Abstract
Background
The association of social isolation or lack of social network ties in older adults is unknown. This knowledge gap is important since the risk of heart failure (HF) and social isolation increase with age. The study examines whether social isolation is associated with incident HF in older women, and examines depressive symptoms as a potential mediator and age and race and ethnicity as effect modifiers.
Methods and Results
This study included 44 174 postmenopausal women of diverse race and ethnicity from the WHI (Women's Health Initiative) study who underwent annual assessment for HF adjudication from baseline enrollment (1993–1998) through 2018. We conducted a mediation analysis to examine depressive symptoms as a potential mediator and further examined effect modification by age and race and ethnicity. Incident HF requiring hospitalization was the main outcome. Social isolation was a composite variable based on marital/partner status, religious ties, and community ties. Depressive symptoms were assessed using CES‐D (Center for Epidemiology Studies‐Depression). Over a median follow‐up of 15.0 years, we analyzed data from 36 457 women, and 2364 (6.5%) incident HF cases occurred; 2510 (6.9%) participants were socially isolated. In multivariable analyses adjusted for sociodemographic, behavioral, clinical, and general health/functioning; socially isolated women had a higher risk of incident HF than nonisolated women (HR, 1.23; 95% CI, 1.08–1.41). Adding depressive symptoms in the model did not change this association (HR, 1.22; 95% CI, 1.07–1.40). Neither race and ethnicity nor age moderated the association between social isolation and incident HF.
Conclusions
Socially isolated older women are at increased risk for developing HF, independent of traditional HF risk factors.
Registration
URL: http://www.clinicaltrials.gov; Unique identifier: NCT00000611.
Keywords: heart failure, older adults, social isolation, women
Subject Categories: Epidemiology, Women, Aging, Race and Ethnicity
Nonstandard Abbreviations and Acronyms
- WHI
Women’s Health Initiative
Clinical Perspective
What Is New?
The association between social isolation, defined as objective lack of or disengagement from social ties, institutional connections, or community participation, and incident heart failure (HF) hospitalization in older adults is unknown.
This study showed that socially isolated women had a higher risk of incident HF hospitalization than nonisolated women even after accounting for traditional risk factors for HF.
The association between social isolation and incident HF hospitalization was not explained by greater depressive symptoms among socially isolated women and did not differ by race and ethnicity or age at baseline.
What Are the Clinical Implications?
Healthcare providers should recognize and assess for social isolation as a psychosocial risk factor for incident HF hospitalization in postmenopausal women with traditional risk factors for HF.
The National Academy of Medicine has recommended routine assessment for social isolation in healthcare settings.
Social isolation—defined as objective lack of or disengagement from social ties, institutional connections, or community participation 1 —is strongly associated with all‐cause and cardiovascular mortality, particularly in older adults. 2 , 3 Few prospective studies have examined the role of social isolation in heart failure (HF) incidence. 4 Using data from the ARIC (Atherosclerosis Risk in Communities) study, Cené and colleagues demonstrated that social isolation is a predictor of incident HF among middle‐aged adults (mean age of 57 years) at study enrollment. 5 However, it is unclear whether social isolation predicts incident HF in older adults and whether the association is stronger with increasing age. Like HF prevalence, the risk of social isolation also increases with age as network disruption occurs because of life course factors such as retirement, bereavement, declining health status, or physical disability. A recent poll of a nationally representative sample of over 2000 adults aged 50 to 80 years found that 27% of respondents reported being socially isolated. 6 Strong evidence shows a nearly 2‐fold higher cardiovascular risk among those who are socially isolated. 7 , 8
There are several pathways by which social isolation might influence HF risk. 9 , 10 Social isolation is considered a stressor that has several effects. It can induce a negative psychological state which can increase neuroendocrine responses; reduce levels of protective hormones, leading to adverse effects on heart rate, blood pressure, and repair of blood vessel walls; downregulate the immune system; and suppress immune function. Because of their lack of social networks or support, socially isolated individuals may suffer more stress than others. They may also be more likely to become depressed and disengage from participating in health‐promoting activities. Strong evidence suggests that lack of social integration is associated with depression. 11 , 12 Few studies have sought to examine potential mediating pathways between social isolation and incident HF. 2 Our study objectives were to: (1) examine whether socially isolated participants in the WHI (Women’s Health Initiative) study are at higher risk for developing HF compared with those who are not socially isolated, (2) assess whether an observed association is modified by age and race and ethnicity, and (3) examine depression as a mediator of any observed association between HF and social isolation.
METHODS
The data, analytic methods, and study materials will be made available to other researchers for purposes of reproducing the results or replicating the procedure. 13
Data Source
The WHI is a US study funded by the National Heart, Lung, and Blood Institute. Postmenopausal women were followed for >20 years, being evaluated for cardiovascular disease, cancer, and osteoporosis. 14 , 15 , 16 The original WHI sample at 40 clinical centers across the United States from 1993 to 1998 included 161 809 postmenopausal women aged 50 to 79 years, making it one of the largest studies of women. Participants were enrolled in either ≥1 randomized clinical trials or an observational study. They were then followed at least annually to track vital status and medical outcomes until study closeout (October 2004–March 2005). 17 After closeout, all participants were invited to enroll in 2 extension studies (2010–2015, 2015–2020) to continue tracking their health status. The study included self‐reported medical information collected through interviews and surveys, anthropometric measurements by WHI personnel, and review of medical records for outcomes determination. To ensure uniform data collection; standardized written protocols, centralized training of staff, and quality assurance visits by the Clinical Coordinating Center were used.
