- Magazine list
- HHS author manuscripts
- PMC4368483
As a library, NLM offers access to scientific literature. Inclusion in any NLM database does not imply approval or endorsem*nt of the contents by NLM or the National Institutes of Health.
Learn more:PMC-disclaimer|PMC Copyright Notice
J Health as you grow older.Author manuscript; available in PMC 2015 April 1.
Published in final edited form as:
Published online September 23, 2014. doi:10.1177/0898264314551172
Yang Claire Yang, PhD,1,2,3 Courtney Boen, MA, mph,1,2InKathleen Mullan Harris, PhD1,2
Author information Copyright and licensing information PMC-disclaimer
The publisher's final edited version of this article is available atJ Health as you grow older
Abstract
Goal
Social relationships are widely believed to be important in maintaining and improving health and longevity, but it remains unclear how different dimensions of social relationships operate through similar mechanisms to influence biophysiological markers of aging-related diseases over time.
Methods
This study used longitudinal data from a nationally representative sample of older adults from the National Social Life, Health, and Aging Project (2005 – 2011) to examine the potential associations between social integration and social support and change in systolic blood pressure and the risk of hypertension over time. time .
Results
Although both dimensions of social relationships have significant physiological effects, their relative importance varies by outcome. Low social support was predictive of increased systolic blood pressure, while low social integration was predictive of increased risk of hypertension.
discussion
The differential roles of relationship characteristics in predicting changes in physiological outcomes suggest specific biophysiological stress responses and behavioral mechanisms that have important implications for both the scientific understanding and effective prevention and management of an important chronic condition in later life.
Keyword:social integration, social support, blood pressure, hypertension, stress response, longitudinal analyses
Introduction
Understanding the role of social connections in shaping health and well-being has been a focus of scientific inquiry since the late 1970s, and a wide range of research from various disciplines now provides overwhelming evidence that social relationships are essential for maintaining and improving of health, functioning and longevity in society. kind (Berkman et al., 2000;Cassel, 1976;Cobb, 1976;McClintock et al., 2005). A growing body of literature has further linked lack of social integration to specific disease conditions such as cardiovascular disease.Berkman et al., 1993;Orth-Gomer et al., 1993;Eng et al., 2002), Cancer (), depression (George et al., 1989;), biomarkers of physiological dysregulation such as inflammation (Kiecolt-Glaser et al., 2010;Yang et al., 2013) and infection (Cohen et al., 1997; Pressman et al., 2005), as well as higher rates of total and cause-specific mortality (;Holt-Lunstad et al., 2010;Thoits, 1995;Yang et al., 2013). Meta-analyses show that the mortality risk of social isolation is comparable to the risk associated with cigarette smoking and obesity (;Holt-Lunstad et al., 2010Research also suggests that social relationships may be particularly important for health in later life, as individuals transition into different social roles and experience more stressful life events, such as the loss of a spouse or friends.;).
Although research on the health effects of social relationships has increased in recent years, three critical gaps remain in the literature. First, although social relationships are generally conceptualized as multidimensional and their association with health is believed to be multifaceted, few empirical analyzes have simultaneously incorporated multiple measures of different aspects of social relationships in studies of large population-based samples. A wide variety of indicators of social relationships have been used in studies that fall under the broad umbrella of social support. The literature generally distinguishes between the structural and functional dimensions of social relationships. While structural dimensions indicate the pattern and organization of social ties, the functional dimension of social connections generally refers to the social support that an individual receives or perceives from their social ties.Thoits, 2011;Uchino, 2004). Empirical analyzes measure the structural dimension of social relationships as the amount or number and nature of ties in an individual's social network that define levels of social integration or disconnection. Examples of indicators include housing arrangements such as living alone (Magaziner et al., 1988;Dean et al., 1992;), Marital status(;Litwak et al. 1989), network reach or diversity (), frequency of contact with network members (Brummttet et al., 2001), and the size of the social network, which includes and summarizes all the above domain-specific ties (,). Measures of the functional dimension of social relationships, including emotional, informational, and instrumental functions performed by network members, typically include perceived levels of social support.) and feelings of loneliness (Hawkley et al., 2006;Cacioppo et al., 2002).
Restriction to single measures of social relationships in health and well-being research therefore has limitations. First, this makes it difficult to identify the unique influences of certain network or support characteristics on health (Although research generally shows that the links between social integration and health are strong, counting social ties is not informative about the actual function or quality of social ties. Individuals may still experience isolation despite having many social relationships; while conversely, having one or two close social connections can lead to greater perceived support (Kiecolt-Glaser et al., 2010). Some studies that have focused on the functional aspects of social support – that is, the role of networks in providing support – have ignored the structural aspects, so less is known about how an individual network of social relationships – the number, the nature, types and characteristics of relationships – influences health (). There is evidence that multidimensional measures that include both structural and functional aspects of social relationships are better predictors of mortality than either alone.Holt-Lunstad et al., 2010). Furthermore, the focus on individual measures of social relationships makes it difficult to see the link between different dimensions of integration and support in relation to specific health indicators. For example, it remains unknown whether social withdrawal and perceived isolation are two independent processes that contribute to health risk, or whether perceived isolation simply reflects the degree of actual isolation.Steptoe et al., 2013). Some recent studies that have assessed both structural and functional aspects of social relationships show that the number of social network connections is more importantly related to mortality than to perceived social support.Stringhini et al., 2012) or loneliness (Steptoe et al., 2013). Whether social integration and social support are two independent or interrelated measures of social relationships that jointly contribute to other markers of disease risk deserves further investigation.
The second major gap is that while much research has documented the association between social connections and self-reported health outcomes, the biobehavioral mechanisms underlying the observed associations, such as those related to physiological stress responses in older adults, have not been well specified or tested. (Yang et al., 2013). Blood pressure is of particular interest to this study for several reasons. First, hypertension is a powerful risk factor for cardiovascular disease, heart attack, stroke, and kidney disease, all of which are strong predictors of mortality among the elderly.). Because approximately 67% of adults over age 60 have high blood pressure (Yoon et al., 2012), hypertension plays an important role in shaping the health trajectories of older adults (). Second, blood pressure is an indicator of physiological stress response. The activity of both the sympathetic nervous system (SNS) and the hypothalamic-pituitary-adrenocortical (HPA) systems increases in response to environmental or social stressors.So 1974;). In addition to releasing stress hormones, the SNS and HPA systems also play a crucial role in regulating blood pressure (Christou et al., 2005;Hawkley et al., 2006). Although transient activation of the SNS and HPA in response to acute injury or pathogens is necessary to maintain healthy metabolic and immune function, the increased frequency, intensity, and duration of SNS and HPA activity that results from exposure to chronic, low-grade stress associated with the development and worsening of hypertension (Hawkley et al. 2006;McEwen, 1998;;;Whitworth et al., 1995).
An examination of the link between blood pressure and social relationships may therefore shed light on the biological mechanisms by which social relationships influence health. On the one hand, the positive affect and effective coping associated with strong social connections can moderate the physiological arousal caused by social and environmental stressors by increasing self-esteem, improving self-efficacy, and promoting a sense of connection and camaraderie.Thoits, 2011;Uchino, 2004). On the other hand, social relationships may also influence disease risk through behavioral mechanisms where individuals receive behavioral guidance through their interactions with their network members (; Marsden & Friedkin, 1993;;). Individuals may feel pressured or encouraged by their network members to adopt health-promoting behaviors and gain meaning by conforming to norms and fulfilling socially mandated roles.Thoits, 2011). Overall, the mechanisms linking social relationships to health remain to be better understood in objective measures of biophysiological functioning in later life.
