Empirical Household Contact Networks and their Implication for Infectious Disease Spread

Pietro Coletti¹, Niel Hens¹

Airborne infectious diseases such as influenza are primarily transmitted from human to human by means of social contacts and thus easily spread within households. Epidemic models, used to gain insight in infectious disease spread and control, typically rely on the assumption of random mixing within households. Until now there was no direct empirical evidence to support this assumption. In this work, we present the first social contact survey designed to study contact networks within households. The survey was conducted in Flanders in 2010-2011 focusing on households with young children. We analysed data from 318 households totaling 1266 individuals with household sizes ranging from 2 to 7 members. Exponential-family random graph models (ERGMs) were fitted to the within-household physical contact networks to reveal the processes driving close contact between household members, both on weekdays and weekends. The fitted ERGMs showed a high degree of clustering, with individuals tightly connected within a household and, specifically on weekdays, decreasing connectedness with increasing household size. Furthermore, we found that the odds of a physical contact between father and child is smaller than for any other pair except for older siblings. This supports the findings of te Beest et al. (2014) i.e. pertussis infection of a child by the father or a sibling takes longer than through other pathways. Epidemic simulations from a discrete-time chain binomial model assuming random mixing within households and using realistic contact networks simulated from the fitted ERGMs, showed only small differences between the two mixing assumptions. Finally, we discuss how the implementation of contact duration may affect the aforementioned picture and we present preliminary results on simulations that take into account the different duration of each contact.