Social Relationships and Health
Carnegie Mellon University
The author discusses 3 variables that assess different
aspects of social relationships—social support, social
integration, and negative interaction. The author argues
that all 3 are associated with health outcomes, that these
variables each inﬂuence health through different
mechanisms, and that associations between these variables
and health are not spurious ﬁndings attributable to our
personalities. This argument suggests a broader view of
how to intervene in social networks to improve health. This
includes facilitating both social integration and social
support by creating and nurturing both close (strong) and
peripheral (weak) ties within natural social networks and
reducing opportunities for negative social interaction.
Finally, the author emphasizes the necessity to understand
more about who beneﬁts most and least from social-
There has been much recent emphasis on the role of social
relationships in our physical health (e.g., Cohen, Gottlieb,
& Underwood, 2000; Uchino, Cacioppo, & Kiecolt-Glaser,
1996). The structure of our social networks (Brissette, Co-
hen, & Seeman, 2000), the support we receive from others
(Cohen et al., 2000), the quality and quantity of our social
interactions (Kiecolt-Glaser & Newton, 2001), and our
feelings of isolation and loneliness (Cacioppo et al., 2002)
have all been identiﬁed as predictors of health and well-
being. However, the interpretation of this literature is
clouded by an approach that often views effects of differ-
ent social environmental and dispositional variables as if
they all are exemplars of the same underlying concept and
mechanism(s). Yet there is increasing evidence that differ-
ent social variables inﬂuence health through entirely differ-
ent and independent mechanisms (Cohen, 1988; Cohen et
al., 2000; Lakey & Cohen, 2000). It is only by understand-
ing the dimensions of our social ties that inﬂuence health
and how they “get under the skin” to do so that psycholo-
gists can successfully apply this knowledge to health pro-
moting interventions. In the service of this goal, I address
ﬁve questions about the association between social ties and
Which characteristics of the social environment are
beneﬁcial for health?
How do these characteristics of our social environ-
ment improve our health?
Can our social environment be destructive to our
Is it the social environment or our personalities that
Can our social environments be changed to improve
These are complex questions, and each deserves consider-
ably more space than I am allocated. I provide preliminary
answers to each question leaning heavily on research and
theory produced in my own laboratory over the last 25
years. For this reason, what I present represents a limited
(but hopefully insightful) approach. My discussion includes
scattered studies of social relationships and mental health,
but my primary focus has been physical health outcomes.
Which Characteristics of the Social Environment Are
Beneﬁcial for Health?
Sociology and social psychology abound with terms that
refer to different properties of the social environment, and
many of these constructs may have implications for health.
However, for both historical and sometimes arbitrary rea-
sons, the literature on social relationship and physical
health is relatively restricted and addresses only a select
group of social constructs. In this section, I focus on two
of these: social integration and social support. They are of
special interest here because I believe that each promotes
health through different mechanisms.
Social support refers to a social network’s provision of
psychological and material resources intended to beneﬁt an
individual’s ability to cope with stress. It is often differen-
tiated in terms of three types of resources: instrumental,
informational, and emotional (e.g., House & Kahn, 1985).
Instrumental support involves the provision of material aid,
for example, ﬁnancial assistance or help with daily tasks.
Informational support refers to the provision of relevant
information intended to help the individual cope with cur-
rent difﬁculties and typically takes the form of advice or
Sheldon Cohen received the Award for Distinguished Sci-
entiﬁc Contributions. Award winners are invited to deliver
an award address at the APA’s annual convention. A ver-
sion of this award address was delivered at the 112th an-
nual meeting, held July 28 –August 1, 2004, in Honolulu,
Hawaii. Articles based on award addresses are reviewed,
but they differ from unsolicited articles in that they are
expressions of the winners’ reﬂections on their work and
their views of the ﬁeld.
676 November 2004
guidance in dealing with one’s problems. Emotional sup-
port involves the expression of empathy, caring, reassur-
ance, and trust and provides opportunities for emotional
expression and venting. Such typologies of support provide
a basis for determining whether the effectiveness of differ-
ent kinds of support differs by the nature of stressful
events or by the characteristics of persons suffering
Social integration is deﬁned as participation in a broad
range of social relationships (Brissette et al., 2000). It is a
multidimensional construct thought to include a behavioral
component—active engagement in a wide range of social
activities or relationships—and a cognitive component—a
sense of communality and identiﬁcation with one’s social
roles (Brissette et al., 2000). The concept of social integra-
tion is rooted in Durkheim’s (1897/1951) seminal work on
social conditions and suicide. Durkheim proposed that sta-
ble social structure and widely held norms are protective
and serve to regulate behavior.
How Do These Characteristics of Our Social
Environment Improve Our Health?