Finally, most existing findings regarding associations between social relationships and biomarkers of physical health are based on cross-sectional studies. The potential problem of reverse causality is difficult to address using cross-sectional designs. There is evidence from animal and laboratory studies that an increased stress response, indicated by an increase in chronic inflammation, can affect the brain via the vagus nerve and increase illness behaviors such as withdrawal from social interactions and depression.). Thus, it is possible that stress-related illnesses, such as high blood pressure, limit social activities and result in social isolation or lack of support. The lack of longitudinal research using prospective biomarker data limits researchers' ability to draw causal conclusions about the effects of social relationships on physiological regulation and disease risk.
In summary, it remains unclear how different dimensions of social relationships operate through similar or different mechanisms to influence biophysiological markers of aging-related diseases over time. This study fills these gaps by conducting a parallel analysis of the potential associations between social integration and social support and change in risk for blood pressure and hypertension, using panel data from a nationally representative sample of older adults in the United States.
Data and methods
Data and example
Data come from the National Social Life, Health and Aging Project (NSHAP), a nationally representative longitudinal study of community-dwelling older adults aged 57–85 years in 2005–2006 (wave 1) and followed in 2010–2011 (wave 1). 2) ). African Americans, Latinos, men, and older adults (75–84 years at the time of screening) were oversampled. A full description of the NSHAP study design can be found in. The overall response rate for Wave 1 was 75.5%, and the response rate for eligible respondents reinterviewed at Wave 2 was 87.8%. TheNSHAP collected extensive information about respondents' intimate social relationships, physical and mental health, health behaviors and medication use, mainly through in-depth home interviews. The NSHAP also includes several biomarkers measured at waves 1 and 2 for a subset of the sample.
The original Wave 1 sample included 3,005 respondents, and 2,261 of these Wave 1 respondents were re-interviewed during Wave 2. This survey included 1,264 individuals who had data on all variables used in the analysis. Among respondents excluded from the study, most were missing complete data on measures of social relationships (471 missing) and smoking (318 missing). Compared to respondents in the final sample, those excluded were more likely to be black or Hispanic (S< 0.001), had a lower education level (S< 0.001), lower BMI's (S< 0.001), higher systolic blood pressure (S=0.003 for wave 1;S= 0.020 for Wave 2) and higher rates of hypertension (S= 0.030 for wave 1;S= 0.001 for wave 2). To the extent that the analytic sample had higher SES and was healthier, the results of our analyzes may be conservative given the positive associations between high SES and social network support (Stringhini et al., 2012).
Dependent variables: systolic blood pressure and hypertension
Blood pressure (BP) is the outcome of interest for this study. As part of the biomarker collection process, respondents took two or three seated blood pressure measurements on their left arm. We used the average of each respondent's blood pressure measurements based on previous studies of this measure in the NSHAP (Cornwell en Waite, 2012). Blood pressure measurements were available for both waves 1 and 2. We included bothsystolic blood pressure (SBP)Inhypertensionas the dependent variables in our analyses. Following previous research among older adults (Kulminski et al. 2013), we used a log-transformed SBP measure to adjust for bias. Hypertension is defined as blood pressure (untransformed) above clinical limits (systolic ≥ 140 mmHg or diastolic ≥ 90 mmHg) or if high blood pressure has ever been diagnosed by a doctor.
We included both continuous and dichotomous measures of BP to test whether different dimensions of social relationships influence physiological function (as measured by SBP) and disease risk (as measured by hypertension) in similar or different ways. Because SBP is a measure of metabolic and cardiovascular function, our SBP analyzes shed light on the ways in which social integration and support may influence the physiological stress response. While younger patients are more likely to have high diastolic blood pressure, SBP is affected by increased arterial stiffness, contributing to a high incidence of systolic hypertension after middle age.Franklin et al., 1997). Unlike diastolic blood pressure, which has been shown to remain constant or decrease after age 50-60, SBP dominates and continues to increase in later life.Franklin et al., 1997;Burt et al., 1995). SBP alone has been shown to correctly classify hypertension status in 99% of older-age subjects (Lloyd-Jones et al., 1999Furthermore, previous epidemiological research suggests that SBP is a much stronger predictor of subsequent vascular and other chronic disease events and mortality across the life course.Yang en Kozloski, 2012).
Our second outcome of interest – hypertension – is a measure of disease. While SBP more directly measures the physiological response on a continuum, hypertension reflects the status of BP at the clinical threshold. It thus includes both physiological processes and health-related factors such as diagnosis and management. In this way, using hypertension as an outcome can inform us about factors that encourage individuals to undergo preventive screenings and contribute to disease management (). Thus, hypertension analyzes seek to improve understanding of how social relationships influence disease onset, diagnosis, and treatment.
Social relationship measures
We measured social relationships using two composite scales.Social integrationreflects the structural dimension of social relationships and indicates the number and nature of a person's social relationships. We measured social integration as a summary index of six measures of social connectedness: marital status, religious participation, frequency of social contacts with family and friends, frequency of volunteer work, frequency of social contacts with neighbors, and participation in organized gatherings.Social supportmeasures the functional dimension of social connections and indicates how respondents think about the quality of their social ties. The Social Support Scale is a summary index of six measures of relationship quality: how often a respondent can open up to their partner, family and friends; and how often a respondent can trust their partner, family and friends. Both measures are consistent with those used in previous studies of social relationships and health (;,2012;Yanget al., 2013,2014) and is further confirmed by a principal component analysis using oblique models. The social integration index included two factors (number of social ties with self-esteem = 1.83; interaction with social ties with self-esteem = 1.21) explaining 50.75% of the total variance. The social support scale consisted of two factors (support from spouse with Eigenesteem=1.73; support from family/friends with Eigenesteem=1.90) that explained 60.45% of the total variance. Each of the item residual correlations for the individual measures used in the scales exceeded the cutoff of 0.20, which is considered satisfactory for reliability (Kline, 1986). The social relations scales were constructed using social relations data collected at Wave 1. Detailed information on the construction of these scales can be found intable 1. In addition to continuous measures, each scale was also included as a three-level categorical measure (1=low, 2=moderate, 3=high) to capture nonlinearity in their effects on the outcome variables.
table 1
Content and coding of social integration and support measures
| Scale overview | Are | Item coding | Scale coding |
|---|---|---|---|
| Social integration | Frequency of socializing with family or friends | 1= ≥eenmaal per week; 0=anders | The sum of 6 elements (0-6): |
| Marital status | 1=married/cohabiting; 0=other | 1=lav (0-1) | |
| Frequency of socializing with neighbors | 1= ≥several times a month; 0=other | 2=moderate (2-4) | |
| Frequency of religious attendance | 1= ≥elke week; 0=anders | 3=high (5-6) | |
| Participate frequently in organized meetings | 1= ≥eenmaal per week; 0=anders | ||
| Frequency Volunteer work | 1= ≥once a month; 0=other | ||
| Social support | How often can you open up to your partner? | 1 = often; 0=sometimes or rarely | The sum of 6 elements (0-6): |
| How often can the husband be called upon? | 1 = often; 0=sometimes or rarely | 1=lav (0-1) | |
| How often can you open up to the family? | 1 = often; 0=sometimes or rarely | 2=moderate (2-4) | |
| How often can the family be called upon? | 1 = often; 0=sometimes or rarely | 3=high (5-6) | |
| How often can you open up to friends | 1 = often; 0=sometimes or rarely | ||
| How often can you trust friends | 1 = often; 0=sometimes or rarely | ||
Covariate
We adjusted for factors that have been associated with blood pressure in previous research, including demographic characteristics (age, gender, and race), education, use of antihypertensive medications, psychosocial stressors (perceived social stress and depressive symptoms), health behaviors (smoking, physical activity ). and drinking) and physical conditions (BMI and diabetes). All covariates were assessed at wave 1. In preliminary cross-sectional analyzes using only baseline data, we also experimented with a broader definition of hypertension by including those taking antihypertensive medications. The results are not sensitive to the choice of alternative definitions. Because information on medications at Wave 2 was not available at the time of analysis, we could not use this definition for the final longitudinal analyses. However, checking for medication at baseline effectively takes into account its effect on the change in blood pressure and hypertension later in time. Additional information on the coding and distribution of all the above measures can be found intable 2.
