behind the social bubble strategy is to exploit the clustering of social contacts such a policy imposes. Managed appropriately, a social bubble strategy can give support to the households that need it most throughout lockdown, such as single occupancy households and households with young children, without increasing the R number above 1.

How did you assess the strategy?

We designed a network-based model of 10,000 households. Within this model, individuals had three potential sources of infection: from within their own household, from the other household within their bubble, and from the community at large. The size and age-structure of households in the model was informed by the most recent census of England and Wales.

Using this model, we were able to assess the impact of different social bubble strategies, that targeted different subsets of households, on epidemic risk and fatalities. By using a network approach, we were able to compare the increase in transmission under social bubble strategies to the increase in transmission from individuals extending their social contacts in an unclustered fashion, to assess the effectiveness of social bubble strategies.

Can social bubbles reduce infection and mortality risk?

While allowing all households to form social bubbles posed the risk of increasing R above 1, social bubble strategies that allowed particular types of household that are likely to benefit most had a relatively minor impact on transmission. In particular, we found that allowing single occupancy households to form social bubbles with other single occupancy households, or households with young children to form social bubbles with other households with young children, would be unlikely to increase R above 1 across the range of assumptions considered. In all cases, social bubbles were effective at reducing the mortality risk associated with increasing a household’s social contacts, compared to unclustered increases in social contacts.

What could impede the success of the strategy? 

For the strategy to be successful, it is important that participating households adhere to the exclusivity of social bubbles. We found that a lack of adherence could result in chains of transmission spanning many households, undermining the effectiveness of social bubble strategies. 

The impact of new variants could also challenge the strategy’s viability. Something we can gather from data is the probability an infected individual will infect other members of their household, known as the household secondary attack rate. We used the best available data at the time to inform the household secondary attack rate of COVID-19 assumed in our model. If COVID-19 has a much larger household secondary attack rate than assumed in the model, then allowing social bubbles could result in a larger increase in transmission. Because of this, monitoring the impact of new variants on the household secondary attack rate of COVID-19 will be important to inform the suitability of social bubble strategies. 

How can social bubbles work alongside other measures? 

Social bubble strategies are effective because they result in exclusive groups of clustered social contacts. If people within the bubble have other social contacts that are unclustered, this undermines the effectiveness of such a strategy. The impact of social bubble strategies alongside other relaxation measures was something we did not explore in the paper, but aligning the social contacts that occur because of social bubbles with other contacts that occur because of other relaxation measures may reduce the epidemic risk that results from relaxation. For example, if schools are open, a family forming a social bubble with a family with children in the same class may be effective at reducing epidemic risk, because some of those social contacts are already occurring.

As the UK is now in its third lockdown, do you think this strategy is still a vital tool going forwards? 

We are still grappling with the best way to relax lockdown policies while minimising the increase in transmission and fatalities that result from such a relaxation. While we remain in lockdown, social bubbles are providing much needed support for many households that need extra support or social contact. Social bubble strategies, and other strategies that exploit the clustering of social contacts, should be considered as potential relaxation measures as we transition from a lockdown to a non-lockdown situation, and will remain relevant for a long while to come.  

“This is a valuable piece of work that I hope has contributed to the evidence for policymaking during emergence from lockdown after the first wave of the pandemic in the UK,” says Jessica Enright, School of Computing Science, University of Glasgow, UK in her review of the article. Read the full study and the peer review reports at Wellcome Open Research.