Social factors can promote health through two generic
mechanisms: stress-buffering and main effects (Cohen,
1988; Cohen & Wills, 1985; House, 1981). Speciﬁc pro-
cesses underlying these mechanisms are presented in the
ﬁrst two rows of Table 1.
The primary model considered by psychologists, especially
those interested in intervention, has been stress buffering.
This model asserts that social connections beneﬁt health by
providing psychological and material resources needed to
cope with stress. The model predicts that social support is
beneﬁcial for those suffering adversity but does not play a
role in health for those without highly stressful demands.
Statistically, the stress-buffering model is supported by an
interaction of stress and social support.
Stress is thought to inﬂuence health both by promoting
behavioral coping responses detrimental to health (smok-
ing, drinking alcohol, illicit drug use, sleep loss) and by
activating physiological systems such as the sympathetic
nervous system and the hypothalamic-pituitary-adrenal cor-
tical axis (Cohen, Kessler, & Gordon, 1995). Prolonged or
repeated activation of these systems is thought to place
persons at risk for the development of a range of physical
and psychiatric disorders.
The current literature suggests that the critical factor in
social support operating as a stress buffer is the perception
that others (even one reliable source) will provide appropri-
ate aid (Cohen, 1988; Cohen & Wills, 1985; Uchino et al.,
1996). In this view, the belief that others will provide nec-
essary resources may bolster one’s perceived ability to
cope with demands, thus changing the appraisal of the situ-
ation and lowering its effective stress (Cohen & Wills,
1985; Thoits, 1986; Wethington & Kessler, 1986). Belief
that support is at hand may also dampen the emotional and
physiological responses to the event or alter maladaptive
behavioral responses (e.g., Wills & Cleary, 1996).
In our early work, we proposed that social support is
effective in reducing the effects of stressful events only in
so far as the form of assistance matches demands of the
event (Cohen & Wills, 1985; also see Cutrona & Russell,
1990). For example, having someone lend you money may
be useful in the face of a temporary job loss but useless in
the face of the death of a friend.
There is substantial evidence that the perceived avail-
ability of social support buffers the effect of stress on psy-
chological distress, depression, and anxiety (reviewed by
Cohen & Wills, 1985; Kawachi & Berkman, 2001). For
example, we found that both student and adult samples
reported more symptoms of depression and of physical ail-
ments under stress but that these associations were attenu-
ated among those who perceived that support was available
from their social networks (see Figure 1; Cohen, Mermel-
stein, Kamarck, & Hoberman, 1985). When types of per-
ceived support were broken down, emotional support
Mechanisms Through Which Different Types of Social Constructs Influence Physical Health
Social construct Mechanism Specific processes
Social support Stress buffering Eliminates or reduces effects of stressful experiences by promoting less threatening
interpretations of adverse events and effective coping strategies.
Social integration Main effect
Promotes positive psychological states (e.g., identity, purpose, self-worth, and
positive affect) that induce health-promoting physiological responses. Provides
information and is a source of motivation and social pressure to care for oneself.
Negative interactions Relationships as a
source of stress
Elicits psychological stress and in turn behavior and physiological concomitants that
increase risk for disease.
worked in the face of a variety of types of stressful events,
whereas other types of support (e.g., instrumental, informa-
tional) responded to speciﬁc needs elicited by an event.
The most striking evidence for stress buffering in the
physical health realm is reported in a prospective study of
healthy Swedish men aged 50 years and over (Rosengren,
Orth-Gomer, Wedel, & Wilhelmsen, 1993). Those with
high numbers of stressful life events in the year before the
baseline exam were at substantially greater risk for mortal-
ity over a seven-year follow-up period. However, this ef-
fect was ameliorated among those who perceived that high
levels of emotional support were available to them (see
Figure 2). In contrast, perceived emotional support made
no difference for those with few stressful events. Social
integration did not act as a stress buffer.
Beyond perceptions, the actual receipt of support could
also play a role in stress buffering. Support may alleviate
the impact of stress by providing a solution to the problem,
by reducing the perceived importance of the problem, or
by providing a distraction from the problem. It might also
facilitate healthful behaviors such as exercise, personal hy-
giene, proper nutrition, and rest (cf. Cohen, 1988; House,
1981). My colleagues and I have pursued this issue in an
experimental primate model in which we randomly as-
signed animals to chronically high stress (unstable) or low
stress (stable) social environments and coded their naturally
occurring afﬁliative behaviors (Cohen, Kaplan, Cunnick,
Manuck, & Rabin, 1992). There, 43 male cynomolgus
monkeys were randomly assigned to stable or unstable so-
cial conditions for 26 months. In the stable condition, ani-
mals remained in a single cage with four other animals. In
the unstable condition, the animals were rotated into cages
with (at least three of the four) different animals on a
monthly basis. The proportion of time spent in afﬁliative
behaviors was assessed by 30-minute observations of each
group made twice per week. T-cell immune response was
assessed weekly for three weeks immediately following the
26 months of manipulation. That afﬁliative behavior repre-
sented an attempt to cope with social stress was supported
by greater afﬁliation among animals in the unstable com-
pared with the stable condition. Although immune response
was suppressed among animals in the unstable social con-
dition, animals in the unstable condition with the highest
levels of afﬁliation were partly protected from this stress-
induced suppression (see Figure 3).