table 2
Sample characteristics and descriptive statistics-in: NSHAP 2005-2011 (N = 1,264)
| Bee | Social integration | Social support | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Low (N=267) | Moderate (N=585) | High (N=412) | p-valueB | Low (N=191 | Moderate (N=541) | High (N=532) | p-valueB | ||
| Systolic blood pressure (SBP) (mmHg), mean (SD) 2005-2006 | 135,68 (20,21) | 138,67 (19,88) | 135,73 (20,91) | 133,77 (19,20) | 0,019 | 135,47 (20,02) | 137,27 (21,06) | 134,22 (19,31) | 0,076 |
| Systolic blood pressure (SBP) (mmHg), mean (SD) 2010–2011 | 137,19 (21,90) | 136,69 (21,46) | 137,92 (21,60) | 136,49 (22,48) | 0,664 | 139,66 (22,39) | 138,77 (22,91) | 134,83 (20,49) | 0,014 |
| Hypertension,%, 2005-2006 | 69,6 | 81,9 | 69,4 | 62,5 | <0,001 | 75,0 | 71,7 | 65,8 | 0,081 |
| Systolic blood pressure ≥140 mmHg, % | 38,9 | 46,2 | 38,5 | 34,8 | 0,100 | 40,1 | 41,4 | 36,0 | 0,293 |
| Diastolic blood pressure ≥ 90 mmHg, % | 24,0 | 30,8 | 22.9 | 21.2 | 0,060 | 26.4 | 23.8 | 23.3 | 0,839 |
| Ever diagnosed with high blood pressure, % | 53,2 | 65,9 | 52,5 | 46,3 | <0,001 | 54,8 | 55,7 | 50,3 | 0,278 |
| Hypertension,%, 2010-2011 | 72,4 | 82,9 | 72,0 | 66,4 | 0,005 | 78,1 | 75,4 | 67,6 | 0,013 |
| Systolic blood pressure ≥140 mmHg, % | 41,6 | 40,1 | 41,9 | 42,1 | 0,926 | 43,6 | 45,4 | 37,2 | 0,101 |
| Diastolic blood pressure ≥ 90 mmHg, % | 19.7 | 23.1 | 20.3 | 16.7 | 0,210 | 21.8 | 18.5 | 20.2 | 0,730 |
| Ever diagnosed with high blood pressure, % | 55,8 | 73,1 | 52,9 | 48,8 | <0,001 | 60,0 | 59,8 | 50,4 | 0,037 |
| Demographic characteristics | |||||||||
| Age, mean (SD) | 67,29 (7,79) | 66,15 (8,02) | 67,72 (7,87) | 67,41 (7,46) | 0,087 | 68,77 (8,82) | 68,00 (7,94) | 66,10 (7,09) | <0,001 |
| Gender (1= female), % | 53,0 | 46,2 | 51,5 | 59,3 | 0,066 | 52,4 | 49,4 | 56,7 | 0,147 |
| Race/Ethnicity, % | |||||||||
| wit | 87,0 | 81,6 | 89,1 | 87,5 | 0,032 | 84,5 | 83,7 | 91,0 | 0,001 |
| Kind | 7.1 | 10.4 | 4.7 | 8.5 | 7.5 | 8.8 | 5.4 | ||
| Spanish speaking | 5.9 | 8,0 | 6.2 | 4.1 | 8,0 | 7.5 | 3.6 | ||
| Education (1= BA degree or higher), % | 30,8 | 25.8 | 30.1 | 34,8 | 0,214 | 29.4 | 29,9 | 32,0 | 0,799 |
| Psychosocial stressors | |||||||||
| Scale of perceived social stress (range= 0 – 12), mean (SD) | 1,53 (2,13) | 2,06 (2,43) | 1,50 (2,08) | 1,24 (1,94) | <0,001 | 2,71 (1,57) | 2,67 (2,07) | 1,09 (1,81) | <0,001 |
| Depressive Symptoms (CES-D) (range= 0 – 27), mean (SD) | 4,26 (4,36) | 5,75 (5,07) | 4,17 (4,38) | 3,46 (3,57) | <0,001 | 6,33 (5,29) | 4,29 (4,25) | 3,53 (3,87) | <0,001 |
| Health behavior | |||||||||
| Use of antihypertensive medications, % | 57,1 | 65,2 | 55,2 | 54,5 | 0,083 | 58,3 | 59,1 | 54,7 | 0,466 |
| Cigarette use (1=former/current smoker), % | 52,3 | 53,9 | 55,7 | 46,9 | 0,143 | 51,4 | 53,1 | 52,0 | 0,929 |
| Physical activity (1=3+ times/week), % | 67,6 | 56,0 | 64,7 | 78,7 | <0,001 | 62,1 | 68,1 | 69,0 | 0,321 |
| Alcoholic drinks per week, mean (SD) | 3,18 (6,17) | 3,62 (6,97) | 3,56 (6,56) | 2,40 (4,95) | 0,006 | 3,30 (6,64) | 3,58 (6,83) | 2,75 (5,29) | 0,106 |
| Physical conditions | |||||||||
| Body mass index (BMI), mean (SD) | 29,65 (6,32) | 30,94 (6,72) | 29,38 (6,62) | 29,22 (5,52) | 0,011 | 29,59 (5,64) | 29,82 (6,71) | 29,51 (6,14) | 0,671 |
| Diabetes (1=diabetes), % | 17.3 | 20.9 | 13.7 | 19.9 | 0,143 | 16.2 | 17.7 | 17.3 | 0,927 |
-inSurvey design adjusted and weighted to account for probability of selection, with post-stratification adjustments for non-response
Bp-value of the t-test or chi-square test of the difference in means between individuals with low, medium and high levels of social integration and support
Note: BP measurements are not transformed; Covariates, including demographics, psychosocial stressors, health behaviors, and physical conditions, were assessed at baseline during 2005–2006.
Methods
In descriptive analyses, we compared sample characteristics across levels of social relationship measures usingTtest and chi-square test. We performed both cross-sectional and longitudinal multivariate analyses. First, we examined the associations between measures of social relationships, including social integration and social support, and blood pressure measures at wave 1. We estimated OLS regression models for log SBP and logistic regression models for hypertension. We then used longitudinal residual change models, also called lagged dependent variable models (Allison, 1990;Halabi, 2004), to examine the effects of social integration and social support at baseline on changes in blood pressure measurements over time. In these models, we regressed Wave 2 blood pressure measures on Wave 1 blood pressure measures and Wave 1 social relationships measures.
We estimated the models in a stepwise manner: 1) basic models including blood pressure measurements and covariates at baseline; 2) models that added social integration; 3) models that added social support; and 4) final models including social integration, social support, and all covariates. The presence of both structural and functional dimensions of social relationships allowed us to further assess potential mediating or moderating effects. Social support may mediate the effects of social integration, in that greater integration may provide the support that results in health benefits. We tested the hypothesized mediating effect using the Sobel-Goodman meditation test (Sobel 1982,1986). According to this test, mediation occurs when (1) the independent variable (social integration) significantly affects the mediator (social support), (2) the independent variable (social integration) significantly affects the dependent variable (BP) in the absence of the mediator (social support), (3) the mediator (social support) has a significant unique effect on the dependent variable (BP), and (4) the effect of the independent variable (social integration) on the dependent variable (BP) decreases when adding the mediator (the support) to the model. We further compared the goodness-of-fit statistics between the models for each outcome. A moderating effect suggests that social support would be more (or less) beneficial for those who had no (or sufficient) social integration. We weighted this by including interaction terms between social integration and social support. The selection of continuous and categorical measures for social relationship variables was based on tests of statistical significance for coefficient estimates and model fit statistics.
All analyzes were conducted in Stata 12 and adjusted for study design effects and non-response using sampling weights.
