Contacts in context: large-scale setting-specific social mixing matrices from the BBC Pandemic project

Petra Klepac; Adam J Kucharski; Andrew JK Conlan; Stephen Kissler; Maria Tang; Hannah Fry; Julia R Gog

To date, the gold-standard for modelling age-specific contact patterns in many settings has been the POLYMOD dataset collected in 2005/06 [22] . In Great Britain, the POLYMOD study had 1,012 participants reporting a total of 11,876 contacts (11.74 contacts on average). This is higher than the daily average reported number of contacts in the BBC contact dataset with 36,155 participants reporting 378,599 contacts (10.47 contacts reported on average). For most age groups, the total number of contacts per day for an average person (assuming 2018 population structure [1] ) is remarkably similar between BBC and POLYMOD datasets, especially for ages over 60 (Fig 6A) . In BBC contact data a smaller proportion of those contacts are physical (and correspondingly a higher proportion of contacts are conversational only). The reduction in the average number of contacts for ages 10-19 is striking between POLYMOD and BBC datasets. While for the 10-14 age group this may be affected with the fact that we are only sampling a subset of this age group and assuming that 13-and 14-year-olds are representative of the entire age group, for 15-to 19-year-olds this reduction is real. This could be the effect of our sampled population, or it might be a signature of a real change in social contacts of teenagers since POLYMOD. A survey of over 1,000 13-to 17-year-olds in the US in the 2018 showed that compared to 2012 teens' preference for direct face-to-face communication with friends has declined substantially (from a half to a third) while the interactions over social media have increased [24] . A similar trend in digitisation of teenager interactions is likely to be present also in the rest of the world with comparable teenage mobile phone use, and it is possible that the reduction in both conversational and physical contacts in our dataset is a reflection of teenage interactions moving away from face-to-face to social media.