The relative importance of frequency of contacts and time of exposure for the spread of directly transmitted infections

Alessia Melegaro, Università Bocconi
Elisabetta De Cao, Università Bocconi
Emilio Zagheni, Max Planck Institute for Demographic Research
Piero Manfredi, Università di Pisa

The impact of public health interventions against infectious diseases critically depends on how individuals mix and the social context in which this mixing occurs. The increasing availability of time use data and survey data on social contacts has improved our understanding of the relationship between social interaction and close-contact childhood infection processes. In the literature, the two data sources have been analyzed independently of each other. We propose a novel model that integrates the two independent data sources (time use data and contact surveys). We estimate age-specific mixing matrices that describe more precisely the relevant social structure that explains the spread of infections, by assuming that potentially infectious contacts are proportional to both self-reported number of social contacts and time of exposure in social activities. The mixing matrix and transmission parameters are jointly estimated using the Bayesian melding approach, to take into account the overall uncertainty. We apply our methods to Italian data on social contacts, time use, and age-specific seroprevalence of varicella and parvovirus B19. We estimate that about 23% of the time of exposure per social contact is suitable for varicella transmission, whereas only around 1% is suitable for Parvovirus B19. The Reproductive number for varicella is around 6 (95% CI: 5.8-6.7), while for B19 is around 3 (95% CI: 2.8-3.1). Social contacts involving physical touch appear to be the most relevant ones to explain the spread of respiratory infections.