Exploiting Social Networks to Tackle Obesity

Readers of these pages will probably recall the work by Christakis and Fowler demonstrating the “contagious” spread of obesity in a large social network (NEJM 2007).

In a paper appearing in this month’s issue of OBESITY, David Bahr and colleagues from Denver, Colorado, USA, use computer simulation to determine if these findings can be exploited to reverse the obesity epidemic by intervening in social networks.

The paper makes a fascinating read, although the key message is very sobering (despite the rather optimistic interpretation by the authors).

In essence, the simulations suggest that because clusters dominate individual behaviours, once a large cluster of obese individuals has formed it becomes self-sustaining, because an individual in the middle of the cluster (e.g., social network of obese friends) will have a very difficult time sustaining weight loss. The surrounding sea of obesity ensures that even a temporary loss of weight in a few individuals is rapidly reversed, a result that remarkable resembles what is frequently observed in weight loss intervention studies.

Rather than recruiting friends to help with weight loss, the simulations suggest that it may be a better strategy to recruit friends of friends, who help establish contacts to members of other networks.

From a population perspective, one of the more effective strategies could be to target well-connected individuals on the edge of a cluster (i.e., those whose social network contains individuals of more than one BMI cluster). In contrast, targeting poorly connected individuals in a tight network of other obese individuals is likely doomed to failure.

Unfortunately, given the high prevalence of obesity in the US (not so different from that in Canada), the majority of the population already lives in social networks that are obese rather than normal weight. Therefore, finding a critical mass of key “agents of change” will pose challenging, even if the simulations show that these key individuals only have to make up around 1% of the population across BMI ranges.

While all of this sounds great in theory, the paper of course is based on computer simulations – whether or not this knowledge can actually be exploited in real life remains to be seen.

Given my natural skepticism, I am certainly not holding my breath.

Hamilton, Ontario