The main-effect model argues that social connectedness is
beneﬁcial irrespective of whether one is under stress.
There is reason to believe that social integration operates
primarily through main effects. Individuals who participate
in a social network are subject to social controls and peer
pressures that inﬂuence normative health behaviors. For
example, their networks might inﬂuence whether they exer-
cise, eat low-fat diets, smoke, or take illicit drugs. Integra-
tion may also engender feelings of responsibility for others
resulting in increased motivation to take care of oneself so
that responsibility can be fulﬁlled.
Social integration is also thought to inﬂuence one’s
sense of self and one’s emotional tone. Role concepts that
are shared among a group of people help to guide social
Each Increase in Perceived Availability of Social
Support Is Associated With a Further Reduction in the
Association Between Psychological Stress and
Depressive Symptoms in College Students
Note. CES-D ⫽ Center for Epidemiologic Study of Depression Scale; ISEL ⫽
Interpersonal Support Evaluation List.
Perceived Availability of Emotional Support Buffers the
Association of the Number of Stressful Life Events and
Mortality in Initially Healthy Swedish Men Aged 50
Years and Older
Note. From “Stressful Life Events, Social Support, and Mortality in Men Born in
1933,” by A. Rosengren, K. Orth-Gomer, H. Wedel, and L. Wilhelmsen, 1993,
British Medical Journal, 307, p. 1104. Copyright 1993 by BMJ Publishing
Group. Reprinted with permission.
678 November 2004
interaction by providing a common set of expectations
about how people should act in different roles. In meeting
normative role expectations, individuals gain a sense of
identity, predictability and stability; of purpose; and of
meaning, belonging, security, and self-worth (Cassel, 1976;
Cohen, 1988; Thoits, 1983; Wills, 1985). Interacting with
others is also thought to aid in emotional regulation in-
creasing positive affect and helping limit the intensity and
duration of negative affective states (Cohen, 1988). These
positive cognitions and emotions are presumed to be bene-
ﬁcial because they reduce psychological despair (Thoits,
1983), result in greater motivation to care for oneself (e.g.,
Cohen, 1988), or result in suppressed neuroendocrine re-
sponse and enhanced immune function (Cohen, 1988;
Uchino et al., 1996). Having a wide range of network ties
also provides multiple sources of information that could
inﬂuence health-relevant behaviors, result in more effective
use of available health services, or help one to avoid
stressful or other high-risk situations.
Attention was drawn to social integration as a predictor
of physical health by a report of a prospective study of
residents of Alameda County, California, by Berkman and
Syme (1979). They found that healthy adults who were
more socially integrated (were married, had close family
and friends, belonged to social and religious groups) at
study onset were more likely to still be living at the nine-
year follow-up than their more isolated counterparts (see
Figure 4). The association between social integration and
mortality has since been replicated in over a dozen pro-
spective community-based studies (reviewed by Berkman
& Glass, 2000), whereas other studies have found that
greater integration predicts survival from heart attacks, less
risk for cancer recurrence, less depression and anxiety, and
less severe cognitive decline with aging (see Cohen et al.,
2000). Although both men and women seem to beneﬁt
from social integration, there is evidence that men beneﬁt
more than women (House, Landis, & Umberson, 1988).
This has been attributed to integration having a cost for
women that isn’t experienced by men—the responsibility
that women take for the fate of other network members
(Kessler, McLeod, & Wethington, 1985). However, it
could also be attributable to women being more sensitive
to the quality and content of their relationships than are
men (Coriell & Cohen, 1995). Hence men may beneﬁt
from lower quality relationships, whereas women do not.
I have been particularly interested in the role of social
integration in the body’s ability to ﬁght off infectious dis-
ease. If social integration confers resistance to infectious
illness, this could account for many of the associations dis-
cussed earlier because infections may underlie mortality
and the risk for asthma, certain cancers, and coronary heart
disease. In my studies of the common cold, I assessed psy-
chosocial factors in healthy adults and then experimentally
exposed them to an infectious agent that causes the com-
mon cold. When exposed to such an agent, only about one
third of the subjects developed clinically veriﬁable disease.