Study Population
Our analytic sample included a subcohort of the original WHI population, selected to participate in a study on the epidemiology of HF in postmenopausal women. This sample included 44 174 postmenopausal women of diverse race and ethnicity, all who underwent annual assessment for HF adjudication from baseline enrollment (1993–1998) through 2018. It also included all participants who were randomized to the WHI hormone therapy trial (n=27 347) as well as an oversampling of minorities comprising all non‐hormone trial Black (n=11 880) and Hispanic (n=4947) women. We excluded women with a diagnosis of HF at baseline (n=430) and those missing data on variables used to define social isolation (n=3210) and other key covariates (n=3768). The final analytic sample included 36 457 participants. (Figure 1) The study was approved by the Human Subjects Review Committees at each WHI participating institution, and all participants provided written informed consent.
Figure 1. Analytic sample.
BMI indicates body mass index; and HF, heart failure; SF36, Short Form 36.
Outcome of Interest
Participants were followed from study enrollment until incident HF requiring hospitalization. Adjudication for HF hospitalization was based upon annual review of medical records and self‐report of hospitalizations. WHI defined incident HF hospitalization as “definite or probable,” according to symptoms, physical exam, clinical data, and medical therapy provided during hospitalizations as previously reported. 18
Exposure of Interest
Using an adapted version of the Berkman–Syme social network index, social isolation was broadly defined as a relative lack of social network ties 19 , 20 and was assessed at 4 time periods. This index weights 3 different types of participant‐reported social network ties: intimate contact (spousal ties), religious ties, and community/group membership. To assess marital status, we asked, “Are you currently married or in an intimate relationship with at least one person?” (yes or no). Specific wording for the questions on religious ties was as follows: “How often have you gone to a religious service or to a church in the past month?” and “How often have you gone to the meetings of clubs, lodges, or parent groups in the past month?” Response options for religious and community participation questions were on a 1 to 6 scale with 1 representing not at all and 6 representing every day. Women who indicated that they were not married/partnered, had no religious ties, and no community ties were considered socially isolated. Our analysis compares women who were socially isolated versus not socially isolated. Unlike the original Berkman–Syme social network index, we did not include number and frequency of contacts with children, close friends, and close relatives in our definition of social isolation because this variable was not assessed in WHI at baseline.
We constructed a composite social isolation variable that considered marital status, religious ties, and community/group membership. In sensitivity analyses, we considered “living alone” as a proxy for social isolation.
Covariates
Guided by the Berkman–Glass conceptual model of social networks on health outcomes, 21 we created a causal diagram to identify potential confounders of the association between social isolation and incident HF hospitalization. 22 We grouped covariates into 3 groups of factors: demographic, health status, and lifestyle/behavioral and clinical. Demographic factors included: age (continuous); race and ethnicity (White, Black, Hispanic, Asian/Pacific Islander, American Indian/Alaskan Native, Other/Multiple), education, and WHI enrollment (clinical trial versus observational study). Health status factors included: self‐rated overall health (first item of Short‐Form 12; 5‐point Likert scale‐ Excellent to Poor) 23 , 24 and physical functioning. Lifestyle/behavioral and clinical factors included: sedentary behavior (<500 metabolic equivalent of task minute/week) 25 ; body mass index (continuous and categorical body mass index ≥30 versus <30); smoking—defined as prior hospitalization for myocardial infarction, prior coronary artery bypass graft surgery, prior percutaneous coronary intervention (coronary artery bypass graft, percutaneous transluminal coronary angioplasty), history of carotid artery disease, stroke/transient ischemic attack or peripheral vascular disease; history of treated diabetes mellitus (yes/no); history of treated high cholesterol (yes/no); and hypertension status (never hypertensive, currently treated, currently untreated). We also included incident myocardial infarction as a time‐dependent covariate.
Mediator
We hypothesized that depressive symptoms may mediate the association between social isolation and incident HF. Depressive symptoms were measured at 4 different time periods using the 8‐item Burnham short version of the CES‐D (Center for Epidemiologic Studies‐Depression). 26 Responses for each item were weighted according to the Burnham algorithm with a final range from 0 to 1—higher scores indicate greater likelihood of depression. Scores >0.06 are indicative of significant depressive symptoms, and scores ≤0.06 are indicative of no/minimal depressive symptoms.
Effect Moderator Variables
We examined age at baseline and race and ethnicity as potential modifiers of the association between social isolation and incident HF hospitalization.
Statistical Analysis
Women’s characteristics overall and by baseline social isolation status were summarized using means (SD) for continuous variables and frequencies (percentages) for categorical variables. For continuous variables and categorical variables, we used Student t‐test and Chi‐square tests to examine differences between social isolation groups. In 3 models, after sequential adjustment for potential confounding variables, we used Cox proportional hazards regression with time‐dependent covariates (social isolation composite and its individual indicators, depressive symptoms, and incident myocardial infarction) to estimate hazard ratios (HR) and 95% CI for the association between social isolation and time to incident HF hospitalization. In the minimally adjusted model (model 0), we adjusted for age, race, education level, study arm (clinical trial versus observational study), and hormone therapy arm. In the first fully adjusted model (Model 1), we included covariates in Model 0, as well as treated diabetes, hypertension status, treated high cholesterol, body mass index, sedentary behavior, prevalent cardiovascular disease (CVD) and incident myocardial infarction, smoking, general health status, and physical functioning. In the second fully adjusted model (Model 2), we added depressive symptoms to the covariates in Model 1. When fitting models, for race and ethnicity, we combined American Indian/Alaskan Native and other/Mixed as 1 category.
