Social Networks and Incident Stroke Among Women With Suspected Myocardial Ischemia

Abstract

To describe the prospective relationship between social networks and nonfatal stroke events in a sample of women with suspected myocardial ischemia. Social networks are an independent predictor of all-cause and cardiovascular mortality, but their relationship with stroke events in at-risk populations is largely unknown.
A total of 629 women (mean age = 59.6 +/- 11.6 years) were evaluated at baseline for cardiovascular disease risk factors as part of a protocol including coronary angiography; the subjects were followed over a median 5.9 years to track the incidence of cardiovascular events including stroke. Participants also completed the Social Network Index (SNI), measuring the presence/absence of 12 types of common social relationships.
Stroke events occurred among 5.1% of the sample over follow-up. More isolated women were older and less educated, with higher rates of smoking and hypertension, and increased use of cardiovascular medications. Women with smaller social networks were also more likely to show elevations (scores of > or =10) on the Beck Depression Inventory (54% versus 41%, respectively; p = .003). Relative to women with higher SNI scores, Cox regression results indicated that more isolated women experienced strokes at greater than twice the rate of those with more social relationships after adjusting for covariates (hazard ratio = 2.7; 95% Confidence Interval = 1.1-6.7).
Smaller social networks are a robust predictor of stroke in at-risk women, and the magnitude of the association rivals that of conventional risk factors.

Full text

Social Networks and Incident Stroke Among Women With Suspected
Myocardial Ischemia
THOMAS RUTLEDGE,PHD, SARAH E. LINKE, BA, MARIAN B. OLSON, MS, JENNIFER FRANCIS,PHD, B. DELIA JOHNSON,PHD,
VERA BITTNER, MD, MSPH, KAKI YORK,PHD, CANDACE MCCLURE,PHD, SHERYL F. KELSEY,PHD, STEVEN E. REIS,MD,
CAROL E. CORNELL,PHD, VIOLA VACCARINO, MD, PHD, DAVID S. SHEPS, MD, MSPH, LESLEE J. SHAW,PHD,
DAVID S. KRANTZ,PHD, SUSMITA PARASHAR, MD, MPH, MS, AND C. NOEL BAIREY MERZ,MD
Objective: To describe the prospective relationship between social networks and nonfatal stroke events in a sample of women with
suspected myocardial ischemia. Social networks are an independent predictor of all-cause and cardiovascular mortality, but their
relationship with stroke events in at-risk populations is largely unknown. Method: A total of 629 women (mean age ⫽59.6 ⫾11.6
years) were evaluated at baseline for cardiovascular disease risk factors as part of a protocol including coronary angiography; the
subjects were followed over a median 5.9 years to track the incidence of cardiovascular events including stroke. Participants also
completed the Social Network Index (SNI), measuring the presence/absence of 12 types of common social relationships. Results:
Stroke events occurred among 5.1% of the sample over follow-up. More isolated women were older and less educated, with higher
rates of smoking and hypertension, and increased use of cardiovascular medications. Women with smaller social networks were also
more likely to show elevations (scores of ⱖ10) on the Beck Depression Inventory (54% versus 41%, respectively; p⫽.003).
Relative to women with higher SNI scores, Cox regression results indicated that more isolated women experienced strokes at greater
than twice the rate of those with more social relationships after adjusting for covariates (hazard ratio ⫽2.7; 95% Confidence
Interval ⫽1.1– 6.7). Conclusions: Smaller social networks are a robust predictor of stroke in at-risk women, and the
magnitude of the association rivals that of conventional risk factors. Key words: social networks, coronary artery disease, women,
prospective, stroke.
CAD ⫽coronary artery disease; SES ⫽socioeconomic status;
CVD ⫽cardiovascular disease; WISE ⫽Women’s Ischemia Syndrome
Evaluation; PCI ⫽percutaneous coronary intervention; CABG ⫽
coronary artery bypass graft; SNI ⫽Social Network Index; BDI ⫽
Beck Depression Inventory; HR ⫽hazard ratio.
INTRODUCTION
Stroke is a leading cause of morbidity and mortality in the
US, trailing only coronary artery disease (CAD) as a spe-
cific cause of death (1–3). The burden of stroke is dispropor-
tionately carried by women, who account for ⬎60% of total
stroke deaths (3). Known risk factors for the development and
prevention of stroke parallel those of CAD, including hyper-
tension, diabetes, smoking, obesity, and dyslipidemia, among
others (4 – 6). However, whereas the relationship between
CAD and psychosocial factors such as low socioeconomic
status (SES), depression, and social relationships is supported
by a large empirical literature (7–11), specific associations
between psychosocial factors and cerebrovascular disease
(CVD) incidence are comparatively rare (12–16).