Results and findings
Descriptive statistics
Although high blood pressure and hypertension were quite common in this group of older adults,table 2shows a clear gradient in BP measurements by social relationship status. Those with the highest levels of integration and support differed significantly from those with low levels of integration and support at baseline in terms of their BP profile, both at baseline and during follow-up, and other baseline characteristics, such as demographics, levels of psychosocial stress. , health behavior and physical conditions. First, negative bivariate associations emerged between social integration and BP measures, with less integrated individuals having significantly higher SBP at Wave 1 (S= 0.019) and higher rates of hypertension (S< 0.001) in both waves 1 and 2 than more integrated individuals. The difference in the frequency of hypertension was particularly large. For example, 82.9% of those with low social integration had hypertension at wave 2, compared with 66.4% of those with high social integration. People with a lower level of integration were more often men,S= 0.066), black or Hispanic (S= 0.032), reported higher levels of perceived social stress (S< 0.001) and more depressive symptoms (S<0.001), had less physical activity (S< 0.001), more alcoholic drinks (S= 0.006) and higher BMIs (S= 0.011) than more integrated individuals.
Similar health disparities were present by level of social support. Compared to individuals with higher levels of support, those with low support had significantly higher SBP at baseline and were more likely to have hypertension at both waves. In contrast to the results for social integration, the gradient in SBP by social support was larger and more significant at wave 2 (S= 0.014) than at wave 1 (S= 0.081). The gradient in hypertension rates by level of social support was less pronounced in magnitude than by social integration, but was nevertheless statistically significant at both waves. Elderly (S<0.001) and blacks and Hispanics (S= 0.001) reported significantly less support than younger respondents and whites, respectively. Low support was also positively associated with perceived stress (S< 0.001) and depressive symptoms (S<0.001). There were no significant differences in health behaviors or physical conditions between levels of social support.
Cross-sectional models
The results of cross-sectional analyzes show associations between the two measures of social relationships with log SBP and hypertension risk at Wave 1 that are broadly consistent with those in descriptive analyses. Using categorical variables for social relationships yielded more statistically significant coefficient estimates as well as better model fit than continuous measures.Figure 1shows the corresponding results in terms of predicted blood pressure outcomes with three levels of social integration and social support, adjusting for age, gender, race, and antihypertensive medication. Higher social integration was associated with significantly lower log SBP (Figure 1A:B= 0,035,S=0.011), a pattern not repeated for social support iFigure 1B. Although both higher social integration and social support were associated with a lower risk of hypertension, the association was much stronger and more important for social integration. Compared to the most socially integrated people, those with the lowest levels of integration were more than twice as likely (Odds Ratio {OR} = 2.72,[95% Confidence Interval {CI}]=[1.75, 4.21],S< 0.001) to hypertensive (Figure 1C). Social support was less related to the risk of hypertension than to social integration.Figure 1D). The difference between low and moderate support groups was not statistically significant, but between moderate and high support groups the difference was marginally significant (OR[95%CI] = 1.29 [0.96, 1.75],S= 0,094).
Social integration, social support, and predicted blood pressure outcomes with 95% confidence intervals (CI): cross-sectional associations at Wave1
Note: Figures based on Wave 1 models, adjusted for age, sex, race, and antihypertensive drug use; the model estimates the study design adjusted and weighted to account for the probability of selection, with poststratification adjustments for nonresponse.
Longitudinal models
Compared with cross-sectional associations at baseline, results from longitudinal analyzes provide new findings on the nature and specificity of these associations for different blood pressure outcomes.
table 3shows the results of the longitudinal models of log SBP. Model 1, adjusting only for baseline covariates, shows the strong effect of SBP at wave 1 on SBP at wave 2, which trumps all other risk factors typically assessed in previous cross-sectional studies. In contrast to the significant cross-sectional association between social integration and log-SBP at baseline, model 2 indicates no significant effect of social integration on change in log-SBP from baseline to follow-up. Model 3 shows that, compared to individuals with the highest levels of social support, individuals with the lowest levels of support at Wave 1 experienced a moderate increase in log SBP from Wave 1 to Wave 2 (B=0,030,S= 0.051). When social integration and social support were jointly included in Model 4, the effect of social support on SBP change increased. As shown inFigure 2ARespondents with low and moderate levels of support had greater increases in logSBP than those with the highest levels of support, with increases being greatest for those with the lowest levels of support (B= 0,034,S= 0.027). We note that in these models that include measures of social relationships, only two covariates (age and BMI) have often been examined in previous cross-sectional studies, because risk factors for hypertension show significant prospective associations with change in logSBP.
Social support, social integration, and predicted change in BP scores from wave 1 to wave 2 with 95% CIs
Note: Figures based on longitudinal models adjusting for age, gender, race, use of antihypertensive medications, education, perceived social stress, depressive symptoms, smoking, physical activity, drinking, BMI, and diabetes; model estimates study design adjusted and weighted to account for the probability of selection, with poststratification adjustments for nonresponse. Hypertension scores reflect the predicted probability of hypertension at wave 2 for individuals without hypertension at wave 1.
table 3
Associations between social relationship characteristics and change in LogSBP: NSHAP 2005–2011 (N = 1,264)
| Variable | Model 1 Coefficient (LIKE THIS) | Model 2 Coefficient (LIKE THIS) | Model 3 Coefficient (LIKE THIS) | Model 4 Coefficient (LIKE THIS) |
|---|---|---|---|---|
| log SBP at wave 1 | 0,462***(0,032) | 0,463***(0,033) | 0,460***(0,032) | 0,463***(0,033) |
| Social integration (reference = 3 (high)) | ||||
| 1 (lav) | -0,007 (0,014) | -0,015 (0,014) | ||
| 2 (moderate) | 0,008 (0,013) | 0,006 (0,012) | ||
| Social support (reference = 3 (high)) | ||||
| 1 (lav) | 0,030†(0,015) | 0,034*(0,015) | ||
| 2 (moderate) | 0,017 (0,010) | 0,018†(0,010) | ||
| The | 0,001**(0,0005) | 0,001*(0,0005) | 0,001*(0,0005) | 0,001†(0,0005) |
| Gender (1=female) | 0,0003 (0,008) | 4.49E-05 (0.008) | 0,001 (0,008) | 0,001 (0,008) |
| Race (reference = white) | ||||
| Kind | 0,021 (0,023) | 0,022 (0,022) | 0,019 (0,022) | 0,020 (0,021) |
| Spanish speaking | 0,025 (0,018) | 0,026 (0,017) | 0,021 (0,018) | 0,022 (0,018) |
| Antihypertensives at wave 1 (1=yes) | -0,016 (0,011) | -0,015 (0,011) | -0,015 (0,011) | -0,014 (0,011) |
| Education (1=BA degree or higher) | 0,015 (0,011) | 0,015 (0,011) | 0,014 (0,011) | 0,014 (0,011) |