We used an ad hoc approach to examine the role of depressive symptoms as a potential mediator. First, we used depressive symptoms as predictor in Models 0 and 1. We then fit a Generalized Estimating Equations model to analyze the associations between social isolation and depressive symptoms, considering the within‐subject correlations. Model 3 was used to evaluate the mediating impact of depressive symptoms on social isolation.
We further analyzed each social isolation indicator separately. Based on Model 1, we examined interactions by race and ethnicity and age. Evidence of interaction was determined based on a statistically significant P value for the interaction term at the P<0.05 level.
We conducted sensitivity analysis to examine how a different composite definition of social isolation (adding “live alone”) affected our results. We also performed survival analyses with CVD death as competing risk. All statistical procedures were performed with the use of SAS version 9.4 (SAS Institute Inc, Cary, NC). All tests were 2‐sided with significance level of P<0.05.
RESULTS
From 1993 to 2018 (median follow‐up 15.0 years), we had 36 457 participants in the analytic sample, and 2364 (6.5%) incident HF cases occurred. Mean age of sample was 62.6 years; 32.8% of sample was Black women, and 14% were Latina. In our sample, 2510 (6.9%) participants were socially isolated. Sociodemographic and clinical characteristics at baseline (Table 1) showed statistically significant differences between socially isolated and non‐isolated women. For those in the socially isolated group, the percentage of women with a CES‐D score of >0.06 (indicative of a significant depressive symptoms) was higher (19% versus 12%; P<0.001).
Table 1.
Participant Characteristics at Baseline
| Overall, n (%) | Not socially isolated, n (%) | Socially isolated, n (%) | P value | |
|---|---|---|---|---|
| Total | 36 457 | 33 974 | 2510 | |
| Age (y), mean (SD) | 62.6 (7.2) | 62.6 (7.2) | 62.6 (7.3) | 0.73 |
| Race and ethnicity | ||||
| American Indian/Alaskan Native | 107 (0.3) | 94 (0.3) | 13 (0.5) | <0.001 |
| Asian/Pacific Islander | 481 (1.3) | 434 (1.3) | 47 (1.9) | |
| Black | 11 960 (32.8) | 11 324 (33.4) | 636 (25.3) | |
| Hispanic/Latino | 5058 (13.9) | 4760 (14.0) | 298 (11.9) | |
| White | 18 566 (50.9) | 17 064 (50.3) | 1502 (59.8) | |
| Other/multiple | 285 (0.8) | 271 (0.8) | 14 (0.6) | |
| Education | ||||
| Less than high school | 3456 (9.5) | 3199 (9.4) | 257 (10.2) | 0.005 |
| High school/vocational training | 11 111 (30.5) | 10 365 (30.5) | 746 (29.7) | |
| Some college/associate degree | 10 157 (27.9) | 9397 (27.7) | 760 (30.3) | |
| College/graduate | 11 733 (32.2) | 10 986 (32.4) | 747 (29.8) | |
| Observational study | ||||
| No | 27 231 (74.7) | 25 255 (74.4) | 1976 (78.7) | <0.001 |
| Yes | 9226 (25.3) | 8692 (25.6) | 534 (21.3) | |
| Hormone therapy arm | <0.001 | |||
| Not in hormone therapy trial | 13 704 (37.6) | 12 978 (38.2) | 726 (28.9) | |
| CEE alone control | 4386 (12.0) | 4033 (11.9) | 353 (14.1) | |
| CEE alone intervention | 4404 (12.1) | 4046 (11.9) | 358 (14.3) | |
| CEE+MPA control | 7053 (19.4) | 6483 (19.1) | 570 (22.7) | |
| CEE+MPA intervention | 6910 (18.9) | 6407 (18.9) | 503 (20.0) | |
| Depressive symptom | ||||
| No | 30 933 (87.3) | 28 976 (87.8) | 1957 (80.7) | <0.001 |
| Yes | 4490 (12.7) | 4023 (12.2) | 467 (19.3) | |
| Smoking status | ||||
| Never smoked | 18 676 (51.2) | 17 790 (52.4) | 886 (35.3) | <0.001 |
| Past smoker | 14 095 (38.7) | 13 020 (38.4) | 1075 (42.8) | |
| Current smoker | 3686 (10.1) | 3137 (9.2) | 549 (21.9) | |
| BMI (kg/m2), mean (SD) | 29.6 (6.3) | 29.6 (6.2) | 29.7 (6.6) | 0.30 |
| Obesity | ||||
| No | 21 526 (59.0) | 20 062 (59.1) | 1464 (58.3) | 0.45 |
| Yes | 14 931 (41.0) | 13 885 (40.9) | 1046 (41.7) | |
| High cholesterol requiring pills | ||||
| No | 31 315 (85.9) | 29 145 (85.9) | 2170 (86.5) | 0.40 |
| Yes | 5142 (14.1) | 4802 (14.1) | 340 (13.5) | |
| Hypertension | ||||
| Never hypertensive | 22 298 (61.2) | 20 704 (61.0) | 1594 (63.5) | <0.001 |
| Current/treated | 3294 (9.0) | 3023 (8.9) | 271 (10.8) | |
| Current/untreated | 10 865 (29.8) | 10 220 (30.1) | 645 (25.7) | |
| Treated diabetes | ||||
| No | 33 783 (92.7) | 31 467 (92.7) | 2316 (92.3) | 0.43 |
| Yes | 2674 (7.3) | 2480 (7.3) | 194 (7.7) | |
| Sedentary behavior | ||||
| No | 15 129 (41.5) | 14 277 (42.1) | 852 (33.9) | <0.001 |
| Yes | 21 328 (58.5) | 19 670 (57.9) | 1658 (66.1) | |
| Prevalent CVD | ||||
| No | 33 788 (92.7) | 31 480 (92.7) | 2308 (92.0) | 0.15 |
| Yes | 2669 (7.3) | 2467 (7.3) | 202 (8.0) | |
| Incident MI | ||||
| No | 34 928 (95.8) | 32 549 (95.9) | 2379 (94.8) | 0.008 |
| Yes | 1529 (4.2) | 1398 (4.1) | 131 (5.2) | |
| General health | ||||
| Excellent | 4865 (13.3) | 4545 (13.4) | 320 (12.7) | <0.001 |
| Very good | 13 582 (37.3) | 12 700 (37.4) | 882 (35.1) | |
| Good | 13 549 (37.2) | 12 662 (37.3) | 887 (35.3) | |
| Fair | 4095 (11.2) | 3726 (11.0) | 369 (14.7) | |
| Poor | 366 (1.0) | 314 (0.9) | 52 (2.1) | |
| SF36—physical functioning, mean (SD) | 79.0 (21.5) | 79.2 (21.3) | 76.2 (23.5) | <0.001 |
BMI indicates body mass index; CEE, combined equine estrogens; CVD, cardiovascular disease; MI, myocardial infarction; and MPA, medroxyprogesterone acetate; SF36, Short Form 36.
