The main outcome measure for the study is the Center for Epidemiological

DUCIaI "EIWOTK ueuermlnants 0T UEPTBSSIOI‘I JN Rosenquist‘-"’, JH Fowler3 and NA Christakis“-5 ‘Department of Psychiatry, Massachusetts General Hospital, Boston, MA, USA; 2Department of Health Care Policy, HarvardMedical School, Boston, MA, USA; 3Department of Political Science, University of California, San Diego, CA, USA; “Departmentof Health Care Policy, Harvard Medical School, Cambridge, MA, USA and 5Department of Sociology, Harvard University,Cambridge, MA, USA The etiology of depression has long been thought to Include social environmental factors. Toquantitatively explore the novel possibility of person-to-person spread and network-leveldetermlnatlon of depresslve symptoms, analyses were performed on a densely interconnectedsoclal network of 12067 people assessed repeatedly over 32 years as part of the FraminghamHeart Study. Longitudinal statistlcal models were used to examlne whether depresslvesymptoms In one person were assoclated with similar scores In friends, co-workers, siblings,spouses and neighbors. Depressive symptoms were assessed using CES-D scores that wereavailable for subjects In three waves measured between 1983 and 2001. Results showed bothlow and high CES-D scores (and classification as being depressed) In a given period werestrongly correlated with such scores In one‘s friends and neighbors. This associationextended up to three degrees of separation (to one’s frlends' frlends' friends). Female friendsappear to be especially Influential In the spread of depression from one person to another. Theresults are robust to multiple network simulation and estimation methods, suggesting thatnetwork phenomena appear relevant to the epidemiology of depression and would benefitfrom further study. Molecular Psychiatry (2011) 16, 273—281; doi:10.1038/mp.2010.13; published online 16 March 2010 Keywords: depression; social networks; sociology; social norms; mood Introduction Depression is a significant cause of worldwide morbi-dity and mortality. Current estimates suggest a life-time incidence of between 13.3 and 17.1% in theUnited States and a yearly cross-sectional preva-lence ranging from 2.3—4.996.1 Using any number ofmetrics, the cost of depression is enormous. Forexample, disability, morbidity and mortality resultingfrom depression was estimated to cost $86 billion inthe year 2000 alone.z The etiology of depression as an illness has beenconceptualized to have a number of interactingbiological, psychological and social components.3This idea that social forces may impact mood symp-toms was first hypothesized over 100 years ago in thecontext of suicide by the sociologist Emile Durkheim.He noted that suicide rates stayed the same acrosstime and across groups even though the individualmembers of those groups came and went.‘ Durkheim’sconclusion was that whether people took their ownlives depended in part on the kind of society theyinhabited. He noted that although depression andsuicide were seen as entirely individualistic. theymay be partly driven by social forces. More recent Correspondence: Dr ]N Rosenquist, Health Care Policy, HarvardMedical School, 180 Longwood Avenue, Boston. MA 02115, USA.E-mail: jrosenqufigmailmom Received 29 May 2009; revised 26 December 2009; accepted 27December 2009; published online 16 March 2010 work on the social influences on depression find asignificant correlation between social factors such aschild abuse, disruptions in family functioning, stress-ful life events and neighborhood characteristics?’3 The literature on social determinants of disease hasbeen augmented in recent years by a growingliterature focused on understanding the role of socialnetwork structure on individual outcomes. Recentwork has yielded results suggesting that traits such asobesity, smoking behavior. happiness and lonelinessmay spread along social networks over time.EHS Aperson’s structural position within a network, such astheir transitivity (whether their friends are friendswith each other] and centrality (whether they arelocated in the middle or edge of the network) havebeen found to affect the development of traits andbehaviors. For example, Bearman and Moody foundthat social isolation and (among women) havingfriends who were not friends with each other weretwo factors predictive for suicidal ideation, suggest-ing the structural components of a person's networkimpacted their behavior.“5 In addition to such structural effects of networkposition, there may also be influence effects, wherebydepression might spread among friends, familymembers, co-workers and neighbors. While suchinfluence effects may have an intuitive appeal (mostpeople can no doubt think of instances where theyfound themselves influenced by a family memberor friend), it is crucial to distinguish among three