This study prospectively examined the relationship be-
tween social networks and stroke over a median 5.9-year
follow-up interval among a clinical sample of women with
suspected myocardial ischemia. Women completed a measure
of social networks as part of a protocol including a coronary
angiogram and a CVD risk factor assessment.
METHOD
Participant Recruitment and Entrance Criteria
Women were eligible for participation in the Women’s Ischemia Syndrome
Evaluation (WISE) study if they were ⬎18 years and were referred for a
coronary angiogram to evaluate suspected myocardial ischemia (16). The
WISE study was designed to improve the understanding and diagnosis of
ischemic heart disease in women. Exclusion criteria included major comor-
bidity compromising follow-up, pregnancy, contraindication to provocative
diagnostic testing, cardiomyopathy, severe heart failure, recent myocardial
infarction or revascularization procedures, significant valvular or congenital
heart disease, and language barrier. Data for WISE were collected between
1996 and 2005. All participants provided written informed consent, and
Institutional Review Board approval was obtained for all participating sites.
Measurement of CAD and Clinical Outcome Events
Quantitative analysis of coronary angiograms was performed at the WISE
Angiographic Core Laboratory (Rhode Island Hospital, Providence, Rhode
Island) by investigators blinded to all other subject data (17). Luminal
diameter was measured at all stenoses and at nearby reference segments, using
an electronic cine projector-based “cross-hair” technique (Vanguard Instru-
ment Corporation, Melville, New York). A CAD severity score was also
developed by assigning increasing points to increasing percent stenosis (0 –19,
20 –49, 50 – 69, 70 – 89, 90–98, 99 –100), after adjusting for presence of
collaterals (filling of the occluded vessel or its distal branches anterograde or
retrograde via channels other than the original lumen). Lesion location was
From the Department of Psychiatry, VA San Diego Healthcare System
(T.R.), San Diego, California; Department of Psychiatry, (T.R.), University of
California, San Diego, California; Department of Psychology, (S.E.L.), San
Diego State University/University of California, San Diego, San Diego,
California; Department of Epidemiology, Joint Doctoral Program in Clinical
Psychology (M.B.O., B.D.J., C.M., S.F.K., S.E.R.), University of Pittsburgh,
Pennsylvania; Department of Medical and Clinical Psychology, Uniformed
Services University of the Health Sciences (J.F., D.S.K.), Bethesda, MD;
Department of Medicine, (V.B.), University of Alabama at Birmingham,
Birmingham, Alabama; Department of Medicine, (K.Y., D.S.S.), University
of Florida, Gainesville, Florida; Department of Medicine, (D.S.S.), North
Florida/South Georgia VA Healthcare System, Gainesville, Florida; Depart-
ment of Medicine, (C.E.C., L.J.S.), University of Arkansas for Medical
Sciences, Little Rock, Arkansas; Department of Medicine, (V.V., S.P.),
Emory University, Atlanta, Georgia; and Department of Medicine,
(C.N.B.M.), Cedars-Sinai Medical Center, Los Angeles, California.
Address correspondence and reprint requests to Thomas Rutledge, Psychol-
ogy Service 116B , VA San Diego Healthcare System, Medical Center, 3350
La Jolla Village Drive, San Diego, CA 92161. E-mail: Thomas.Rutledge@
va.gov
All authors participated in the conceptual development and writing of this
manuscript. None of the authors have conflicts of interest with the contents of
this manuscript.
This work was supported by Contracts N01-HV-68161, N01-HV-68162,
N01-HV-68163, N01-HV-68164 from the National Heart, Lung, and Blood
Institutes; GCRC Grant M01-RR00425 from the National Center for Research
Resources; and grants from the Gustavus and Louis Pfeiffer Research Foun-
dation, The Women’s Guild, Cedars-Sinai Medical Center, and the Ladies
Hospital Aid Society of Western Pennsylvania, and QMED, Inc.
Received for publication May 25, 2007; revision received October 24, 2007.
DOI: 10.1097/PSY.0b013e3181656e09
282 Psychosomatic Medicine 70:282–287 (2008)
0033-3174/08/7003-0282
Copyright © 2008 by the American Psychosomatic Society

taken into account in the scoring, with more proximal lesions receiving higher
weighting (18).
