You, Your Friends, Your Friends of Friends


By GINA KOLATA
IN a way it all seems so obvious. Your friend found a lump in her breast, so you have that long-delayed mammogram. One by one your friends stop smoking, so you stop, too. Of course people are affected by their friends’ habits and their health.
But what seems obvious in the abstract can lead to surprising findings. A recent study found that obesity can spread from friend to friend much like a virus. When one person gains weight, close friends tend to gain weight, too.
The study, published recently in The New England Journal of Medicine, involved a detailed analysis of a large social network of 12,067 people who had been closely followed for 32 years, from 1971 to 2003.
Now, scientists believe that social networks not only can spread diseases, like the common cold, but also may influence many types of behavior — negative and positive — which then affect an individual’s health, as well as a community’s.
“In the past few years we have been seeing a network revolution,” says Albert-Laszlo Barabasi, a physics professor at the University of Notre Dame. “People sensed that networks were out there, but they never had large enough data sets to start understanding them in a quantitative fashion.”
For example, he said, sociologists would go into a classroom and ask students to list their friends. That, he said, can be useful, but social networks are huge, and they evolve over time. They involve you, your family, your friends, your friends’ friends and your friends’ friends’ friends.
The researchers who found the obesity effect stumbled upon the data they needed in the Framingham Heart Study. It has gone on for decades and followed most of the population of Framingham, Mass. As part of the study, participants named friends who could help locate them if researchers lost contact. That link was just what was needed to construct a social network and watch it evolve over decades — a web of friends and friends of friends along with family members.
The striking feature of networks, notes Dr. Nicholas Christakis of Harvard Medical School, is that they amplify whatever effect they are propagating. One person catches a cold and spreads it to 10 friends, each of whom spreads it to 10 more friends.
But obesity? Dr. Christakis and his colleague James H. Fowler, a political scientist at the University of California, San Diego, say they do not know how it happened, but the dynamic was clear — when one person became obese, that person’s friends were more likely to become obese and so were their friends and their friends’ friends.
Obesity was just the start, the researchers say. They have already begun asking about other health-related issues. Smoking, for example — have smokers become more isolated over time?
After all, networks evolve, and if smoking becomes unacceptable, smokers might be expected to cluster in their own little orbits, cut off from the mainstream. That, Dr. Fowler said, is exactly what happened. The Framingham data from the 1970s show smokers embedded in social networks just like everyone else. But by the 1990s, smokers began to be shunted to the side, their links to nonsmokers breaking.
Dr. Fowler and Dr. Christakis are now looking at depression, asking whether it spreads from friend to friend. There are hints from another study that it might.
The network in that case was derived from the federal National Longitudinal Study of Adolescent Health, a study of tens of thousands of teenagers that asks them to name their friends and follows them for years. It turned out, says Peter Bearman, a social scientist at Columbia University, that certain friendships increased the likelihood of suicide or suicidal thoughts in teenage girls.
The risky friendships are what Dr. Bearman calls a contradictory network — a teenage girl has two friends who dislike each other. “It tells you about the importance of social relationships for girls’ health and self-esteem,” Dr. Bearman says. “If you are in an unstable triad, it makes it much more difficult to fit in.”
Now Dr. Bearman and his colleagues are studying autism. The number of autistic children has increased rapidly in recent years, but it is not clear how much resulted from increased diagnosis and how much from an increase in the actual disease.
Dr. Bearman is studying how diagnoses of autism spread. When a child is diagnosed, friends of that child’s parents may wonder whether their child has autism as well, and have their child evaluated. Demand for autism evaluations would increase, and doctors and schools would become more sensitive to the disorder and more likely to suspect it. Schools would then provide services for the autistic children in the community, attracting families from other areas where autism was less common and where schools were not as prepared to help.
“There is an enormously important dynamic that draws people into a diagnostic maelstrom,” Dr. Bearman says. “Autism is real, but the epidemic very likely has a very important social network component.”
The challenge in medicine now is to map out complex dynamic networks, Dr. Barabasi said, and he does not just mean networks of people. The proteins and enzymes in a cell also form a closely connected network, Dr. Barabasi says. And when you tweak one protein, the whole web is affected. “That is why drugs have side effects,” Dr. Barabasi said.
And, he adds, “we will not have cures for obesity or cancer until we understand those networks.”
But he’s an optimist, believing that these cellular networks will be mapped sooner or later. “Right now, this is a work in progress,” Dr. Barabasi says, “but I believe we will get there in 10 years.”