| Perceived Social Stress Scale | 0,0008 (0,002) | 0,001 (0,002) | -2.71E-04 (0.002) | -2,81E-04 (0,002) |
| Depressive Symptomator (CES-D) | -0,0005 (0,001) | -3,53E-04 (0,001) | -0,001 (0,001) | -0,001 (0,001) |
| Cigarette use (1=former/current smoker) | -0,002 (0,011) | -0,002 (0,011) | -0,002 (0,011) | -0,003 (0,011) |
| Physical activity (1=3+ times per week) | -0,0004 (0,011) | -1.63E-04 (0.011) | -3.61E-04 (0.011) | -0,001 (0,011) |
| Alcoholic drinks per week | -0,0005 (0,0006) | -4.770E-04 (6.22E-04) | -0.001 (6.29E-04) | -0.001 (6.23E-04) |
| Body Mass Index (BMI) | -0,001†(0,0008) | -0,001†(7.71E-04) | -0,001†(7.65E-04) | -0,001†(7.61E-04) |
| Diabetes (1=diabetes) | 0,010 (0,018) | 0,011 (0,017) | 0,011 (0,017) | 0,012 (0,017) |
| Intercept | 2.593***(0,160) | 2.587***(0,162) | 2.613***(0,160) | 2.605***(0,161) |
| Model fit | ||||
| R2 | 0,197 | 0,199 | 0,202 | 0,205 |
***p<0,001,
**p<0,01,
*p<0,05,
†p<0.1 (two-sided test)
Explanation: Unweighted N=1264; the model estimates the study design adjusted and weighted to account for the probability of selection, with post-stratification adjustments for non-response
Table 4presents the results of longitudinal hypertension models. We found that although both social integration and social support individually protected against the increase in hypertension risk, social integration had a stronger and more significant effect that may be mediated by social support. Model 2 shows that respondents with the lowest level of integration in Wave 1 compared to the most socially integrated respondents experienced an increase of 75.3% (95% CI = [1.04,2.95],S= 0.036) in the risk of hypertension from waves 1 to 2. Note that at a six-year interval for an older sample, an effect of change in the risk of hypertension of this magnitude was quite large. As illustrated inFigure 2BAmong those who did not have hypertension during the first wave, those with higher levels of social integration were less likely to develop hypertension from baseline to follow-up than those with lower levels of social integration. Model 3 shows that social support also reduces the risk of hypertension, but the effect is non-linear. Although the increase in hypertension risk was not statistically significant for respondents with low support, respondents with moderate support had a 33.6% (95% CI = [0.95, 1.87]) increase in hypertension risk which was marginally significant (S=0.092) compared to people with high social support. When social integration and social support are simultaneously included in Model 4, the effect of social support no longer holds, while low social integration continues to predict a significantly higher risk of hypertension. In bothTable 3Inin 4,4Although the large majority of the variance was explained by the lagged dependent variables (BP measurements at wave 1), the addition of social relationship variables slightly improved model fit. The result of the Sobel-Goodman test further suggests that conditions 1), 2), and 4) are met such that the effect of social integration on hypertension is partially mediated by social support. Specifically, we found that social integration significantly influences social support (S<0.001); that social integration significantly influences hypertension, in the absence of social support (as indicated in model 2 ofTable 4); and that the effect of social integration on hypertension is reduced by the addition of support to the models (OR for “low integration” decreases from 1.75 to 1.65 from model 2 to model 4 inTable 4). Because the social support coefficients in model 3 were not highly significant, condition 3) failed to reach statistical significance (S= 0.175). The mediation effect in this sense exists statistically in a weak form. The interaction effects between social integration and support variables were not statistically significant in any analysis and were therefore omitted from the final models. Similar to the SBP models, few other covariates show significant prospective associations with hypertension risk. Apart from antihypertensive medication use, social integration was the only significant factor influencing the change in hypertension risk from wave 1 to wave 2.
Table 4
Associations between social relationship characteristics and change in hypertension risk: NSHAP 2005–2011 (N=1,264)
| Variable | Model 1 ELLER (95% BI) | Model 2 ELLER (95% BI) | Model 3 ELLER (95% BI) | Model 4 ELLER (95% BI) |
|---|---|---|---|---|
| Hypertension at wave 1 (1=yes) | 11.16***(6.85 - 18.17) | 10,72***(6,51 - 17,65) | 11.11***(6,86 - 17,98) | 10,74***(6,57 - 17,58) |
| Social integration (reference = 3 (high)) | ||||
| 1 (lav) | 1,75*(1,04 - 2,95) | 1,65†(0,99 - 2,76) | ||
| 2 (moderate) | 1,23 (0,79 - 1,91) | 1,20 (0,78 - 1,85) | ||
| Social support (reference = 3 (high)) | ||||
| 1 (lav) | 1,46 (0,79 - 2,68) | 1,30 (0,71 - 2,37) | ||
| 2 (moderate) | 1,34†(0,95 - 1,87) | 1,29 (0,92 - 1,81) | ||
| The | 1,00 (0,98 - 1,02) | 1,00 (0,98 - 1,03) | 1,00 (0,97 - 1,02) | 1,00 (0,98 - 1,03) |
| Gender (1=female) | 1,03 (0,72 - 1,48) | 1,08 (0,77 - 1,52) | 1,06 (0,74 - 1,51) | 1,10 (0,79 - 1,54) |
| Race (reference = white) | ||||
| Kind | 1,52 (0,75 - 3,08) | 1,49 (0,76 - 2,93) | 1,46 (0,72 - 2,96) | 1,44 (0,74 - 2,83) |
| Spanish speaking | 1,25 (0,68 - 2,30) | 1,19 (0,65 - 2,17) | 1,17 (0,63 - 2,18) | 1,13 (0,61 - 2,09) |
| Antihypertensives at wave 1 (1=yes) | 1,64**(1,17 - 2,31) | 1,64**(1,17 - 2,29) | 1,65**(1,16 - 2,33) | 1,64**(1,17 - 2,31) |
| Education (1=BA degree or higher) | 1,00 (0,70 - 1,43) | 1,01 (0,70 - 1,44) | 0,99 (0,69 - 1,41) | 1,00 (0,70 - 1,43) |
| Perceived Social Stress Scale | 1,00 (0,90 - 1,12) | 1,00 (0,90 - 1,12) | 0,99 (0,88 - 1,11) | 0,99 (0,88 - 1,12) |
| Depressive Symptomator (CES-D) | 1,02 (0,97 - 1,07) | 1,01 (0,96 - 1,06) | 1,02 (0,97 - 1,07) | 1,01 (0,96 - 1,06) |
| Cigarette use (1=former/current smoker) | 1,22 (0,85 - 1,75) | 1,20 (0,83 - 1,72) | 1,21 (0,85 - 1,74) | 1,19 (0,83 - 1,71) |
| Physical activity (1=3+ times per week) | 1,01 (0,66 - 1,55) | 1,07 (0,70 - 1,64) | 1,02 (0,66 - 1,58) | 1,07 (0,69 - 1,65) |
| Alcoholic drinks per week | 0,98 (0,96 - 1,01) | 0,98 (0,95 - 1,01) | 0,98 (0,95 - 1,01) | 0,98 (0,95 - 1,01) |
| Body Mass Index (BMI) | 0,97 (0,94 - 1,01) | 0,97 (0,94 - 1,01) | 0,97 (0,94 - 1,01) | 0,97 (0,94 - 1,01) |
| Diabetes (1=diabetes) | 1,37 (0,92 - 2,05) | 1,40 (0,92 - 2,12) | 1,38 (0,93 - 2,05) | 1,40 (0,93 - 2,11) |
| Intercept | 0,76 (0,08 - 7,40) | 0,56 (0,06 - 5,67) | 0,84 (0,08 - 8,19) | 0,63 (0,06 - 6,40) |
| Model fit | ||||
| Log-likelihood | -565,49*** | -563,19*** | -565,22*** | -563.03*** |
| Psuedo R-squared | 0,231 | 0,234 | 0,232 | 0,235 |
***p<0,001,
**p<0,01,
*p<0,05,
†p<0.1 (two-sided test)
Explanation: Unweighted N=1264; the model estimates the study design adjusted and weighted to account for the probability of selection, with poststratification adjustments for nonresponse.
discussion
Previous research has clearly linked the characteristics of social relationships to various health outcomes in later life. However, much of the evidence is limited to one dimension of social relationships, self-reported rather than objective assessments of physical functioning, and cross-sectional associations due to the lack of repeated longitudinal measurements of biomarkers. This study used longitudinal data from a nationally representative sample of older adults in the United States to examine the association between social integration and social support and changes in blood pressure measurements over a six-year period. It reveals new knowledge about the physiological effects of social relationships over time, and elucidates key biological mechanisms underlying the social gradient in health in old age.
