| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Division of Geriatrics (T.E.S., L.L.-S.), School of Medicine, University of California, Los Angeles, California; Office of Population Research (B.H.S.), Princeton University, Princeton, New Jersey; and Department of Psychology (C.D.R.) and Institute on Aging (C.D.R., G.D.L.), University of Wisconsin–Madison, Madison, Wisconsin.
Address reprint requests to: Dr. Teresa Seeman, Division of Geriatrics, UCLA School of Medicine, 10945 Le Conte Ave., Suite 2339, Los Angeles, CA 90095-1687. Email: tseeman{at}mednet.ucla.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: One cohort included older adults aged 70 to 79 years (N = 765); the other cohort included persons aged 58 to 59 years (N = 106). Allostatic load was assessed using identical protocols in the two cohorts. Measures of social experience were similar but not identical, reflecting levels of social integration and support for the older cohort vs. childhood and adult experiences of loving/caring relationships with parents and spouse for the younger cohort. Gender-specific analyses were examined to evaluate possible gender differences in patterns of association.
RESULTS: In the younger cohort, positive cumulative relationship experiences were associated with lower allostatic load for men and women. In the older cohort, men who were more socially integrated and those reporting more frequent emotional support from others had lower allostatic load scores; similar but nonsignificant associations were seen for women.
CONCLUSIONS: Evidence from two cohorts provides support for the hypothesis that positive social experiences are associated with lower allostatic load. These findings are consistent with the hypothesis that social experiences affect a range of biological systems, resulting in cumulative differences in risks that in turn may affect a range of health outcomes.
Key Words: social ties, • social relationships, • social support, • allostatic load, • biology.
Abbreviations: AL = allostatic load;; ANOVA = analysis of variance;; BP = blood pressure;; DHEA-S = dehydroepiandrosterone sulfate;; HDL = high-density lipoprotein;; HPA = hypothalamic-pituitary-adrenal axis;; MAC = MacArthur Studies of Successful Aging;; PAIR = Personal Assessment of Intimacy Relationships Inventory;; SNS = sympathetic nervous system;; WLS = Wisconsin Longitudinal Study.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
Importantly, these health effects of the social environment have been shown to exist across the lifespan. A recent review (9), for example, explored the nature of "unhealthy environments" and how they "get under the skin" both at younger and older ages. With respect to effects on children’s health, for example, this review pointed to three characteristics of the family social environment that undermine the health of children and adolescents: 1) a social climate that is conflictual and angry, or even violent and abusive; 2) parent-child relationships that are unresponsive and lacking in cohesiveness, warmth, and emotional support; and 3) parenting style that is either overly controlling and dominating or involved with imposition of rules and structure. Such characteristics were linked to poorer mental and physical health outcomes. With respect to older ages, studies have also linked both positive and negative aspects of social interaction to patterns of physiological arousal, with positive patterns of interaction generally being found to predict lower levels of physiological arousal and negative types of interaction being related to increased cardiovascular and/or neuroendocrine activity (10, 11; see Refs. 12 and 13 for reviews).
With the growing evidence linking aspects of the social environment to health outcomes, researchers have begun to focus on the question of possible biological pathways through which aspects of the social environment may affect health. To date, research has largely focused on understanding links between social environment characteristics and individual biological parameters, providing initial evidence linking both aspects of social interaction to patterns of physiological arousal. Children exposed to more negative social environments (eg, less nurturance, more conflict) are more likely to exhibit dysregulated cortisol activity and show greater cardiovascular and sympathetic nervous system (SNS) reactivity to challenge (see Refs. 9 and 14 for reviews). Studies have also linked both positive and negative aspects of social interaction to patterns of physiological arousal in adults, with positive patterns of interaction predicting lower levels of physiological arousal and negative interactions being linked to increased cardiovascular and/or neuroendocrine activity (see Refs. 12 and 13 for reviews). Presence of a supportive friend or confederate, for example, has been shown to result in lower cardiovascular reactivity to challenge (15–17). By contrast, marital conflict has been linked to high blood pressure (BP) (18), elevated pituitary and adrenal hormones (19), and physiological arousal (20). Though research in adults has not focused only on older adults, where the focus has included specific attention to older ages, results have been consistent with those seen in younger adults (12, 13). Comprehensive reviews of the literature relating interpersonal relationships and the overall social environment to physiology and health have been provided by Seeman and McEwen (12), Uchino et al. (13), and Ryff and Singer (21).