Table 2 lists the unadjusted and adjusted HR of HF incidence based on social isolation status. The minimally adjusted risk of HF was 56% higher in women who were socially isolated (Model 0: HR, 1.56%; 95% CI, 1.37–1.78). Socially isolated women had significantly greater risk of developing HF in the fully adjusted model (Model 1: HR, 1.23; 95% CI, 1.08–1.41). Figure 2 shows the Kaplan–Meier plot of time to incident HF hospitalization by baseline social isolation status. In the ad hoc mediation analysis, Generalized Estimating Equation model results indicated that social isolation increased risk of depressive symptoms (regression coefficient, 0.27; 95% CI, 0.20–0.34; P<0.001). Depressive symptoms were independently associated with incident HF (Model 0: HR, 1.21; 95% CI, 1.06–1.38; P=0.005) in the minimally adjusted model; however, the association was diminished in both fully adjusted models (Model 1: HR, 0.93; 95% CI, 0.81–1.06; P=0.26; Model 2: HR, 0.92; 95% CI, 0.80–1.05; P=0.23). After additionally adjusting for depressive symptoms, the association between social isolation and incident HF remained significant (Model 2: HR, 1.22; 95% CI, 1.07–1.40, P=0.003), with a similar HR, indicating that depressive symptoms might not mediate the association between social isolation and incident HF risk. Effect modification by race and ethnicity or baseline age of the association between social isolation and incident HF was not statistically significant based on the tests of interactions (Table 2). Figure 3 summarizes the results of our findings according to our conceptual model.
Table 2.
Association Between Incident HF Hospitalization and Social Isolation (Total Number of Observations: 36 457; Total Number of Events: 2364)
| Predictor | Model 0* | Model 1 † | Model 2 ‡ | Interaction P value § | Interaction P value ∥ | |||
|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value | |||
| Social isolation ¶ | 1.56 (1.37–1.78) | <0.001 | 1.23 (1.08–1.41) | 0.002 | 1.22 (1.07–1.40) | 0.003 | 0.29 | 0.94 |
| Not married or intimate | 1.25 (1.15–1.35) | <0.001 | 1.11 (1.02–1.20) | 0.020 | 1.10 (1.01–1.20) | 0.024 | 0.21 | 0.16 |
| No religion tie # | 1.19 (1.09–1.30) | <0.001 | 1.06 (0.97–1.16) | 0.22 | 1.06 (0.97–1.16) | 0.23 | 0.50 | 0.92 |
| No community tie # | 1.15 (1.06–1.25) | <0.001 | 1.01 (0.93–1.10) | 0.75 | 1.01 (0.93–1.10) | 0.77 | 0.48 | 0.47 |
| Live alone | 1.00 (0.92–1.09) | 0.98 | 0.97 (0.89–1.06) | 0.11 | 0.97 (0.89–1.06) | 0.52 | 0.14 | 0.27 |
Model 0: adjusted for age, race, education, study arm (clinical trial vs observational study), hormone therapy arm.
Model 1: Adjusted for age, race, education, study arm (clinical trial vs observational study), hormone therapy arm, treated diabetes, hypertension status, treated high cholesterol, obesity, sedentary behavior, prevalent cardiovascular disease and incident myocardial infarction, smoking, self‐rated health status, and physical functioning.
Model 2: model I plus most recent depressive symptoms as time‐varying covariate.
Interaction between race and ethnicity and predictor for Model 1.
Interaction between baseline age and predictor for Model 1.
The sum of not married, no religion tie, and no community tie was calculated. If the sum is 3, then social isolation=“Yes”.
Defined as not at all in the past month.
Figure 2. Kaplan–Meier product limit estimate of the cumulative incidence of heart failure hospitalization over time by social isolation status (socially isolated vs nonisolated) at baseline.
Figure 3. Conceptual model of direct, mediating, and moderating effects of social isolation on incident HF hospitalization in 36 457 older women (mean age, 62 years) from the WHI (Women's Health Initiative) study over mean follow‐up of 15 years.
Table 3 provides the HRs for each covariate in multivariable Model 1. The following covariates were associated with a significantly increased risk of incident HF hospitalization: age, treated diabetes, obesity, prevalent CVD, incident myocardial infarction, and poor general self‐reported health. A significantly decreased risk of incident HF was associated with non‐White race and ethnicity, hypertensive status (never hypertensive or controlled hypertension), and never or past smoking. Adjusting for depressive symptoms did not alter the association between these covariates and risk of incident HF hospitalization (data not shown).