Our study assessed the quantitative and qualitative dimensions of social relationships individually and simultaneously in the context of BP. Consistent with previous findings on the associations between social relationships and general health, our study found associations between low social integration and low social support with adverse blood pressure outcomes. However, results vary by study design, dimension of social relationships, and biomarker results. Cross-sectional analyzes showed that low social integration was much more strongly associated with both higher SBP and hypertension than low social support. These cross-sectional findings suggest that social network size plays an important role in relation to the assessment of relationship quality in shaping health risks, as suggested by some recent studies.Stringhini et al., 2012;Yang et al., 2013). It could also indicate that those who experienced worsening blood pressure had smaller social networks via reverse causality. The longitudinal analysis of change in systolic blood pressure and risk of hypertension helped clarify some of these associations.
In longitudinal analyses, we found that perceived social support is the dominant factor influencing change in SBP, while social integration is the most important factor influencing change in hypertension risk. Furthermore, tests of net effects, when both dimensions of social relationships are considered, demonstrate the link between these concepts and clarify the role of social support. Low social support had a significant impact on increasing SBP, independent of low social integration. That is, the perception of instrumental and functional social support was more important than the number of social connections in the regulation of SBP. The greater number of social connections reduced the risk of hypertension over time, and its effects may occur in part through the increased social support from these connections. The test for the mediation effect of social support shows that mediation statistically exists in a weak form. The lack of significant interaction effects between the two suggests that there is no moderating effect, or that their effects are additive rather than conditional.
The differing findings in longitudinal analyzes of changes in SBP and hypertension further suggest the specificity of the biological mechanisms by which social relationship variables influence physical health characteristics. We included both a continuous measure of blood pressure and a dichotomous measure of hypertension because these had different biological meanings. The significant correlations between social support and the continuous measure of BP suggest that more positive appraisals or better support from network members, regardless of the size of network connections, can dampen physiological arousal in response to social stressors and improve physiological functioning. manifesting in decreases in SBP values along a continuum. Actually,table 2shows that those with higher levels of support had lower SBP than those with lower levels of support in both Waves 1 and 2. It is likely that the SBP influence of social support operates through psychosocial processes.table 2for example, shows that people with more social support reported significantly lower levels of perceived stress and depressive symptoms than people with less social support.
Large social networks or greater social integration, on the other hand, may protect against the manifestation of clinically significant, harmful disease outcomes, such as hypertension. The result insideTable 4that social support mediated the effects of social integration, suggesting that people who were more socially integrated were less likely to experience hypertension, in part because they tended to enjoy higher quality relationships or support that appeared to be effective in lowering from SBP. In addition to social support, social integration may further reduce the risk of hypertension through socially transmitted behavioral mechanisms.table 2indicate that individuals with higher levels of integration engaged in more physical activity, drank fewer alcoholic beverages, and had a lower BMI than those who were less integrated. In contrast, those with varying degrees of perceived support did not differ in these behavioral risk factors for hypertension. There are likely other behaviors not included in this study that may play a role. What the available data shows is that simply being embedded in larger social networks can yield significant health benefits. It can be the most effective way to promote preventive behavior among network members, directly contributing to lower rates of disease or better treatment and management of the disease condition.
In summary, using both cross-sectional and longitudinal study designs, we can begin to understand the interdependent mechanisms involved in structural and functional sources of social support in relation to a physiological measure of health and disease risk such as blood pressure. Structural social support through social attachment and the number of social ties is a necessary condition for promoting health, because without social connections the functional aspects of social support are not possible. Our results show this in both cross-sectional and longitudinal analyses. However, in addition to the presence of social connections, the functional dimensions of social support are also important, because our results show that part of the beneficial effect of having social connections arises from the way these connections promote physiological function and confer disease risks bring over. Having social connections is therefore a necessary but not sufficient condition, because what someone gets from social connections in terms of health knowledge and behavior is also important. Although both dimensions of social relationships have been associated with self-reported general health outcomes in cross-sectional data, our study is the first to test how they may work independently and jointly to moderate changes in an objectively measured marker of cardiovascular health over time. to influence.
It is notable that although the descriptive analysis shows significant bivariate associations between most of the covariates included in the study and the blood pressure outcomes, the multivariate analyzes show that few significantly influence blood pressure changes over time. Apart from taking antihypertensive medication, social integration and social support had the largest effect on blood pressure changes at baseline, effects that were larger than general biobehavioral indicators of physical health, such as age and BMI, and were of significant magnitude in the change models (75% to 65%). increase in the risk of hypertension for the lowest social integration groupTable 4). This is strong longitudinal evidence that highlights the crucial nature of social relationships in protecting against increases in blood pressure and the onset of disease in old age, independent of other known risk factors. These findings underscore the critical importance of considering social relationships in future studies of physiological health indicators as individuals age. However, that the residual effects of social relationship measures remained significant in the final models adjusted for all covariates also suggests the need to consider more psychosocial, behavioral, or biological mechanisms in future studies.
There are several other limitations of this study that invite future research. First, there are residual change models, which although widely used in longitudinal analyzes (e.g.;Umberson et al., 2014), is not without limitations. In particular, the greatest threat to causal inference in these models comes from unobserved variables, where the relationship between the individual error term and the outcome measured at wave 1 remains unknown (Halabi, 2004). To confirm the robustness of the results from the residual change models, we performed additional multinomial logistic regression analyzes for the hypertension outcome, where change in hypertension status was modeled as an outcome in four categories: never hypertensive; no hypertension at wave 1 to hypertensive at wave 2; hypertensive at wave 1 to no hypertension at wave 2; and always hypertension. The results of the additional analyzes were substantially consistent with the residual change models, indicating that our results are robust to alternative modeling strategies. Nevertheless, the results of the other change models are presented inTable 3Inin 44must be interpreted in the context of their potential limitations. Second, although we found evidence for longitudinal associations between social relationships and changes in BPD outcomes over time, the evidence is limited by the data design, which includes a relatively small sample used only once over a period of time. was followed for a few years. The older adults in the sample were hom*ogeneous in baseline BP measurements and had lower SBP levels and hypertension rates than those with missing data. The slightly truncated range of the outcome variables could have minimized change or variation over a six-year period, although even in this setting we found a remarkably large effect of low social integration on hypertension risk (an increase of 75%). Additionally, the analytic sample had higher SES and was generally healthier. Although the use of sample weights still resulted in national estimates, the results are conservative and need to be confirmed in future studies of other large population-based samples assessed at multiple time points over a longer follow-up period. Third, the included biomarkers are quite limited due to a lack of availability in both waves. It was suggested that a broader spectrum of markers of physiological dysregulation in multiple body systems could better characterize the biological stress response in older adults.Seeman et al., 2001) and would thus be more informative about the interrelatedness of social and biological variations over the life course (Yang en Kozloski, 2011). Finally, because the social relationship measures we used include an extensive set of variables for which no repeated measures are currently available for Wave 2, we were unable to assess the extent to which these measures are good indicators of the quantity and quality of long-term relationships. We are also limited in the ability to model the cumulative effects of long-term relationship characteristics on borderline status over time. The future availability of longitudinal data that include repeated assessments of both social and biomarker variables is critical to elucidating the complex pathways underlying these and other similar biosocial associations.
In conclusion, the significance of the problem we investigated and the results we achieved can be better understood in the broader demographic and epidemiological context. As the United States population continues to age, effective prevention, treatment, and management of chronic diseases remain critical to maintaining the nation's health and well-being. About 2 in 3 older adults in the United States have high blood pressure, putting a large and growing portion of the older population at increased risk for a variety of health problems, including cardiovascular disease, heart attack, stroke, kidney disease and death. find that social relationships can reduce the physiological stress response, promote healthy behaviors, and in turn prevent or delay the onset of hypertensive diseases as individuals age. The development and maintenance of robust, supportive networks may therefore be critical to promoting healthy aging.
Recognitions
This research is supported by National Institute of Aging grant number K01AG036745 and University Cancer Research Funds from the Lineberger Cancer Center (to the first author). We are grateful for general support from the Carolina Population Center (R24 HD050924), University of North Carolina at Chapel Hill.
