Network Structure of Swine Flu Pandemic

The Swine Flu has been in the news during April 2009. Initially it spread amongst pigs, and then made the jump to humans that came in contact with the contagious pigs. From the early data, this appears to have happened in Mexico.

We use the social network analysis software, InFlow, to illustrate the spread of the contagion. Below is a social network map of the connections between employees in a large organization [data is real, names have been hidden]. A grey line indicates face-to-face [F2F] contact between two employees.


Now, one of the employees visits relatives in Mexico, who live on a farm and have pigs. The swine flu virus infects this employee, whose immune system has not been previously exposed to Swine Influenza A -- H1N1. The link of that infection is mapped below. The contagious pig is represented by the pink node. The disease transmission vector is represented by the green link.


If there is no human-to-human transmission of the swine flu virus then the disease stops here. One person sick, no contacts infected. Unfortunately, the CDC now acknowledges that transmission of this virus from person-to-person is possible -- one patient visited Mexico, came home sick, and passed the virus to another household contact who had not accompanied him on the trip.

Now the contact network at this workplace becomes a disease transmission network, as is common with airborne contagions. Also, this workplace network overlaps with other networks -- each employee goes home to a family/neighborhood/social network which also include F2F contact. As the virus spreads in the workplace, it will also be transmitted to other networks connected to the workplace network by common members.

The sick employee may still come to work initially and via coughing and sneezing start to spread the virus to others in the same space/location. The first wave of disease transmission is mapped below. We now show only the disease transmission links in green and hi-lite each node [yellow] that has been infected by the virus. For this simple example we are assuming a 100% infection rate [higher than usual] -- if someone coughs or sneezes in your presence, you will catch the flu after the normal incubation period, typically 1 to 3 days.


The map below shows how the virus spreads via person-to-person contact. Even if the original infected individuals are no longer at work, the density of the network, with multiple paths [and F2F contacts] between individuals, ensures that many employees will be exposed.



The virus spreads because infectious individuals constantly come in F2F contact with those who have not been exposed. Because of local density in a typical human network, a healthy person can be exposed to multiple sick individuals, thus increasing the odds of transmission. We can watch the contagion spread to the rest of the population, but in most cases some action would be taken once a significant portion of a local population becomes ill. Either government authorities or the employer will step in to isolate the sick from the healthy and adjust work locations.

Ironically, the network structure that enhances the transmission of good contagions -- such as ideas, solutions, and knowledge, can also transmit bad contagions such as disease and fear. When the network is transporting ideas and knowledge we want to decrease distance between individuals. When the network starts to transmit disease we want to increase distance and fragmentation in the network to isolate the virus and slow/stop the spread -- natural work groups need to be identified [via SNA/ONA] and physically separated [no F2F contact]. Another solution is for many people to remain at home [whenever possible] and connect over the Internet [i.e. Skype, GoToMeeting, WebEx, etc.] to coordinate and collaborate and get their work done.

For a deeper dive on social network analysis and contact tracing applied to public health issues see this CDC paper.

Be careful how you connect in the next few months!