Women were contacted at 6 weeks post baseline and annually thereafter
for a median of 5.9 years (25
th
percentile ⫽2.5 years; 75
th
percentile ⫽6.9
years) to track subsequent cardiovascular events. Follow-up consisted of a
scripted telephone interview by an experienced nurse or physician who
inquired about hospitalization, treatment, or occurrence of myocardial infarc-
tion, congestive heart failure, and stroke. In the event of death, a death
certificate was obtained and reviewed by a blinded WISE physician for
classifying the cause of death. Subtypes of stroke were not differentiated.
Cardiovascular Risk Factor Measurement
Major CVD risk factors in the WISE protocol included smoking (dichot-
omized as current versus former or never smokers), history of dyslipidemia,
history of diabetes, history of hypertension, and waist circumference. Risk
factors were assessed by physical examination (waist circumference), self-
report (smoking), and diagnosis and treatment history (dyslipidemia, diabetes,
hypertension). Women were also assessed for medications used for treatment
of CVD risk factors, including aspirin, lipid lowering medications (statin and
nonstatin agents), and cardiovascular medications (including angiotensin-
converting enzyme (ACE) inhibitors, angiotensin receptor blockers, diuretics,
and vasodilators). For analytic purposes, the multiple lipid and hypertension
medications were simplified into a pair of dichotomous variables (i.e., sepa-
rate yes/no variables for using lipid and cardiovascular medications). Active
treatment in the latter categories was defined by use in the previous week.
Physical measurements of blood pressure, blood glucose levels, and choles-
terol were also collected, but the substitution of these measurements for
treatment history reports made no differences in event analyses. The partic-
ipants’ reported education history was dichotomously coded to indicate less
than high school graduate versus high school diploma or greater. Education is
a stable measure of SES (19). Women’s race was also coded dichotomously
(0 ⫽White, 1 ⫽non-White). Only 1.2% of the sample identified themselves
as other than African-American or White.
Psychosocial Measures
Participants’ baseline responses to the Social Network Index (SNI) (20)
were used to measure social networks. The SNI has been used to predict
inflammation in the Framingham and Third National Health and Nutrition
Examination Survey cohorts (21,22), and total mortality outcomes in WISE
(23). The SNI collects information on 12 types of social relationships,
including friends, employment, neighbors, marriage partners, belonging to a
church, children, parents, in-laws, other relatives, class attendance (e.g.,
university), volunteer work, and group memberships. Scoring of the SNI
produces a measure of social network diversity based on the presence or
absence of each of the 12 relationship domains over a 2-week period, with
scores ranging from 0 to 12.
Participants also completed the Beck Depression Inventory (BDI) to
measure depression symptom severity (24). The BDI is a 21-item question-
naire that has been validated in many clinical populations and linked to poor
CAD outcomes (8).
Statistical Analyses
Descriptive statistics, ttests, and
␹
2
statistics were used to make compar-
isons of more versus less isolated women on CVD risk factors (smoking
history, waist circumference, history of diabetes, dyslipidemia, and hyperten-
sion), demographic characteristics (age, ethnicity, education), angiographic
CAD severity score, and BDI scores. We sequentially built Cox regression
models to adjust for demographic factors, BDI scores, CVD risk factors, and
CAD severity scores. To correct for skewing, angiographic CAD severity
scores were log transformed before inclusion in the analyses.
We first computed hazard ratios (HRs) for social network scores in
continuous form, followed by a secondary analysis using the SNI in categor-
ical form using high and low scorers based on a median split (scores of ⬎6
were above the median), wherein women with larger social networks served
as the reference category. We chose a dichotomous breakdown to maintain
acceptable sample sizes that would have been compromised with additional
groups. For a more detailed graphical display of the SNI-stroke relationship,
we also created quartile groups, corresponding to SNI scores in the ranges of
1 to 5, 6, 7 to 8, and 9 to 11 in this sample. In the hazard models, stroke-free
participants were censored at their last completed follow-up date. Model fit
and validation were assessed on a logistic model containing all covariates
using the goodness-of-fit test (Hosmer-Lemeshow
␹
2
statistic) and the
Shrunken R
2
statistic. Assumptions of equal proportionality held for the Cox
Regression models.
Finally, because the initial analyses indicated that the SNI-stroke relation-
ship contained a nonlinear component, we also examined the SNI-stroke
relationship by completing Cox regression models with both linear and
nonlinear SNI components (including quadratic, cubic, exponential, power,
inverse, S-curve, and logarithmic transformations). All analyses were com-
pleted, using SPSS version 12.0 (SPSS Inc., Chicago, Illinois), with the
criterion for statistical significance set at .05.
