Quantifying social contacts in a household setting of rural Kenya

Everyday, we interact with people from different social, demographic, economic, cultural and mobility backgrounds. These backgrounds determine whom we meet, where we meet, and the duration of the interaction.

In epidemiology, these characteristics are important in the spread of infections that require close or direct physical contact, such influenza, RSV, measles and other respiratory illnesses. For example, the probability of a school child acquiring infection from fellow students or siblings could be higher compared to acquiring it from a friend. This can be explained by the fact that, a school child will spend more time either at school or at home.

Data on individual interactions and how they shape the transmission events are difficult to collect and obtain. A novel way of collecting this data is use of electronic proximity sensors . Each sensor detects and records the “sighting” of a similar tag worn by different participant when they come to close proximity. Therefore, it is possible to determine who an individual meets, at what time, and for how long (maybe even location) by looking at the time-stamp data collected by the electronic sensors.

In order to assess the challenges of deploying electronic sensors to obtain social contact data that would inform on transmission patterns of respiratory illness in a developing country background such as Kenya, we set out to collect close proximity contact data from 100 individuals in 5 households located within the Kilifi Heath Demographic Survey (KDHSS).

Each individual, apart from one infant, was given a sensor to wear all the time for 4 days. They would wear the sensor from the time they wake in the morning up to the time they go to sleep in the night. In addition to the sensor’s proximity data, we collected each individual’s views concerning the sensors for example what do they think of the devices? Would they be willing to wear them for longer periods? What kind of questions were they asked when seen wearing them? Were they concerned about their privacy?

Although this was a small study, we present interesting findings that have not been demonstrated empirically before in a remote rural population from a developing country. An important output of this study was community-based suggestion on how to improve the study methods. Through focus group discussions, participants mentioned that intensive community engagement describing the study tools and procedures was important to allay privacy and confidentiality concerns, especially of non-participants. Within households, we demonstrated the presence of strong ties especially between school going children. In three of the five households with contemporaneous data, adults formed the links between the households. Lastly, we also demonstrated a circadian rhythm of high interactions in the early morning, noon and evening, with dips occurring in between. This can be explained by regular routines of household members.

Although these results may not be generalizable to other contexts, they are important for future study designs that involve larger populations. Currently, plans are underway to conduct large studies that involve schools and households located in urban and rural settings within Kilifi. Results from human network modeling are important in designing interventions against transmission of infections spread via close contact, especially those that are not covered by current vaccines.