Consequently, I can ask whether a particular psychosocial
factor (in this case, social integration) predicts who is re-
sistant to illness. To investigate this issue, my colleagues
and I (Cohen, Doyle, Skoner, Rabin, & Gwaltney, 1997)
developed a measure of social integration that assessed
active (at least once every two weeks) participation in 11
different social roles including being married, a parent, a
parent-in-law, a child, another close family member, a
close neighbor, a friend, a workmate, a schoolmate, a fel-
Affiliative Animals Are Protected From the Effects of
Chronic Social Stress on Cellular Immune Function
Note. CPM ⫽ counts per minute. From “Chronic Social Stress, Affiliation, and
Cellular Immune Response in Nonhuman Primates,” by S. Cohen, J. R. Kaplan,
J. E. Cunnick, S. B. Manuck, and B. S. Rabin, 1992, Psychological Science, 3,p.
303. Copyright 1992 by Blackwell Publishing, Inc. Adapted with permission.
Greater Social Integration Is Associated With Lower
Rates of Mortality
Note. From “Social Networks, Host Resistance, and Mortality: A Nine-Year
Follow-Up Study of Alameda County Residents,” by L. F. Berkman and L. Syme,
1979, American Journal of Epidemiology, 109, p. 190. Copyright 1979 by
Oxford University Press. Adapted with permission.
low volunteer, or a member of groups with or without reli-
gious afﬁliation. We found that greater social integration as
assessed by the numbers of social roles was associated
with less susceptibility to clinical illness (Figure 5). This
association was conﬁrmed across two viruses and was in-
dependent of demographic factors and of immunity to the
experimental virus at baseline (viral-speciﬁc antibody
Although the social integration research I have dis-
cussed is impressive in the reliability of associations with
morbidity and mortality, it does not directly test whether
social integration is operating as a main effect or stress
buffer because none of these studies included measures of
stress. In contrast, the literature on social integration and
psychological well-being indicates that social integration is
associated with better psychological well-being but does
not interact with stress (review by Cohen & Wills, 1985).
A study of stressful life events, social integration, and sur-
vival from breast cancer (Funch & Marshall, 1983) found
that stressful life events were directly associated with lower
rates of survival and social integration was associated with
higher rates. As in the psychological literature, there was
no stress-buffering interaction. Although not reported in the
published work (Cohen et al., 1997), our own research (de-
scribed earlier) similarly failed to ﬁnd an interaction (or
even a consistent pattern) that would support social integra-
tion operating as a buffer. In sum, existing evidence is gen-
erally supportive of social integration inﬂuencing health
independent of whether persons are facing adversity.
Can Our Social Environment Be Destructive to
Social networks provide emotional, informational, and ma-
terial support; regulate behavior; and offer opportunities for
social engagement. They also provide modes of contact to
spread disease and the opportunity for conﬂict, exploita-
tion, stress transmission, misguided attempts to help, and
feelings of loss and loneliness. These potentially negative
aspects of networks can act as psychological stressors re-
sulting in cognitive, affective, and biological responses
thought to increase risk for poor health. The processes un-
derlying this mechanism are summarized in the third row
of Table 1.
Earlier, I discussed the social integration literature as if
being integrated is beneﬁcial and the more integrated one
is, the better it is for one’s health. This argument is sup-
ported by the graded relation between social integration
and health found in many studies (see Figures 4 and 5). An
alternative argument is that it is social isolation that causes
disease. In this view, isolation could be a stressor in its
own right, increasing negative affect and a sense of alien-
ation, loneliness, and stress while decreasing feelings of
control and self-esteem. In turn, these negative psychologi-
cal states could increase neuroendocrine and cardiovascular
responses, suppress immune function, and interfere with
performance of health behaviors (Cacioppo et al., 2002;
Cohen, 1988; Uchino et al., 1996). The hypothesis that
isolation is what is responsible for the reported associations
between social integration and health suggests that there is
some threshold for social contact below which one is at
risk for disease. Interestingly, the social integration litera-
ture can also be interpreted as supporting this hypothesis.
For example, a close examination of Figures 4 and 5 indi-
cates that the most isolated individuals are at greater risk
than one would expect if the relation between social inte-
gration and health was linear. Hence the reliable and strik-
ing associations between social integration and morbidity
and social integration and mortality may be attributable to
both the health-promoting mechanisms associated with in-
tegration and the disease-promoting mechanism that oper-
ates among the most isolated.
In other work on the negative effects of the social envi-
ronment on health, I have examined the inﬂuence of social
conﬂicts on susceptibility to colds. My colleagues and I
(Cohen et al., 1998) assessed whether subjects were in-
volved in serious, enduring (one month or longer) social
conﬂicts using an intensive interview technique. We then
exposed each subject to a virus that causes the common
cold. Those with enduring conﬂicts were more than twice
as likely to develop a cold as persons without any chronic
stressors in their lives. The conﬂicts contributing to this
relationship included enduring problems with spouses,
close family members, and friends. These associations were
Greater Number of Social Roles Is Associated With
Decreased Susceptibility to the Common Cold
Note. From “Social Ties and Susceptibility to the Common Cold,” by S. Cohen,
W. J. Doyle, D. P. Skoner, B. S. Rabin, and J. M. Gwaltney Jr., 1997, Journal of
the American Medical Association, 277, p. 1943. Copyright 1997 by the
American Medical Association. Reprinted with permission.