Table 3.
HRs of Incidence HF Hospitalization of Covariates in Model 1
| Covariates | HR (95% CI) | P value |
|---|---|---|
| Baseline age per year | 1.09 (1.09–1.10) | <0.001 |
| Race and ethnicity (vs White) | <0.001 | |
| Asian/Pacific Islander | 0.71 (0.46–1.11) | |
| Black | 0.78 (0.66–0.91) | |
| Hispanic/Latino | 0.61 (0.49–0.76) | |
| Other | 0.97 (0.67–1.40) | |
| Education (vs college/graduate) | 0.32 | |
| Less than high school | 1.08 (0.92–1.27) | |
| High school/vocational training | 1.19 (0.99–1.22) | |
| Some college/associate degree | 1.08 (0.97–1.20) | |
| Treated diabetes | 2.35 (2.08–2.64) | <0.001 |
| Hypertension | <0.001 | |
| Never hypertensive | 0.58 (0.53–0.64) | |
| Controlled hypertension | 0.70 (0.60–0.81) | |
| Treated high cholesterol | 0.96 (0.86–1.07) | 0.43 |
| Obesity (BMI ≥30) | 1.34 (1.23–1.47) | <0.001 |
| Sedentary behavior | 1.09 (1.00–1.19) | 0.06 |
| Smoking status (vs current smoker) | <0.001 | |
| Never smoked | 0.45 (0.39–0.51) | |
| Past smoker | 0.55 (0.48–0.63) | |
| Prevalent CVD | 1.82 (1.62–2.05) | <0.001 |
| Incident MI | 2.90 (2.53–3.33) | <0.001 |
| General health (vs excellent) | 0.002 | |
| Very good | 1.11 (0.95–1.29) | |
| Good | 1.15 (0.98–1.34) | |
| Fair | 1.44 (1.19–1.76) | |
| Poor | 1.39 (0.94–2.05) | |
| Physical functioning per 1 unit increase | 0.991 (0.989–0.993) | <0.001 |
| Hormone therapy arm (vs not in hormone therapy trial) | 0.002 | |
| CEE alone control | 1.41 (1.14–1.73) | |
| CEE alone intervention | 1.36 (1.10–1.68) | |
| CEE+MPA control | 1.18 (0.95–1.46) | |
| CEE+MPA intervention | 1.24 (1.00–1.53) |
BMI indicates body mass index; CEE, combined equine estrogens; CVD, cardiovascular disease; HF, heart failure; HR, hazard ratio; MI, myocardial infarction; and MPA, medroxyprogesterone acetate.
Survival analyses using CVD death as competing risk against HF did not change the results materially (data not shown).
Notably, “living alone”, which is often used clinically as a proxy for social isolation, was not associated with time to incident HF hospitalization (Model 2: HR, 0.97; 95% CI, 0.89–1.06, P=0.52). Living alone was highly correlated with marital/partnered status (data not shown); therefore, we did not include it in the social isolation composite.
DISCUSSION
Our study yielded 3 main findings. First, social isolation (defined by a lack of marital/partnered, religious, and community ties) was significantly associated with incident HF hospitalization, the risk being, on average, 23% higher in socially isolated versus non‐isolated older women. Second, and contrary to our hypothesis, depressive symptoms did not mediate the association between social isolation and incident HF hospitalization. Third, the association does not differ by race and ethnicity or age.
Few population‐based studies have examined the association between social isolation (or social network characteristics more broadly) and incident HF. 4 , 5 Other studies have also documented strong associations of social isolation with hospital readmission in patients with HF. 27 Our findings confirm previous work conducted in the ARIC (Atherosclerosis Risk in Communities) cohort study, which found that the risk of developing incident HF (defined as incident HF hospitalization or death) was 21% higher in middle‐aged men and women (mean [SD] age of 56.9 [SD, 5.7] years) with moderate/high social isolation risk compared with those with low social isolation. 5 We extend this literature by documenting this independent association in older women (mean [SD] age of 62.6 [SD, 7.2] years) using a different measure of social isolation. On the contrary, Rod and colleagues did not find an association between social network characteristics and incident HF hospitalization. 4 This may be because of differences in study population (Danish sample) and social network measurement (examined living alone and regular contact with family and friends as their social network indicators).
A 2016 systematic review and meta‐analysis examining the prospective association between social isolation or loneliness found that poor social relationships were associated with a 29% increased risk of incident coronary heart disease 28 —a major predisposing factor for HF. Since our social isolation definition did not include an assessment of family and friendship ties, it is possible that the strength of our association is overestimated. However, our previous work in the ARIC study used a measure of social isolation which did include friends and family ties and demonstrated a similar effect size to the current study. 5 Prior literature has noted significant changes in network size and composition with age, with some studies showing that networks shrink, while others show that the total network size remains stable as the number of close relatives increases and the number of friends decreases. 29 Although HF disproportionately affects underrepresented racial and ethnic women, 30 , 31 and some data demonstrate that social network size and social isolation differ by race and ethnicity, 32 we did not find evidence that the association between social isolation and HF risk differed by race and ethnicity.
