A slender person with overweight friends is more likely to gain weight than one whose buddies are svelte. That's the conclusion of a new study that mined 30 years' worth of data and found that obesity correlates strongly with social networks.
Social networks influence smoking patterns, exercise, the likelihood of dying soon after one's spouse, and other medical questions. But most such studies have relied on data collected at one time, notes Duncan Watts, a sociologist at Columbia University. In reality, social networks evolve, and capturing these ever-shifting patterns has been hard.
To determine how webs of relationships affect body weight over time, medical sociologist and physician Nicholas Christakis of Harvard Medical School in Boston and political scientist James Fowler of the University of California, San Diego, turned to health records from the Framingham Heart Study. Launched in 1948 to understand the genesis of heart disease, the project has collected information on thousands of residents of Framingham, Massachusetts, and their descendents. Christakis study drew on a subset of those data, a social network that includes about 5000 individuals and more than 7000 of their parents, siblings, spouses, and friends.
A person's chance of becoming obese significantly increased if he or she had an obese friend, sibling, or spouse, the researchers report in tomorrow's New England Journal of Medicine. The strength of the relationship played a role: If a friendship was mutual, that boosted the chance of obesity 171%; but if one of the friends didn't acknowledge the friendship in the survey, that individual was at no increased risk of obesity. The researchers ruled out confounding factors that might independently make two friends obese, such as the tendency to hang out with similarly sized people or a shared environment, such as a new fast-food restaurant. They conclude that the effect on obesity persists through up to three degrees of separation--for example, a spouse's sister's friend--suggesting that being overweight spreads through social networks almost like a disease.
Watts praises the team for mining a conclusion from such a rich source of data gathered over time: "It's actually extremely difficult to show that the network is doing anything." Hurdles aside, showing a social influence on obesity isn't a surprise, notes David Allison, a biostatistician and obesity researcher at the University of Alabama, Birmingham. He also has reservations, he says, about suggesting that obesity can "spread" like an infectious disease, because he's not convinced that one person's obesity is causing that of others two or three degrees removed from them.