RESULTS
A total of 936 women were enrolled in WISE. From this
group, 297 subjects were enrolled before the initiation of the
psychosocial battery that included the SNI and BDI. An ad-
ditional 10 women were removed due to an absence of fol-
low-up data, leaving a total of 629 participants available for
analysis. Thirty-one nonfatal and one fatal stroke events (5.1%
of sample) were reported over a median 5.9 years of follow-
up. Women categorized by SNI scores differed systematically
(Table 1). Women with lower SNI scores were significantly
older, had lower education levels, and were in poorer health as
documented by CVD risk factors. More isolated women also
had higher rates of depression with 57% versus 43% reporting
BDI scores of ⱖ10 (p⫽.003). Stroke occurred among 4.4%
(17/290) of women without SNI data, and these women did
not differ significantly from those with SNI scores on any risk
factor listed in Table 1 (data not shown).
Medications were commonly prescribed for CVD risk fac-
tor management. A total of 46.4% reported using one or more
cardiovascular medications, and usage rates were higher
among more socially isolated women (56.7% versus 42%; p⫽
.008). There were no differences between SNI groups on rates
of aspirin use or use of lipid-lowering medications.
Social Networks and Stroke Events
Before covariate adjustment, the presence of each addi-
tional relationship on the SNI was associated with a 23%
decrease in stroke risk (HR ⫽0.77; 95% Confidence Interval
(CI) ⫽0.62– 0.94). Church membership and “other friend-
ships” (comprised of friends not linked to SNI item catego-
ries) had the strongest inverse relationships to stroke among
the specific relationship domains measured by the SNI (un-
adjusted RR ⫽0.44, 0.45; 95% CI ⫽0.22– 0.91, 0.21– 0.98,
respectively. Nonchurch members and those without other
friendships served as the reference categories). None of the
individual SNI items was a reliable predictor of stroke after
covariate adjustment. After adjusting for age, education, eth-
nicity, and BDI scores, SNI scores continued to predict stroke
events (HR ⫽0.78; 95% CI ⫽0.63– 0.97). However, this
relationship was no longer significant after adjusting for CVD
risk factors and CAD severity scores (HR ⫽0.82; 95% CI ⫽
0.63–1.07).
SOCIAL NETWORKS AND STROKE
283Psychosomatic Medicine 70:282–287 (2008)

As summarized in Table 2, after including demographic
variables, BDI scores, CVD risk factors, and CAD severity
scores, the relationship between social networks and stroke
was more stable using a dichotomized version of the SNI
scores. In the latter model, more isolated women experienced
strokes at more than twice the rate (HR ⫽2.7; 95% CI ⫽
1.1– 6.5) of those with higher SNI scores. In the final model,
smoking history (HR ⫽3.0; 95% CI ⫽1.2–7.5 for current
versus former or never smokers, respectively) and CAD se-
verity scores (HR ⫽2.1; 95% CI ⫽1.2–3.9) were the only
factors other than SNI scores to predict stroke events. There
was no evidence of a lack of fit in the regression models. The
Hosmer-Lemeshow statistic was nonsignificant (p⬎.80),
indicating good fit, and the population R
2
estimate derived
from the Shrunken R
2
statistic (0.11) differed only slightly
from that observed in the final model including all covariates
(R
2
⫽0.12).
Figure 1 provides a possible explanation for the weaker
linear versus categorical findings, suggesting that the relation-
ship between social network scores and stroke contained a
nonlinear component. Subsequent Cox regression results,
however, in which we tested linear and nonlinear models
Figure 1. Stroke rates (%) across Social Network Index (SNI) quartiles
(quartile values ⫽0 –5, 6, 7–8, and ⬎8). Group 1 represents the socially
isolated women.