Contrary to our hypothesis, we did not find evidence that depressive symptoms mediated the effect of social isolation on incident HF. Among the few studies that have examined the association between depressive symptoms and incident HF, including some focused on older populations, results have been mixed. 33 , 34 , 35 , 36 It is possible that the association is only evident among subgroups, such as individuals with poor health at baseline. 34
The mechanisms through which social isolation influences cardiovascular risk (including HF) are unknown but are hypothesized to include neuroendocrine pathways, health behaviors, and physiologic pathways such as inflammation. 2 , 9 , 37 Social isolation may predispose individuals to chronic psychosocial distress and toxic biological stress response. This stress response causes sympathetic activation which can facilitate pathogenic processes involved in coronary heart disease and ultimately HF. Identification of mediators is crucial so that clinicians and researchers can better assess risk of developing HF and subsequently intervene before HF develops or worsens. Cené and colleagues previously demonstrated that vital exhaustion, an aspect of functional status, which shares common symptomatology with depression (fatigue, irritability, and feelings of demoralization), mediated the association between social isolation and incident HF in middle‐aged adults. 5
Our study had several notable limitations. First, our definition of social isolation did not include frequency of contact with family and friends. This type of social interaction is sometimes used in composite definitions of social isolation and is a known correlate of social isolation. 38 Second, we used HF hospital admissions as a proxy measure of heart failure incidence. However, HF hospitalization was an adjudicated outcome and the specificity for accurately identifying individuals with clinical heart failure is likely high. Third, as with all observational studies, there is a potential for residual confounding because of unmeasured variables, such as loneliness (perceived isolation), 7 other aspects of psychosocial distress (eg, anxiety, stress), or clinical conditions that could impede social connections (eg, hearing or vision loss, limited mobility). Finally, WHI solely includes older women; therefore, results may not be generalizable to men or younger women.
Despite these limitations, our study has several strengths. Notably, we addressed an important and timely research question for older adults and women’s health. 4 A recent publication highlights the need for increased attention to the pathophysiology of heart disease in women and emphasizes the importance of conducting sex‐specific examinations. 39 Second, we leveraged the strengths of the WHI study, including population‐based sampling, large sample size, and extended follow‐up time. The large sample size allowed for greater accuracy of results and the ability to simultaneously adjust for confounders. Finally, our analysis remained sensitive to different definitions of social isolation, thus bolstering the validity of our findings.
Identifying and addressing health implications of social isolation is important from a clinical and public health standpoint. There is increasing interest in social determinants of health (including social isolation) as predictors of health and well‐being. The National Academy of Medicine has recommended routine assessment for social isolation in healthcare settings 40 and they endorse a 5‐item measure of social isolation (based on the same measure we used) that queries the frequency of talking to or getting together with family or friends in a typical week, attending church or religious services, attending meetings of clubs or organizations, and marital status. Healthcare providers should consider screening for social isolation in older adults at risk for HF in addition to traditional HF risk factors and using extant evidence to guide intervention strategies to mitigate social isolation, including referrals to community resources. 41 , 42 Future research should seek to identify mediating pathways through which social isolation influences the development of HF and other clinical conditions.
Sources of Funding
The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, US Department of Health and Human Services through contracts HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C.
Dr Candyce Kroenke has funding from the following sources: NCI Grant R01CA253028, R01CA230440 and K07CA187403 (PI: Kroenke); ACS Research Investigator Award RSG‐16‐167‐01‐CPPB (PI: Kroenke).
Disclosures
None.
Acknowledgments
We would like to acknowledge and thank the WHI participants who generously gave and continue to give of their time to this study. We would like to acknowledge key members of the larger WHI team that made this study possible.
Program Office: (National Heart, Lung, and Blood Institute, Bethesda, Maryland) Jacques Rossouw, Shari Ludlam, Joan McGowan, Leslie Ford, and Nancy Geller.
Clinical Coordinating Center: (Fred Hutchinson Cancer Research Center, Seattle, WA) Garnet Anderson, Ross Prentice, Andrea LaCroix, and Charles Kooperberg.
Investigators and Academic Centers: (Brigham and Women's Hospital, Harvard Medical School, Boston, MA) JoAnn E. Manson; (MedStar Health Research Institute/Howard University, Washington, DC) Barbara V. Howard; (Stanford Prevention Research Center, Stanford, CA) Marcia L. Stefanick; (The Ohio State University, Columbus, OH) Rebecca Jackson; (University of Arizona, Tucson/Phoenix, AZ) Cynthia A. Thomson; (University at Buffalo, Buffalo, NY) Jean Wactawski‐Wende; (University of Florida, Gainesville/Jacksonville, FL) Marian Limacher; (University of Iowa, Iowa City/Davenport, IA) Jennifer Robinson; (University of Pittsburgh, Pittsburgh, PA) Lewis Kuller; (Wake Forest University School of Medicine, Winston‐Salem, NC) Sally Shumaker; (University of Nevada, Reno, NV) Robert Brunner.
For Sources of Funding and Disclosures, see page 10.