680 November 2004
independent of preexposure immunity to the virus (speciﬁc
antibodies), demographics, and a host of other controls.
In sum, social environments and one’s responses to
them can have powerful detrimental effects. It is likely that
these effects are primarily mediated through one’s apprais-
als of social conditions as stressful and the consequential
changes in health behaviors, endocrine, immune, and car-
Is It the Social Environment or Our Personalities That
Social personality traits such as attachment, extraversion,
agreeableness, and hostility are thought to play a key role
in molding our social environments. Hence, it is possible
that one or more aspects of these traits account for associa-
tions between social environments and health that I have
discussed earlier. Are social traits responsible (acting as
third or spurious factors) for associations that have tradi-
tionally been attributed to the social environment?
A number of traits that are thought to inﬂuence the de-
gree and quality of one’s social relationships have been
implicated in health outcomes. For example, hostile people
are at greater risk for coronary artery disease and possibly
other physical health problems (Smith, 1992), and argu-
ments (but as yet little evidence) have been recently made
that those with more secure attachment styles may be
partly protected from disease risk. My work has focused on
sociability, a disposition that is generally recognized as a
determinant of quality and quantity of social interaction. I
deﬁne sociability as “the quality of seeking others and be-
My colleagues and I found that sociability was associ-
ated with greater resistance to developing colds when per-
sons were experimentally exposed to a cold virus (Cohen,
Doyle, Turner, Alper, & Skoner, 2003). That this associa-
tion was found after entering multiple controls (covariates
in the equation) including preexisting immunity (speciﬁc
antibody), gender, age, education, season of the year, body
mass index, and type of virus is notable.
If sociability is responsible for our creating and main-
taining of social ties, and sociability is itself directly bene-
ﬁcial to health, then the associations of social ties and
health we discussed earlier may just be a reﬂection of the
beneﬁts of sociability and not attributable to our social ties
and interactions. This is especially important because it
suggests that changing social ties would not inﬂuence
health outcomes. There is limited evidence in regard to this
question. When my colleagues and I controlled for markers
of sociability when predicting health from social integra-
tion or from social conﬂicts (Cohen et al., 1997, 1998), the
associations between the social environmental variables and
disease susceptibility were reduced but remained substan-
tial and signiﬁcant.
In an earlier study, we were interested in determining
whether other dispositional factors might account for per-
ceived social support inﬂuences in stress buffering (Cohen,
Sherrod, & Clark, 1986). As expected, we found that per-
ceived social support buffered the effects of psychological
stress on depression. This association held up even after
we controlled for both the main effects and interactions
with stress of social competence, social anxiety, and the
tendency to disclose intimate feelings to others. Similarly,
Rook (1984) found that the association between having ties
who were sources of negative interactions and poorer psy-
chological well-being was not altered after controlling for
indices of social competence. In sum, social personality
traits do play a role in health. However, the few studies
that address the overlap of these traits with social integra-
tion, perceived social support, and negative interaction sug-
gest that the trait and environmental effects are at least
partly independent of one another (also see Uchino et al.,
Can Our Social Environments Be Changed to Improve
The provocative and consistent positive associations of so-
cial integration and supports with physical health discussed
in this article are based entirely on the measures of rela-
tionships within natural social networks. In contrast, pub-
lished interventions are based almost entirely on support
provided by strangers. Moreover, evidence of the effective-
ness of social interventions is much weaker than the evi-
dence from correlational studies. There are fewer studies,
and overall their results have been disappointing (Cohen et
Social support interventions have often been aimed at
improving health outcomes in patients with serious, life-
threatening diseases. Patients are approached after an acute
event (heart attack, stroke) or in the course of a chronic,
debilitating disease (cancer, HIV) and offered an interven-
tion aimed at reducing secondary events and improving
function. Most often, patients are offered peer support
groups. Early studies, though often small, were sometimes
promising (e.g., Fawzy et al., 1993; Frasure-Smith &
Prince, 1989); one showed, for instance, that women with
metastatic breast cancer who were offered group psycho-
therapy lived longer (Spiegel, Bloom, & Kraemer, 1989).