TABLE 1. Mean ⴞStandard Deviation Values (Unless Otherwise Indicated) and Stroke Risk Factors Among Women Categorized by Social
Network Index (SNI) Scores (nⴝ629)
a
Low SNI (n⫽188)
b
High SNI (n⫽441) p
Age 60.8 ⫾10.9 56.6 ⫾11.2 ⬍.001
Race (% non-White) 19.1 14.5 .12
Percent completing high school 70.2 87.8 ⬍.001
Beck Depression Inventory 12.4 ⫾9.3 9.6 ⫾7.5 ⬍.001
Coronary artery disease severity score 14.9 ⫾13.5 12.5 ⫾11.9 .04
History of hypertension (%) 64.2 53.3 .04
History of diabetes (%) 28.2 20.1 .06
History of dyslipidemia (%) 56.7 50.6 .26
Smoking history (%)
Never smoker 34 54.1 ⬍.001
Former smoker 42 29.5 ⬍.01
Current smoker 23.9 16.4 ⬍.05
Waist circumference (inches) 37.7 ⫾7.7 35.4 ⫾6.5 ⬍.001
Cardiovascular disease medications (%)
c
56.7 42.0 .008
Aspirin (%) 61.1 58.5 .72
Lipid-lowering medications (%) 35.6 29.3 .41
Stroke events (n(%)) 16 (8.5%) 16 (3.6%) .006
a
Group differences evaluated with tests of means (t tests) and categories (
␹
2
).
b
Low SNI scores consisted of women with a scale score of ⱕ6.
c
Includes use of angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, diuretics, and vasodilators.
TABLE 2. Cox Regression Model Describing the Relationship Between Social Networks And Incident Stroke (nⴝ629)
Dependent Variable: Total Stroke Events HR Estimate for Low Versus
High SNI Scorers
Lower
95% CI
Upper
95% CI
Unadjusted SNI association 2.5 1.3 5.1
SNI—adjusted for demographics
a
2.3 1.1 4.7
Adjusted for demographics and CAD risk factors
b
2.7 1.1 6.5
HR ⫽hazard ratio; SNI ⫽Social Network Index; CI ⫽confidence interval; CAD ⫽coronary artery disease.
a
Includes age, education history, ethnicity, and Beck Depression Inventory scores.
b
Includes diabetes, smoking, dyslipidemia and hypertension histories, waist-circumference, and CAD severity score.
T. RUTLEDGE et al.
284 Psychosomatic Medicine 70:282–287 (2008)

using a variety of nonlinear SNI transformations, failed to
support this hypothesis, indicating that none of the quadratic,
cubic, exponential, power, inverse, S-curve, or logarithmic
transformations added to the model after controlling for the
linear SNI effect.
DISCUSSION
This study is among the first to demonstrate a prospective
relationship between social relationships and the risk of stroke
in a clinical sample of women with suspected CAD. These results
are consistent with previous studies using the SNI and other
social network measures to describe the size of an individual’s
social circle and frequency of social contact, which have
described that those reporting impoverished social relation-
ships have an elevated risk for a variety of health events. For
example, multiple population studies support an association
between smaller social networks and all-cause and CVD mor-
tality (25–28). Reduced social contacts may, at least in part,
explain the well-established relationship between low socio-
economic status and health (29). Finally, social isolation is
closely associated with depression, an established predictor of
CVD incidence and progression (7,30,31).
To our knowledge, the relationship between social relation-
ships and stroke has been investigated in four previous studies,
with mixed results. Vogt and colleagues (16) reported social
network effects on a 15-year incidence of mortality and spe-
cific disease incidence, including stroke. They described
strong social network associations with mortality, but in-
creases in stroke risk only among young participants (age
range ⫽30 – 44 years), suggesting that the impact of social
networks may be greater in the aftermath of disease onset. No
gender-specific analyses for stroke were presented. In a study
of ⬎15,000 patients with symptoms consistent with acute
myocardial infarction (32), the authors reported on relation-
ships between living alone and short-term (30-day and 1-year)
mortality and stroke events. After covariate adjustment, no
relationships between living alone and mortality or stroke
were present. Tomaka and colleagues (15) reported cross-
sectional associations between social network and support
measures and disease including stroke, observing stroke rela-
tionships with self-reported loneliness and family support.
Most recently, a study of workplace stress also showed that
social support was a predictor of subsequent stroke and myo-
cardial infarction events, relationships that also held among
women (14).
Combined with the current results, the above findings
suggest both promise and ongoing challenges for future re-
search in this area. The study of social relationships remains
limited by a multiplicity of terminology and measurement
approaches that makes it difficult to compare findings across
investigations that already vary substantially in demographic
and clinical characteristics. Social relationships are also highly
dynamic, although few or no health studies assess these char-
acteristics repeatedly over time, and are likely a consequence
of health status changes (15,16) as well as possible cause. Our
exploratory findings of single items from the SNI also sug-
gested that the study of specific relationship subtypes beyond
the standard marital status or living alone categories might
also be fruitful in future research.