REFERENCES
- 1. Seeman TE. Social ties and health: the benefits of social integration. Ann Epidemiol. 1996;6:442–451. doi: 10.1016/S1047-2797(96)00095-6 [DOI] [PubMed] [Google Scholar]
- 2. National Academies of Sciences, Engineering, and Medicine . Social Isolation and Loneliness in Older Adults: Opportunities for the Health Care System. Washington, DC: The National Academies Press; 2020. [PubMed] [Google Scholar]
- 3. Pantell M, Rehkopf D, Jutte D, Syme SL, Balmes J, Adler N. Social isolation: a predictor of mortality comparable to traditional clinical risk factors. Am J Public Health. 2013;103:2056–2062. doi: 10.2105/AJPH.2013.301261 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Rod NH, Andersen I, Prescott E. Psychosocial risk factors and heart failure hospitalization: a prospective cohort study. Am J Epidemiol. 2011;174:672–680. doi: 10.1093/aje/kwr144 [DOI] [PubMed] [Google Scholar]
- 5. Cene CW, Loehr L, Lin FC, Hammond WP, Foraker RE, Rose K, Mosley T, Corbie‐Smith G. Social isolation, vital exhaustion, and incident heart failure: findings from the Atherosclerosis Risk in Communities Study. Eur J Heart Fail. 2012;14:748–753. doi: 10.1093/eurjhf/hfs064 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Solway E, Piette J, Kirch M, Singer D, Kullgren J, Malani P. Loneliness and health among older adults: results form the University of Michigan National Poll on Healthy Aging. Innov Aging. 2019;3(Suppl 1):S600–S601. doi: 10.1093/geroni/igz038.2233 [DOI] [Google Scholar]
- 7. Leigh‐Hunt N, Bagguley D, Bash K, Turner V, Turnbull S, Valtorta N, Caan W. An overview of systematic reviews on the public health consequences of social isolation and loneliness. Public Health. 2017;152:157–171. doi: 10.1016/j.puhe.2017.07.035 [DOI] [PubMed] [Google Scholar]
- 8. Steptoe A, Kivimäki M. Stress and cardiovascular disease: an update on current knowledge. Annu Rev Public Health. 2013;34:337–354. doi: 10.1146/annurev-publhealth-031912-114452 [DOI] [PubMed] [Google Scholar]
- 9. Xia N, Li H. Loneliness, social isolation, and cardiovascular health. Antioxid Redox Signal. 2018;28:837–851. doi: 10.1089/ars.2017.7312 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Grant N, Hamer M, Steptoe A. Social isolation and stress‐related cardiovascular, lipid, and cortisol responses. Ann Behav Med. 2009;37:29–37. doi: 10.1007/s12160-009-9081-z [DOI] [PubMed] [Google Scholar]
- 11. Santini ZI, Koyanagi A, Tyrovolas S, Mason C, Haro JM. The association between social relationships and depression: a systematic review. J Affect Disord. 2015;175:53–65. doi: 10.1016/j.jad.2014.12.049 [DOI] [PubMed] [Google Scholar]
- 12. Schwarzbach M, Luppa M, Forstmeier S, König HH, Riedel‐Heller SG. Social relations and depression in late life—a systematic review. Int J Geriatr Psychiatry. 2014;29:1–21. doi: 10.1002/gps.3971 [DOI] [PubMed] [Google Scholar]
- 13. Women's Health Initiative . Researchers. Available at: https://www.whi.org/page/working‐with‐whi‐data. Accessed September 1, 2021.
- 14. Ritenbaugh C, Patterson RE, Chlebowski RT, Caan B, Fels‐Tinker L, Howard B, Ockene J. The Women's Health Initiative Dietary Modification trial: overview and baseline characteristics of participants. Ann Epidemiol. 2003;13:S87–S97. doi: 10.1016/S1047-2797(03)00044-9 [DOI] [PubMed] [Google Scholar]
- 15. Stefanick ML, Cochrane BB, Hsia J, Barad DH, Liu JH, Johnson SR. The Women's Health Initiative postmenopausal hormone trials: overview and baseline characteristics of participants. Ann Epidemiol. 2003;13:S78–S86. doi: 10.1016/S1047-2797(03)00045-0 [DOI] [PubMed] [Google Scholar]
- 16. Langer RD, White E, Lewis CE, Kotchen JM, Hendrix SL, Trevisan M. The Women's Health Initiative observational study: baseline characteristics of participants and reliability of baseline measures. Ann Epidemiol. 2003;13:S107–S121. doi: 10.1016/S1047-2797(03)00047-4 [DOI] [PubMed] [Google Scholar]
- 17. Anderson GL, Manson J, Wallace R, Lund B, Hall D, Davis S, Shumaker S, Wang C‐Y, Stein E, Prentice RL. Implementation of the Women's Health Initiative study design. Ann Epidemiol. 2003;13:S5–S17. doi: 10.1016/S1047-2797(03)00043-7 [DOI] [PubMed] [Google Scholar]
- 18. Heckbert SR, Kooperberg C, Safford MM, Psaty BM, Hsia J, McTiernan A, Gaziano JM, Frishman WH, Curb JD. Comparison of self‐report, hospital discharge codes, and adjudication of cardiovascular events in the Women’s Health Initiative. Am J Epidemiol. 2004;160:1152–1158. doi: 10.1093/aje/kwh314 [DOI] [PubMed] [Google Scholar]
- 19. Berkman LF, Syme SL. Social networks, host resistance, and mortality: a nine‐year follow‐up study of Alameda County residents. Am J Epidemiol. 1979;109:186–204. doi: 10.1093/oxfordjournals.aje.a112674 [DOI] [PubMed] [Google Scholar]
- 20. Seeman TE, Kaplan GA, Knudsen L, Cohen R, Guralnik J. Social network ties and mortality among the elderly in the Alameda County Study. Am J Epidemiol. 1987;126:714–723. [DOI] [PubMed] [Google Scholar]
- 21. Berkman LF, Glass T. Conceptual model of how social networks impact health. In: Berkman LF, Kawachi I, eds. Social Epidemiology. New York: Oxford University Press; 2000:144. [Google Scholar]
- 22. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology. 1999;10:37–48. doi: 10.1097/00001648-199901000-00008 [DOI] [PubMed] [Google Scholar]