References
- Allison police. Change scores as dependent variables in regression analysis.Sociological method.1990: 93-114. doi:10.2307/271083.[CrossRef][Google Scholar]
- Berkman LF, Glas T, Brissette I, Seeman TE. From social integration to health: Durkheim in the new millennium.Social Sciences and Medicine.2000;51(6):843-857. doi:10.1016/s0277-9536(00)00065-4.[PubMed] [CrossRef][Google Scholar]
- Berkman LF, Syme L. Social networks, host resistance, and mortality: a nine-year follow-up study among Alameda County residents.American Journal of Epidemiology.1979;109(2):186-204.[PubMed][Google Scholar]
- Berkman LF, Vaccarino V, Seeman T. Gender differences in cardiovascular morbidity and mortality: the contributions of social networks and support.Annals of Behavioral Medicine.1993;15(2-3):112-118. [Google Scholar]
- Brummett BH, Barefoot JC, Siegler IC, Clapp-Channing NE, Lytle BL, Bosworth HB, Williams Redford B, Mark DB. Characteristics of socially isolated patients with coronary artery disease at increased risk of mortality.Psychosomatic medicine.2001;63(2):267-272.[PubMed][Google Scholar]
- Burt VL, Whelton P, Roccella EJ, Brown C, Cutler JA, Higgins M, Horan MJ, Labarthe D. Prevalence of hypertension in the US adult population: results from the third National Health and Nutrition Examination Survey, 1988-1991.Hypertension.1995;25(3):305-313. doi:10.1161/01.HYP.25.3.305.[PubMed] [CrossRef][Google Scholar]
- Cacioppo JT, Hawkley LC, Crawford LE, Ernst JM, Burleson MH, Kowalewski RB, Malarkey WB, Van Cauter E, Berntson GG. Loneliness and health: potential mechanisms.Psychosomatic medicine.2002;64(3):407-417.[PubMed][Google Scholar]
- Cassel J. The contribution of the social environment to harboring resistance.American Journal of Epidemiology.1976;104:107-123.[PubMed][Google Scholar]
- Christou DD, Jones PP, Jordan J, Diedrich A, Robertson D, Seals RD. Women have lower tonic autonomic support of arterial blood pressure and a less effective baroreflex buffer than men.Circulation.2005;111:494-498. doi:10.1161/01.CIR.0000153864.24034.A6.[PubMed] [CrossRef][Google Scholar]
- Cobb S. Social support as a moderator of life stress.Psychosomatic medicine.1976;38(5):300.[PubMed][Google Scholar]
- Cohen SD, William J, Skoner DP, Rabin BS, Gwaltney JM., Jr. Social ties and susceptibility to colds.JAMA: The Journal of the American Medical Association.1997;277(24):1940-1944. doi:10.1001/jama.1997.03540480040036.[PubMed] [CrossRef][Google Scholar]
- Cornwell B, Laumann EO, Schumm LP. The social connectedness of older people: a national profile.American Sociological Review.2008;73(2): 185. doi:10.1177/000312240807300201. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Cornwell EY, Waite LJ. Connection, perceived isolation, and health among older adults.Journal of Health and Social Behavior.2009;50(1):31-48. doi:10.1177/002214650905000103. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Cornwell EY, Waite LJ. Social network resources and the management of hypertension.Journal of Health and Social Behavior.2012;53:215-231. doi:10.1177/0022146512446832. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Dean A, Kolody B, Wood P, Matt G. The impact of living alone on depression in older adults.Journal of Aging and Health.1992;4(1):3–18. doi:10.1177/089826439200400101.[CrossRef][Google Scholar]
- Eng PM, Rimm EB, Fitzmaurice G, Kawachi I. Social ties and change in social ties in relation to subsequent total and cause-specific mortality and coronary heart disease in men.American Journal of Epidemiology.2002;155(8):700-709. doi:10.1093/aje/155.8.700.[PubMed] [CrossRef][Google Scholar]
- Franklin SS, Gustin W, Wong ND, Larson MG, Weber MA, Kannel WB, Levy D. Hemodynamic patterns of age-related changes in blood pressure: the Framingham Heart Study.Circulation.1997;96(1):308-315. doi:10.1161/01.CIR.96.1.308.[PubMed] [CrossRef][Google Scholar]
- George LK, Blazer DG, Hughes DC, Fowler N. Social support and the outcome of major depression.British Journal of Psychiatrie.1989;154(4):478-485. doi:10.1192/bjp.154.4.478.[PubMed] [CrossRef][Google Scholar]
- Haines Valerie A, Hurlbert Jeanne S. Network supply and health.Journal of Health and Social Behavior.1992;33:254-266. doi:10.2307/2137355.[PubMed] [CrossRef][Google Scholar]
- Halaby C.N. Panel models in sociological research: theory to practice.Annual Review of Sociology.2004;30:507-544. [Google Scholar]
- Hawkley LC, Masi CM, Berry JD, Cacioppo JT. Loneliness is a unique predictor of age-related differences in systolic blood pressure.Psychology and aging.2006;21(1):152-164. doi:10.1037/0882-7974.21.1.152.[PubMed] [CrossRef][Google Scholar]
- Heikkinen RL, Kauppinen M. Depressive symptoms in later life: a 10-year follow-up.Archive for gerontology and geriatrics.2004;38(3):239-250. doi:10.1016/j.archer.2003.10.004.[PubMed] [CrossRef][Google Scholar]
- Holt-Lunstad J, Smith TB, Layton JB. Social relationships and mortality risk: a meta-analytic review.PLoS medicine.2010;7(7):e1000316. doi:10.1371/journal.pmed.1000316. [PMC free article][PubMed] [CrossRef][Google Scholar]
- House JS, Landis KR, Umberson D. Social relationships and health.Science.1988;241(4865):540-545. doi:10.1126/science.3399889.[PubMed] [CrossRef][Google Scholar]
- Hu Y, Goldman N. Mortality differences by marital status: an international comparison.Demographics.1990;27(2):233-250. doi:10.2307/2061451.[PubMed] [CrossRef][Google Scholar]
- Hughes M, Gove WR. Living alone, social integration and mental health.American Journal of Sociology.1981;87(1):48-74. doi:10.1086/227419.[PubMed] [CrossRef][Google Scholar]
- Kiecolt-Glaser JK, Gouin JP, Hantsoo L. Close relationships, inflammation and health.Neuroscience and Biobehavioral Reviews.2010;35(1):33–38. doi:10.1016/j.neubiorev.2009.09.003. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Kline P.A handbook of test construction: introduction to psychometric design.New York: Methuen; 1986.[Google Scholar]
- Kulminski AM, Arbeev KG, Christensen K, Stallard E, Miljkovic I, Barmada M, Yashin AI. Biogenetic mechanisms that predispose to complex phenotypes in parents may function differently in their children.The Journals of Gerontology Series A: Biological Sciences and Medical Sciences.2013;68(7):760-768. doi:10.1093/gerona/gls243. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Litwak E, Messeri P, Wolfe S, Gorman S, Silverstein M, Guilarte M. Organization theory, social support, and mortality: atheoretical convergence.American Sociological Review.1989;54:49-66. doi:10.2307/2095661.[CrossRef][Google Scholar]
- Lloyd-Jones DM, Evans JC, Larson MG, O'Donnell CJ, Levy D. Differential impact of systolic and diastolic blood pressure levels on JNC-VI staging.Hypertension.1999;34(3):381-385. doi:10.1161/01.HYP.34.3.381.[PubMed] [CrossRef][Google Scholar]
- Magasin J, Cadigan DA, Hebel JR, Parry RE. Health and living conditions of older women: Living alone increases the risk of disease.Journal of Gerontology.1988;43(5):M127-M133. doi:10.1093/geronj/43.5.M127.[PubMed] [CrossRef][Google Scholar]
- Maier SF, Watkins LR. Cytokines for psychologists: implications of bidirectional communication between the immune system and the brain for understanding behavior, mood, and cognition.Psychological assessment.1998;105(1):83. doi:10.1037/0033-295X.105.1.83.[PubMed] [CrossRef][Google Scholar]