There are multiple behavioral and pathophysiological path-
ways by which social networks may affect CVD risk. Protec-
tive effects of social contacts may be due to the tangible
support provided by others, emotional benefits of social rela-
tionships, by promoting physical activity (e.g., going to
church, work, or a friend’s place) or a combination of these
and other pathways. Poor social connectedness is further as-
sociated with increased sympathetic nervous system reactivity
to stress, heightened neurohormonal activation (e.g., elevated
cortisol levels) and compromised immune function, which
may increase susceptibility to infections and inflammation
(20 –22,33,34). There is currently no evidence to suggest that
the mechanisms potentially linking social networks to in-
creased stroke risk differ from those proposed to explain
previously observed relationships with mortality or CAD;
however, we do not believe any research has specifically
addressed this point.
The WISE protocol includes a number of methodological
features that improve the reliability of the social networks-
stroke relationship reported here. The baseline examination
included a thorough measurement of standard risk factors for
CVD and psychosocial factors, use of CVD risk factor med-
ications known to affect prognosis, and a coronary angiogram
as a standardized measure of CAD severity. As a result of
these design components, we were able to evaluate the pre-
dictive value of social relationships after adjusting for a num-
ber of important risk factors. In our analyses, the risk factor
profiles of women with smaller social networks were consis-
tently worse across demographic variables and CVD risk
factors. Adjusting for these risk factors, however, accounted
only partially for the observed SNI-stroke relationship. Our
observations concerning a nonlinear trend in the relationship
between SNI scores and stroke is in contrast to previous social
network reports (25,27), including previous WISE data de-
scribing SNI relationships with mortality (23), which reported
robust linear associations. Although it is possible that the
present findings are capturing relationship elements unique to
the WISE sample or methodology, the more probable expla-
nation is that the distribution of stroke events was unstable due
to the small sample. The fact that SNI scores continued to
predict stroke events despite these inconsistencies reinforces
the potential importance of the relationship, but the need for
confirmatory research from additional studies is clear.
Study Limitations
Despite the study assets, the relationship between social
networks and stroke in the WISE population is observational,
and should not be misconstrued as implying a causal associ-
ation. The measurement protocol does not include assessments
of other speculated mechanisms linking psychosocial factors
and CVD (e.g., neurohormonal activity, autonomic dysfunc-
tion, atrial fibrillation, medication and/or treatment adher-
ence). Stroke is a broad diagnostic category, assuming several
SOCIAL NETWORKS AND STROKE
285Psychosomatic Medicine 70:282–287 (2008)

specific diagnoses with differing etiologies that we were un-
able to differentiate. Stroke events occurred at a low rate in the
WISE sample, and the recording of CVD events was primarily
based on standardized clinical interviews rather than hospital
record documentation. Although we have no evidence to sug-
gest that our interview method was differentially biased by
women with lower versus higher SNI scores, the combination
of this documentation method with the small number of stroke
events encourages caution. The WISE sample had a low rate
of angiographically significant CAD, as substantiated by rates
of obstructive coronary stenoses (⬎50% occlusion) present in
the angiograms of ⬍40% of participants (14), but carried a
heavy disease burden as substantiated by the high rates of
CVD risk factors and use of risk factor medications. Due to
these characteristics, caution must be drawn in extrapolating
the current findings to dissimilar populations including men,
older-age samples, and asymptomatic or healthy women,
among others. We measured social relationships only at base-
line, whereas the size of women’s social networks probably
changed in multiple ways over nearly 6 years of follow-up.
There is no universal definition of small social network val-
ues; we grouped women according to a median and quartile
distribution of SNI scores that is unlikely to replicate perfectly
in other cohorts.
Conclusion
This study demonstrates an association between social net-
works and incident stroke in a cohort of women with sus-
pected CAD. Over a median 5.9 years of follow-up, more
isolated women experienced a stroke rate greater than twice
the rate of those with larger social networks. These findings
add to an already broad literature demonstrating associations
between smaller social networks and an increased risk of
all-cause and CVD mortality (27,28). Although social isola-
tion is a recognized problem in the aftermath of stroke (35),
these findings suggest that social relationships may also be
important to women at the stages of primary and secondary
prevention.
REFERENCES
1. Gorelick PM, Sacco RL, Smith DB, Alberts M, Mustone-Alexander L,
Rader D, Ross JL, Raps E, Ozer MN, Brass LM, Malone ME, Goldberg
S, Booss J, Hanley DF, Toole JF, Greengold NL, Rhew DC. Prevention
of a first stroke: a review of guidelines and a multidisciplinary consensus
statement from the National Stroke Association. JAMA 1999;281:
1112–20.