- 23. Chamberlain AM, Manemann SM, Dunlay SM, Spertus JA, Moser DK, Berardi C, Kane RL, Weston SA, Redfield MM, Roger VL. Self‐rated health predicts healthcare utilization in heart failure. J Am Heart Assoc. 2014;3:e000931. doi: 10.1161/JAHA.114.000931 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Ware J Jr, Kosinski M, Keller SD. A 12‐Item Short‐Form Health Survey: construction of scales and preliminary tests of reliability and validity. Med Care. 1996;34:220–233. doi: 10.1097/00005650-199603000-00003 [DOI] [PubMed] [Google Scholar]
- 25. Pandey A, LaMonte M, Klein L, Ayers C, Psaty BM, Eaton CB, Allen NB, de Lemos JA, Carnethon M, Greenland P, et al. Relationship between physical activity, body mass index, and risk of heart failure. J Am Coll Cardiol. 2017;69:1129–1142. doi: 10.1016/j.jacc.2016.11.081 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Karim J, Weisz R, Bibi Z, ur Rehman S. Validation of the eight‐item Center for Epidemiologic Studies Depression Scale (CES‐D) among older adults. Curr Psychol. 2015;34:681–692. doi: 10.1007/s12144-014-9281-y [DOI] [Google Scholar]
- 27. Gorji MH, Fatahian A, Farsavian A. The impact of perceived and objective social isolation on hospital readmission in patients with heart failure: a systematic review and meta‐analysis of observational studies. Gen Hosp Psychiatry. 2019;60:27–36. doi: 10.1016/j.genhosppsych.2019.07.002 [DOI] [PubMed] [Google Scholar]
- 28. Valtorta NK, Kanaan M, Gilbody S, Ronzi S, Hanratty B. Loneliness and social isolation as risk factors for coronary heart disease and stroke: systematic review and meta‐analysis of longitudinal observational studies. Heart. 2016;102:1009–1016. doi: 10.1136/heartjnl-2015-308790 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Van Tilburg T. Losing and gaining in old age: changes in personal network size and social support in a four‐year longitudinal study. J Gerontol B Psychol Sci Soc Sci. 1998;53:S313–S323. doi: 10.1093/geronb/53B.6.S313 [DOI] [PubMed] [Google Scholar]
- 30. Eaton CB, Abdulbaki AM, Margolis KL, Manson JE, Limacher M, Klein L, Allison MA, Robinson JG, Curb JD, Martin LA, et al. Racial and ethnic differences in incident hospitalized heart failure in postmenopausal women: the Women's Health Initiative. Circulation. 2012;126:688–696. doi: 10.1161/CIRCULATIONAHA.111.066688 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Go AS, Mozaffarian D, Roger VL, Benjamin EJ, Berry JD, Blaha MJ, Dai S, Ford ES, Fox CS, Franco S, et al. Heart disease and stroke statistics‐2014 update. Circulation. 2014;129:e28–e292. doi: 10.1161/01.cir.0000441139.02102.80 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Cornwell B, Laumann EO, Schumm LP. The social connectedness of older adults: a national profile. Am Sociol Rev. 2008;73:185–203. doi: 10.1177/000312240807300201 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Rutledge T, Reis VA, Linke SE, Greenberg BH, Mills PJ. Depression in heart failure: a meta‐analytic review of prevalence, intervention effects, and associations with clinical outcomes. J Am Coll Cardiol. 2006;48:1527–1537. doi: 10.1016/j.jacc.2006.06.055 [DOI] [PubMed] [Google Scholar]
- 34. Ogilvie RP, Everson‐Rose SA, Longstreth W Jr, Rodriguez CJ, Diez‐Roux AV, Lutsey PL. Psychosocial factors and risk of incident heart failure: the Multi‐Ethnic Study of Atherosclerosis. Circ Heart Fail. 2016;9:e002243. doi: 10.1161/CIRCHEARTFAILURE.115.002243 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Abramson J, Berger A, Krumholz HM, Vaccarino V. Depression and risk of heart failure among older persons with isolated systolic hypertension. Arch Intern Med. 2001;161:1725–1730. doi: 10.1001/archinte.161.14.1725 [DOI] [PubMed] [Google Scholar]
- 36. May HT, Horne BD, Carlquist JF, Sheng X, Joy E, Catinella AP. Depression after coronary artery disease is associated with heart failure. J Am Coll Cardiol. 2009;53:1440–1447. doi: 10.1016/j.jacc.2009.01.036 [DOI] [PubMed] [Google Scholar]
- 37. Penwell LM, Larkin KT. Social support and risk for cardiovascular disease and cancer: a qualitative review examining the role of inflammatory processes. Health Psychol Rev. 2010;4:42–55. doi: 10.1080/17437190903427546 [DOI] [Google Scholar]
- 38. Umberson D, Karas MJ. Social relationships and health: a flashpoint for health policy. J Health Soc Behav. 2010;51:S54–S66. doi: 10.1177/0022146510383501 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Vaccarino V, Badimon L, Corti R, de Wit C, Dorobantu M, Hall A, Koller A, Marzilli M, Pries A, Bugiardini R. Ischaemic heart disease in women: are there sex differences in pathophysiology and risk factors? Position paper from the working group on coronary pathophysiology and microcirculation of the European Society of Cardiology. Cardiovasc Res. 2011;90:9–17. doi: 10.1093/cvr/cvq394 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Committee on the Recommended Social and Behavioral Domains and Measures for Electronic Health Records . Capturing Social and Behavioral Domains in Electronic Health Records. National Academies Press; 2014. [PubMed] [Google Scholar]
- 41. Dickens AP, Richards SH, Greaves CJ, Campbell JL. Interventions targeting social isolation in older people: a systematic review. BMC Public Health. 2011;11:647. doi: 10.1186/1471-2458-11-647 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Gardiner C, Geldenhuys G, Gott M. Interventions to reduce social isolation and loneliness among older people: an integrative review. Health Soc Care Community. 2018;26:147–157. doi: 10.1111/hsc.12367 [DOI] [PubMed] [Google Scholar]