- Marsden PV, Friedkin NE. Network studies of social influence.Sociological methodological research.1993;22(1):127-151. doi:10.1177/0049124193022001006.[CrossRef][Google Scholar]
- McClintock MK, Conzen SD, Gehlert S, Masi C, Olopade F. Breast cancer and social interactions: identifying multiple environments that regulate gene expression across the lifespan.The Journals of Gerontology Series B: Psychological Sciences and Social Sciences.2005;60(Særnummer 1): 32–41. doi:10.1093/geronb/60.Special_Issue_1.32.[PubMed] [CrossRef][Google Scholar]
- McEwen B.S. Stress, Adjustment and Disease - Allostasis and AllostaticLoad.Annals of the New York Academy of Sciences.1998;840(1):33–44. doi:10.1111/j.1749-6632.1998.tb09546.x.[PubMed] [CrossRef][Google Scholar]
- Merton RK, Rossi AS. Contribution to the theory of reference group behavior. In: Merton RK, editor.Social theory and social structure.New York:NY: Free Press; 1968. pp. 100–1 279–3[Google Scholar]
- O'Muircheartaigh C, Eckman S, Smith S. Statistical design and estimation for national social life, health and aging.The Journals of Gerontology Series B: Psychological Sciences and Social Sciences.2009;64(suppl. 1):i12–i19. doi:10.1093/geronb/gbp045. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Orth-Gomér K, Rosengren A, Wilhelmsen L. Lack of social support and incidence of coronary heart disease in middle-aged Swedish men.Psychosomatic medicine.1993;55(1):37-43.[PubMed][Google Scholar]
- Penwell LM, Larkin KT. Social support and risk of cardiovascular disease and cancer: a qualitative review examining the role of inflammatory processes.Health Psychology Review.2010;4(1):42-55. doi:10.1080/17437190903427546.[CrossRef][Google Scholar]
- Pressman SD, Cohen S, Miller GE, Barkin A, Rabin BS, Treanor JJ. Loneliness, social network size, and immune response to influenza vaccination in first-year college students.Health psychology.2005;24(3):297.[PubMed][Google Scholar]
- Rozanski A, Blumenthal JA, Kaplan J. Impact of psychological factors on the pathogenesis of cardiovascular disease and implications for therapy.Circulation.1999;99:2192-2217. doi:10.1161/01.CIR.99.16.2192.[PubMed] [CrossRef][Google Scholar]
- Seeman TE, Berkman TE. Structural features of social networks and their relationship to social support in older adults: who provides support.Social Sciences and Medicine.1988;26(7):737-749.[PubMed][Google Scholar]
- Seeman TE, McEwen BS, Rowe JW, singer BH. Allostatic load as a marker of cumulative biological risk: the MacArthur studies of successful aging.Proceedings of the National Academy of Sciences of the United States of America.2001;98(8):4770-4775. doi:10.1073/pnas.081072698. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Selye H.Stress without necessity.Philadelphia: JB Lippincott; 1974.[Google Scholar]
- Smith KP, Christakis NA. Social networks and health.Annual Review of Sociology.2008;34:405-429. doi:10.1146/annurev.soc.34.040507.134601.[CrossRef][Google Scholar]
- Sable ME. Asymptotic confidence intervals for indirect effects of structural equation models.Sociological method.1982;13(1982):290-312. doi:10.2307/270723.[CrossRef][Google Scholar]
- Sable ME. Some new results on indirect effects and their standard errors in covariance structure models.Sociological method.1986;16:159-186. doi:10.2307/270922.[CrossRef][Google Scholar]
- Steptoe A, Shankar A, Demakakos P, Wardle J. Social isolation, loneliness and all-cause mortality in older men and women.Proceedings of the National Academy of Sciences.2013;110(15):5797-5801. doi:10.1073/pnas.1219686110. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Stringhini S, Berkman L, Dugravot A, Ferrie JE, Marmot M, Kivimaki M, Singh-Manoux A. Socioeconomic status, structural and functional measures of social support and mortality: the British Whitehall II Cohort Study, 1985-2009.American Journal of Epidemiology.2012;175(12):1275-1283. doi:10.1093/aje/kwr461. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Thoit is PA. Stress, coping and social support processes: where are we? What's next?Journal of Health and Social Behavior.1995;35:53-79. doi:10.2307/2626957.[PubMed] [CrossRef][Google Scholar]
- Thoit is PA. Mechanisms linking social ties and support to physical and mental health.Journal of Health and Social Behavior.2011;52(2):145-161. doi:10.1177/0022146510395592.[PubMed] [CrossRef][Google Scholar]
- Uchino Bert N.Social support and physical health: Understanding the health consequences of relationships.Yale University Press; 2004.[Google Scholar]
- Umberson D, Crosnoe R, Reczek C. Social relationships and health behavior across the life course.Annual Review of Sociology.2010;36(1):139-157. doi:10.1146/annurev-soc-070308-120011. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Umberson D, Williams K, Thomas PA, Liu H, Thomeer MB. Race, gender, and chains of disadvantage: Childhood adversity, social circ*mstances, and health.Journal of Health and Social Behavior.2014;55(1): 20–38. [PMC free article][PubMed][Google Scholar]
- Van Uum SHM, Lender JWM, Hermus ARMM. Cortisol, 11β-hydroxysteroid dehydrogenaser and hypertension.Seminars in vascular medicine.2004;4(2):121-128. doi:10.1055/s-2004-835369.[PubMed] [CrossRef][Google Scholar]
- Wethington E, Kessler RC. Perceived support, received support, and adjustment to stressful events.Journal of Health and Social Behavior.1986;27(1):78-89. doi:10.2307/2136504.[PubMed] [CrossRef][Google Scholar]
- Whitworth JA, Brown MA, Kelly JJ, Williamson PM. Mechanisms of cortisol-induced hypertension in humans.Steroids.1995;60:76-80. doi:10.1016/0039-128X(94)00033-9.[PubMed] [CrossRef][Google Scholar]
- Wolfe B, Evans WN, Seeman TE, editors.The biological consequences of socio-economic inequalities.Russell Sage Foundation; 2012.[Google Scholar]
- Yang Y, George LK. Functional limitations, transitions to disability, and depressive symptoms in later life.Journal of Aging and Health.2005;17(3):263-292.[PubMed][Google Scholar]
- Yang Y, Kozloski M. Gender differences in age trajectories of physiological dysregulation: inflammation, metabolic syndrome, and allostatic load.Journal of Gerontology: Biological Sciences.2011;66A(5):493-500. doi:10.1093/gerona/glr003. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Yang Y, Kozloski M. Changing gender differences in total and cause-specific mortality across the life course in the United States.Annals of Epidemiology.2012;22(2):94-103. doi:10.1016/j.annepidem.2011.06.006. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Yang YC, McClintock MK, Kozloski M, Li T. Social isolation and mortality among adults: the role of chronic inflammation and gender differences.Journal of Health and Social Behavior.2013;54(2):183-203. doi:10.1177/0022146513485244. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Yang YC, Schorpp K, Harris KM. Social support, social tension, and inflammation: Evidence from a national longitudinal survey of US adults.Social Sciences and Medicine.2014;107:124-135. doi:10.1016/j.socscimed.2014.02.013. [PMC free article][PubMed] [CrossRef][Google Scholar]
- Yoon PW, Gillespie CD, George MG, Wall HK. Control of hypertension in adults - National Health and Nutrition Examination Survey, United States, 2005-2008.Weekly Morbidity and Mortality Report.2012;61:19-25.[PubMed][Google Scholar]