2. Goldstein LB, Adams R, Alberts MJ, Appel LJ, Brass LM, Bushnell CD,
Culebras A, DeGraba TJ, Gorelick PM, Guyton JR, Hart RG, However G,
Kelly-Hayes M, Nixon JV, Sacco RL. Primary prevention of ischemic
stroke: a guideline from the American Heart Association/American
Stroke Association Stroke Council. Circulation 2006;113:e873– e923.
3. American Heart Association. Heart disease and stroke statistics—2007
update. Heart disease and stroke statistics—2007 update (at-a-glance
version). Available at www.americanheart.org.
4. Mokdad AH, Stroup DF, Giles WH. Public health surveillance for be-
havioral risk factors in a changing environment: recommendations from
the behavioral risk factor surveillance team. MMWR Recomm Rep
2003;52:1–12.
5. Yusuf S, Hawken S, Ounpuu S, Dans T, Avezum A, Lanas F, McQueen
M, Budaj A, Pais P, Varigos J, Lisheng L. INTERHEART study inves-
tigators. Effect of potentially modifiable risk factors associated with
myocardial infarction in 52 countries (the INTERHEART study): case-
control study. Lancet 2004;364:937–52.
6. Lichtman JH, Krumholz HM, Wang Y, Radford MJ, Brass LM. Risk and
predictors of stroke after myocardial infarction among the elderly: results
from the cooperative cardiovascular project. Circulation 2002;105:
1082–7.
7. Rosengren A, Hawken S, Ounpuu S, Sliwa K, Zubaid M, Almahmeed
WA, Blackett KN, Sitthi-amorn C, Sato H, Yusuf S. INTERHEART
investigators. Association of psychosocial risk factors with risk of acute
myocardial infarction in 11119 cases and 13648 controls from 52 coun-
tries (the INTERHEART study): case-control study. Lancet 2004;364:
953– 62.
8. Rozanski A, Blumenthal JA, Kaplan J. Impact of psychological factors on
the pathogenesis of cardiovascular disease and implications for therapy.
Circulation 1999;99:2192–217.
9. Rumsfeld JS, Havranek E, Masoudi FA, Peterson ED, Jones P, Tooley JF,
Krumholz HM, Spertus JA. Depressive symptoms are the strongest
predictor of short-term declines in health status in patients with heart
failure. J Am Coll Cardiol 2003;42:1811–7.
10. Hemingway H, Marmot M. Evidence based cardiology: psychosocial
factors in the aetiology and prognosis of coronary heart disease: system-
atic review of prospective cohort studies. BMJ 1999;318:1460 –7.
11. Orth-Gomer K, Wamala SP, Horsten M, Schenck-Gustafsson K,
Schneiderman N, Mittleman MA. Marital stress worsens prognosis in
women with coronary heart disease: the Stockholm female coronary
risk study. JAMA 2000;284:3008 –14.
12. Everson SA, Kaplan GA, Goldberg DE, Lakka TA, Sivenius J, Salonen
JT. Anger expression and incident stroke: prospective evidence from the
Kuopio ischemic heart disease study. Stroke 1999;30:523– 8.
13. Eng PM, Fitzmaurice G, Kubzansky LD, Rimm EB, Kawachi I. Anger
expression and risk of stroke and coronary heart disease among male
health professionals. Psychosom Med 2003;65:100 –10.
14. Andre-Petersson L, Engstrom G, Hedblad B, Janzon L. Social support at
work and the risk of myocardial infarction and stroke in women and men.
Soc Sci Med 2007;64:830 –41.
15. Tomaka J, Thompson S, Palacios R. The relation of social isolation,
loneliness, and social support to disease outcomes among the elderly.
J Aging 2006;18:359 – 84.
16. Vogt TM, Mullooly JP, Ernst D, Pope CR, Hollis JF. Social networks as
predictors of ischemic heart disease, cancer, stroke and hypertension:
incidence, survival and mortality. J Clin Epidemiol 1992;45:659 –66.
17. Bairey Merz CN, Kelsey SF, Pepine CJ, Reichek N, Reis SE, Rogers WJ,
Sharaf BL, Sopko G. The women’s ischemia syndrome evaluation
(WISE) study: protocol design, methodology, and feasibility report. J Am
Coll Cardiol 1999;33:1453– 61.
18. Sharaf BL, Pepine CJ, Kerensky RA, Reis SE, Reichek N, Rogers WJ,
Sopko G, Kelsey SF, Holubkov R, Olson M, Miele NJ, Williams DO,
Bairey Merz CN. Detailed angiographic analysis of women with sus-
pected ischemic chest pain (pilot phase data from the NHLBI-sponsored
women’s ischemia syndrome evaluation [WISE] study angiographic core
laboratory). Am J Cardiol 2001;87:937– 41.