However, as trials became larger, followed more rigorous
protocols, and involved multiple sites and interventionists,
the promising results were not replicated. Recently, for
example, several clinical trials with postmyocardial infarc-
tion patients (e.g., The ENRICHD Investigators, 2003; Fra-
sure-Smith et al., 1997) and metastatic breast cancer pa-
tients (e.g., Cunningham et al., 1998; Goodwin et al.,
2001) have found no effects of social support interventions
on recurrent disease or mortality.
Our own intervention work similarly failed to support
the effectiveness of providing seriously ill patients with
emotional support provided by a support group of peers. In
a clinical trial led by my colleague Vicki Helgeson (Helge-
son, Cohen, Schulz, & Yasko, 1999), we found that the
provision of information, but not “peer” emotional support,
facilitated psychological and physical adjustment to breast
cancer. More important, the effectiveness of providing dif-
ferent types of support was dependent on individual differ-
ences in the strength and nature of natural support systems
(Helgeson, Cohen, Schulz, & Yasko, 2000). For example,
peer emotional support groups helped women who lacked
support from their partners or physicians but harmed
women who had high levels of natural support.
In sum, the existing intervention literature has been dis-
appointing. In general, investigators have approached a
complex problem with good intentions but often without
deep theoretical analysis or a strong base of prior research.
My intent here is not to suggest that social network and
support interventions are not potentially effective. On the
contrary, it is to encourage the design of interventions that
better reﬂect what is known from existing research on the
role of social relationships in health.
The work discussed in this article highlights a number
of directions for future interventions. First, it suggests a
special emphasis on intervention in natural social networks.
Emphasis on natural networks is consistent with the corre-
lational evidence presented in this article but is also more
promising than social support groups as a means of making
changes in social relationships that one can rely on months
and years beyond the period of active intervention (see
Gottlieb, 2000, on when to intervene in the natural social
Second, it suggests intervening in several different as-
pects of the social environment. These include (a) increas-
ing the availability of social support within existing social
networks by improving individual social skills or by build-
ing stronger ties to existing network members, (b) increas-
ing social integration by creating and nurturing close and
peripheral ties between an individual and his or her com-
munity, and (c) reducing negative interactions.
Finally, it suggests that researchers need to identify the
characteristics of those who beneﬁt most and least from
social integration and support interventions. For example,
variability in the effectiveness of support may be attribut-
able to differences in the participants’ social skills, social
traits, or network relationships. It is conceivable that only
the most socially isolated individuals reap signiﬁcant bene-
ﬁts from their contact with new sources of support. Or it
may be that those who feel the greatest emotional isolation,
or experience the most conﬂict with their existing associ-
ates, are the least able to take advantage of new opportuni-
ties to strengthen their network. As suggested by our work
with breast cancer patients, this may not merely be an is-
sue of identifying who will beneﬁt but also who may be
harmed by interventionists’ positive intentions.
In sum, I have argued that three different social relation-
ship variables—social integration, social support, and nega-
tive interaction—are all associated with health outcomes,
that these variables each inﬂuence health through different
mechanisms, and that these associations are not spurious
ﬁndings attributable to personality and hence they are
likely subject to intervention. Further, I feel that this litera-
ture suggests a broader view of how to intervene that in-
cludes creating and strengthening a diverse natural social
network, increasing the availability of social support in
natural networks, and reducing negative interactions within
one’s network. Finally, I emphasize the necessity to under-
stand more about who beneﬁts most and least from social
This research was supported by National Institute of Men-
tal Health (NIMH) Research Scientist Development Awards
K02 MH00721 and K05 MH00721, National Cancer Insti-
tute Grants CA38243 and CA61303, National Institute of
Allergies and Infectious Diseases Grant AI23072, National
Heart Lung and Blood Institute Grant HL29547, and
NIMH Grants MH47234 and MH50429. Preparation of
this article was facilitated by Pittsburgh Mind–Body Center
Grants HL65111 and HL65112.
I thank my colleagues on the Robert Wood Johnson
Planning Group on Social Connectedness and Health, Lisa
Berkman, John Cacioppo, Tom Cook, Robert Rose, and
John Sheridan; my collaborators on an edited volume on
social support, Ben Gottlieb and Lyn Underwood, for stim-
ulating discussions of the status of this research literature
and its implications for intervention; and my fellow mem-
bers of the MacArthur Foundation Network on Socioeco-
nomic Status and Health. I thank Vicki Helgeson, Tamar
Krishnamurti, and Tom Wills for their helpful comments on
an earlier draft.
Correspondence concerning this article should be ad-
dressed to Sheldon Cohen, Department of Psychology, Car-
negie Mellon University, Pittsburgh, PA 15213. E-mail:
Berkman, L. F., & Glass, T. (2000). Social integration,
social networks, social support, and health. In L. F. Berk-
man & I. Kawachi (Eds.), Social epidemiology (pp. 137–
173). New York: Oxford University Press.