There are two features to this previous research that are of special note in the context of this article. One, as noted above and elaborated below, is that research on physiological parameters has tended to focus on individual biological parameters. The second point is that much previous research on the social environment has focused on single point-in-time assessments of relationships (eg, in childhood, adulthood, or later life). Although useful, such investigations do not provide insight into long-term profiles across multiple significant relationships and thereby do not permit evaluation of the health risks associated with cumulative relational adversity. Some who experience conflictual relationships in adulthood, for example, may have also experienced relational problems with parents in childhood. On the positive side, there may also be continuity between nurturing and supportive relationships in childhood and emotional intimacy with a spouse or partner in adulthood. Relational experience, positive or negative, may thus cumulate over time, and differences in such cumulative experience can be hypothesized to affect health risks across multiple physiological systems.
A focus on individual biological parameters does not address the possibility that physiological risks cumulate across systems and across time, resulting in differential health risks within populations. The concept of allostatic load (AL), introduced by McEwen and Stellar (22), reflects this more cumulative view of physiological risk. The basic proposition is that wear and tear across multiple physiological systems is a significant contributor to overall health risk. Such wear and tear is hypothesized to ensue, at least in part, from repeated exposure to life challenges, which may include social relational conflict or adversity. Further conceptual elaboration of AL and its relationship to life experiences has been provided by McEwen (23) and McEwen and Seeman (24).
An initial operationalization of AL—using measures of hypothalamic-pituitary-adrenal (HPA) axis, SNS, and cardiovascular activity; metabolism and adipose tissue deposition; and glucose metabolism—was developed by Seeman et al. (25). Higher AL was shown to predict later life mortality, incident cardiovascular disease, and decline in cognitive and physical functioning (25, 26). Others using the same operationalization have linked higher AL to exposure to cumulative economic and social relational adversity from childhood through age 59 in a 40-year longitudinal study (27). Furthermore, a slightly modified version of the original operationalization of AL has been associated with lower levels of education and greater hostility (28).
The idea of cumulative biological risk is similar to other multifactorial approaches to modeling risk for various health outcomes. Both emphasize the importance of considering multiple factors in estimating overall risks. The primary difference between AL and more traditional multifactorial approaches lies in the emphasis of AL on cumulative risk across multiple regulatory systems and across time. Traditional multifactorial models of cardiovascular risk, for example, tend to examine the independent contributions from multiple individual risk factors (BP, cholesterol, glucose, etc.), asking whether each of these various factors contributes significantly to risk for the outcome. By contrast, the concept of AL focuses on the cumulative, overall risk that results from the combined effects of multiple factors across multiple physiological regulatory systems. Indeed, previous analyses from the MacArthur Studies of Successful Aging (25) have shown that although the overall summary measure of AL significantly predicts risk for major health outcomes, including mortality, none of the individual components is a significant independent "risk factor." Thus, it is the cumulative impact of a set of more modest individual dysregulations across multiple physiological systems, rather than individual dysregulations, that is most consequential in terms of health risks.
Importantly, wear and tear across multiple physiological systems is consistent with evidence that many people, particularly at later ages, suffer from multiple, co-occurring chronic conditions (ie, multiple pathophysiologies). Forty-five percent of women and 35% of men age 60 to 69 report two or more chronic conditions; these figures rise to 61% of women and 47% of men age 70 to 79 and 70% of women and 53% of men age 80 to 89 (29). Though research on mental health outcomes (eg, psychiatric diagnoses) includes some attention to comorbidity (for review see Ref.(30), the prevailing tendency is to focus on a single system of risk (eg, cardiovascular) and single outcomes (eg, cardiovascular disease). In addition, it is important to emphasize that no single form of comorbidity occurs with high frequency; rather, a multiplicity of diverse combinations are seen (eg, osteoarthritis and diabetes, colon cancer and coronary heart disease, or depression and hypertension) (29). This diversity highlights the need for early warning systems of biomarkers that provide evidence of initial dysregulations across a multiplicity of physiological systems. The concept of AL has been proposed as one warning system.