19. Sharaf BL, Williams DO, Miele NJ, McMahon RP, Stone PH, Bjer-
regaard P, Davies R, Goldberg AD, Parks M, Pepine CJ, Sopko G,
Conti CR. A detailed angiographic analysis of patients with ambula-
tory electrocardiographic ischemia: results from the asymptomatic
cardiac ischemia pilot (ACIP) study angiographic core laboratory.
J Am Coll Cardiol 1997;29:78 –84.
20. Cohen S, Doyle WJ, Skoner DP, Rabin BS, Gwaltney JM. Social ties and
susceptibility to the common cold. JAMA 1997;277:1940 –4.
21. Loucks EB, Sullivan LM, D’Agostino RB Sr, Larson MG, Berkman LF,
Benjamin EJ. Social networks and inflammatory markers in the Framing-
ham heart study. J Biosoc Sci 2006;38:835– 42.
22. Ford ES, Loucks EB, Berkman LF. Social integration and concentrations
of C-reactive protein among US adults. Ann Epidemiol 2006;16:78 –84.
23. Rutledge T, Reis SE, Olson M, Owens J, Kelsey SF, Pepine CJ, Mankad
S, Rogers WJ, Bairey Merz CN, Sopko G, Cornell CE, Sharaf B,
Matthews KA. National Heart, Lung, and Blood Institute. Social net-
works are associated with lower mortality rates among women with
suspected coronary disease: the National Heart, Lung, and Blood Insti-
tute-sponsored women’s ischemia syndrome evaluation study. Psycho-
som Med 2004;66:882– 8.
24. Beck AT. Depression Inventory. Philadelphia: Center for Cognitive Therapy;
1978.
25. Berkman LF, Syme SL. Social networks, host resistance, and mortality:
T. RUTLEDGE et al.
286 Psychosomatic Medicine 70:282–287 (2008)

a nine-year follow-up study of Alameda County residents. Am J Epide-
miol 1979;109:186 –204.
26. House JS, Landis KR, Umberson D. Social relationships and health.
Science 1988;241:540 –5.
27. Schoenbach VJ, Kaplan BH, Fredman BH, Kleinbaum DG. Social ties
and mortality in Evans County. Am J Epidemiol 1986;123:577–91.
28. Kaplan GA, Salonen JT, Cohen RD, Brand RJ, Syme SL, Puska P. Social
connections and mortality from all causes and from cardiovascular disease:
prospective evidence from Eastern Finland. Am J Epidemiol 1988;128:370 –80.
29. Cohen S, Schwartz JE, Epel E, Kirschbaum C, Sidney S, Seeman T.
Socioeconomic status, race, and diurnal cortisol decline in the coronary
artery risk development in young adults (CARDIA) study. Psychosom
Med 2006;68:41–50.
30. Pressman SD, Cohen S, Miller GE, Barkin A, Rabin BS, Treanor JJ.
Loneliness, social network size, and immune response to influenza vac-
cination in college freshmen. Health Psychol 2005;24:297–306.
31. Hawkley LC, Burleson MH, Berntson GG, Cacioppo JT. Loneliness in
everyday life: cardiovascular activity, psychosocial context, and health
behaviors. J Pers Soc Psychol 2003;85:105–20.
32. O’Shea JC, Wilcox RG, Skene AM, Stebbins AL, Granger CB, Arm-
strong PW, Bode C, Ardissino D, Emanuelsson H, Aylward PE, White
HD, Sadowski Z, Topol EJ, Califf RM, Ohman EM. Comparison of
outcomes of patients with myocardial infarction when living alone versus
those not living alone. Am J Cardiol 2002;90:1374 –7.
33. Wulsin LR, Singal BM. Do depressive symptoms increase the risk for the
onset of coronary disease? A systematic quantitative review. Psychosom
Med 2003;65:201–10.
34. Barth J, Schumacher M, Herrman-Lingen C. Depression as a risk factor
for mortality in patients with coronary heart disease. Psychosom Med
2004;66:802–13.
35. Boden-Albala B, Litwak E, Elkind MS, Rundek T, Sacco RL. Social
isolation and outcomes post stroke. Neurology 2005;64:1888 –92.
SOCIAL NETWORKS AND STROKE
287Psychosomatic Medicine 70:282–287 (2008)