Berkman, L. F., & Syme, L. (1979). Social networks, host
resistance, and mortality: A nine-year follow-up study of
682 November 2004
Alameda county residents. American Journal of Epidemiol-
ogy, 109, 186 –203.
Brissette, I., Cohen, S., & Seeman, T. E. (2000). Measur-
ing social integration and social networks. In S. Cohen, L.
Underwood, & B. Gottlieb (Eds.), Measuring and interven-
ing in social support (pp. 53– 85). New York: Oxford Uni-
Cacioppo, J. T., Hawkley, L. C., Crawford, E., Ernst,
J. M., Burleson, M. H., Kowalewski, R. B., et al. (2002).
Loneliness and health: Potential mechanisms. Psychoso-
matic Medicine, 64, 407– 417.
Cassel, J. (1976). Social science in epidemiology: Psycho-
social processes and stress theoretical formulation. In E. L.
Struening & M. Guttentag (Eds.), Handbook of evaluation
research (Vol. 1, p. 537–549). London: Sage.
Cohen, S. (1988). Psychosocial models of social support in
the etiology of physical disease. Health Psychology, 7,
Cohen, S., Doyle, W. J., Skoner, D. P., Rabin, B. S., &
Gwaltney, J. M., Jr. (1997). Social ties and susceptibility to
the common cold. Journal of the American Medical Asso-
ciation, 277, 1940 –1944.
Cohen, S., Doyle, W. J., Turner, R. B., Alper, C. M., &
Skoner, D. P. (2003). Sociability and susceptibility to the
common cold. Psychological Science, 14, 389 –395.
Cohen, S., Frank, E., Doyle, W. J., Skoner, D. P., Rabin,
B. S., & Gwaltney, J. M., Jr. (1998). Types of stressors
that increase susceptibility to the common cold in adults.
Health Psychology, 17, 214 –223.
Cohen, S., Gottlieb, B., & Underwood, L. (2000). Social
relationships and health. In S. Cohen, L. Underwood, & B.
Gottlieb (Eds.), Measuring and intervening in social sup-
port (pp. 3–25). New York: Oxford University Press.
Cohen, S., Kaplan, J. R., Cunnick, J. E., Manuck, S. B., &
Rabin, B. S. (1992). Chronic social stress, afﬁliation, and
cellular immune response in nonhuman primates. Psycho-
logical Science, 3, 301–304.
Cohen, S., Kessler, R. C., & Gordon, L. U. (1995). Strate-
gies for measuring stress in psychiatric and physical disor-
ders. In S. Cohen, R. C. Kessler, & L. U. Gordon (Eds.),
Measuring stress (pp. 3–28). New York: Oxford University
Cohen, S., Mermelstein, R., Kamarck, T., & Hoberman, H.
(1985). Measuring the functional components of social sup-
port. In I. G. Sarason & B. R. Sarason (Eds.), Social sup-
port: Theory, research and application (pp. 73–94). The
Hague, Holland: Martinus Nijhoff.
Cohen, S., Sherrod, D. R., & Clark, M. S. (1986). Social
skills and the stress-protective role of social support. Jour-
nal of Personality and Social Psychology, 50, 963–973.
Cohen, S., & Wills, T. A. (1985). Stress, social support,
and the buffering hypothesis. Psychological Bulletin, 98,
Coriell, M., & Cohen, S. (1995). Concordance in the face
of a stressful event: When do members of a dyad agree
that one person supported the other? Journal of Personality
and Social Psychology, 69, 289 –299.
Cunningham, A. J., Edmonds, C. V. I., Jenkins, G. P., Pol-
lack, H., Lockwood, G. A., & Warr, D. (1998). A random-
ized controlled trial of the effects of group psychological
therapy on survival in women with metastatic breast can-
cer. Psycho-oncology, 7, 508 –517.
Cutrona, C. E., & Russell, D. W. (1990). Type of social
support and speciﬁc stress: Toward a theory of optimal
matching. In B. R. Sarason, I. G. Sarason, & G. R. Pierce
(Eds.), Social support: An interactional view (pp. 319 –
366). New York: Wiley.
Durkheim, E. (1951). Suicide. New York: Free Press.
(Original work published 1897)
The ENRICHD Investigators. (2003). Effects of treating
depression and low perceived social support on clinical
events after myocardial infarction: The enhancing recovery
in coronary heart disease patients (ENRICHD) randomized
trial. Journal of the American Medical Association, 28,
Fawzy, F. I., Fawzy, N. W., Hyun, C. S., Elashoff, R.,
Guthrie, D., Fahey, J. L., & Morton, D. L. (1993). Malig-
nant melanoma: Effects of an early structured psychiatric
intervention, coping, and affective state on recurrence and
survival 6 years later. Archives of General Psychiatry, 50,
Frasure-Smith, N., L’Esperance, F., Prince, R. H., Verrier,
P., Garber, R., Juneau, M., et al. (1997). Randomized trial
of home-based psychosocial nursing: Intervention for pa-
tients recovering from myocardial infarction. The Lancet,
350, 473– 479.
Frasure-Smith, N., & Prince, R. (1989). Long-term fol-
low-up of the Ischemic Heart Disease Life Stress Monitor-
ing Program. Psychosomatic Medicine, 51, 485–513.
Funch, D. P., & Marshall, J. (1983). The role of stress,
social support, and age in survival from breast cancer.
Journal of Psychosomatic Research, 27, 77– 83.
Goodwin, P. J., Leszcz, M., Ennis, M., Koopmans, J., Vin-
cent, L., Guther, H., et al. (2001). The effect of group psy-
chosocial support on survival in metastatic breast cancer.
New England Journal of Medicine, 345, 1719 –1726.
Gottlieb, B. H. (2000). Selecting and planning support in-
terventions. In S. Cohen, L. Underwood, & B. H. Gottlieb
(Eds.), Social support measurement and interventions: A
guide for health and social scientists (pp. 195–220). New
York: Oxford University Press.
Helgeson, V. S., Cohen, S., Schulz, R., & Yasko, J.
(1999). Effects of education and peer discussion group in-
terventions on 6-month adjustment to Stage I and II breast
cancer. Archives of General Psychiatry 56, 340 –347.
Helgeson, V. S., Cohen, S., Schulz, R., & Yasko, J.
(2000). Group support interventions for people with cancer:
Who beneﬁts from what? Health Psychology, 19, 107–114.
House, J. S. (1981). Work stress and social support. Read-
ing, MA: Addison-Wesley.
House, J. S., & Kahn, R. L. (1985). Measures and concepts
of social support. In S. Cohen & S. L. Syme (Eds.), Social
support and health (pp. 83–108). New York: Academic
House, J. S., Landis, K. R., & Umberson, D. (1988, July
29). Social relationships and health. Science, 241, 540 –
Kawachi, I., & Berkman, L. F. (2001). Social ties and
mental health. Journal of Urban Health: Bulletin of the
New York Academy of Medicine, 78, 458 – 467.
Kessler, R. C., McLeod, J. D., & Wethington, E. (1985).
The costs of caring: A perspective on the relationship be-
tween sex and psychological distress. In I. G. Sarason &
B. R. Sarason (Eds.), Social support: Theory, research, and
applications (pp. 491–507). Boston: Martinus Nijhoff.
Kiecolt-Glaser, J. K., & Newton, T. L. (2001). Marriage
and health: His and hers. Psychological Bulletin, 127, 472–
Lakey, B., & Cohen, S. (2000). Support theory and mea-
surement. In S. Cohen, L. Underwood, & B. H. Gottlieb
(Eds.), Social support measurement and interventions: A
guide for health and social scientists (pp. 29 –52). New
York: Oxford University Press.
Rook, K. S. (1984). The negative side of social interaction:
Impact on psychological well-being. Journal of Personality
and Social Psychology, 46, 1097–1108.
Rosengren, A., Orth-Gomer, K., Wedel, H., & Wilhelmsen,
L. (1993). Stressful life events, social support, and mortal-
ity in men born in 1933. British Medical Journal, 307,
Smith, T. W. (1992). Hostility and health: Current status of
a psychosomatic hypothesis. Health Psychology, 11, 139 –
Spiegel, D., Bloom, J. R., & Kraemer, H. C. (1989). Effect
of psychosocial treatment on survival of patients with met-
astatic breast cancer. The Lancet, 2, 888 – 891.
Thoits, P. A. (1983). Multiple identities and psychological
well-being: A reformation and test of the social isolation
hypothesis. American Sociological Review, 48, 174 –187.
Thoits, P. A. (1986). Social support as coping assistance.
Journal of Consulting and Clinical Psychology, 54, 416 –
Uchino, B. N., Cacioppo, J. T., & Kiecolt-Glaser, J. K.
(1996). The relationship between social support and physi-
ological processes: A review with emphasis on underlying
mechanisms and implications for health. Psychological
Bulletin, 119, 488 –531.
Wethington, E., & Kessler, R. C. (1986). Perceived sup-
port, received support, and adjustment to stressful life
events. Journal of Health and Social Behavior, 27, 78 – 89.
Wills, T. A. (1985). Supportive functions of interpersonal
relationships. In S. Cohen & S. L. Syme (Eds.), Social
support and health (pp. 61– 82). New York: Academic
Wills, T. A., & Cleary, S. D. (1996). How are social sup-
port effects mediated: A test for parental support and ado-
lescent substance use. Journal of Personality and Social
Psychology, 71, 937–952.
684 November 2004