Following on our own prior research on AL (25–27), the goal of the present study was to advance our understanding of life-course linkages between social experiences and biological risk profiles by drawing on longitudinal data from two cohorts, one situated in midlife and the other in old age. Each study includes the same measures of cumulative biological risk (ie, AL) but provides distinct assessments of social experiences, thereby offering different pieces regarding the larger mosaic of cumulative psychosocial and biological risk. The younger midlife cohort, from the Wisconsin Longitudinal Study (WLS), allows longitudinal, life-history assessment of cumulative social experiences (positive and negative). The older cohort, from the MacArthur Studies of Successful Aging (MAC), allows comparative, point-in-time assessments of structural (social integration) and qualitative (social support) aspects of the social environment. The examination of both cohorts provides an opportunity to test the replicative consistency of observed associations between social experience and AL across two independent samples. In addition, the 10- to 20-year age difference between the two cohorts provides the opportunity to compare patterns at different periods in the life course, with the prediction that a higher proportion of the older cohort will show high AL as a result of longer accumulation of physiological wear and tear. In sum, we sought to advance our understanding of life-course links between social experience and biological risk profiles by drawing on two data sets that include comparable measures of biological risk (ie, AL) along with distinct assessments of the social environment.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
Subjects
WLS cohort.
The younger cohort represents a subsample (N = 106) of participants from the WLS (N = 10,317). The WLS is a random sample of men and women who graduated from Wisconsin high schools in 1957 and were reinterviewed in 1975 and 1992–1993 when they were age 58 to 59 (for an overview of the WLS, see Refs. 31 and 32). The WLS biological subsample has a wide range of biomarker assessments and survey responses about social relationships (not available in the larger WLS sample). This sample was selected to match the income distributions of respondents’ family household income in 1957 and own household income in 1992–1993 for the full WLS population. In 1997, each person completed a social relationship questionnaire, underwent a physical health examination, and contributed blood and urine samples. In the WLS cohort, AL data were obtained during an overnight visit to the General Clinical Research Center on the University of Wisconsin–Madison campus. Nursing staff obtained blood samples and other measures needed for AL assessments as well as the 12-hour urine samples (collected overnight from 8 PM to 8 AM).
MAC cohort.
The older cohort is from the MacArthur Studies of Successful Aging. MAC subjects were subsampled on the basis of age and both physical and cognitive functioning from three community-based cohorts in Durham, North Carolina, East Boston, Massachusetts, and New Haven, Connecticut, that were part of the Established Populations for Epidemiological Studies of the Elderly. Men and women (N = 4030) were screened on the basis of four criteria of physical functioning and two criteria of cognitive functioning to identify those functioning in the top third of their age group. Of the 4030 age-eligible men and women, a cohort of 1313 subjects met all screening criteria; 1189 (90.6%) agreed to participate and provided informed consent. Baseline data collection was completed between May 1988 and December 1989, when subjects were ages 70 to 79. This included a 90-minute, face-to-face interview covering detailed assessments of physical and cognitive performance, health status, social and psychological characteristics, as well as other lifestyle characteristics. Subjects were also asked to provide blood samples and 12-hour overnight urine samples; 80.3% agreed to provide blood samples, and 85.8% consented to provide urine samples.
Biological Measures
Allostatic load.
As described in previous work (25, 27), our measure of AL is designed to summarize levels of physiological activity across multiple regulatory systems pertinent to disease risks (22). Parameters included systolic and diastolic BP (indices of cardiovascular activity), waist/hip ratio (an index of metabolism and adipose tissue deposition) (33), serum high-density lipoprotein (HDL) and total cholesterol (markers known to influence the development of atherosclerosis), blood plasma levels of glycosylated hemoglobin (HbA1c; an integrated measure of glucose metabolism over several days’ time), serum dehydroepiandrosterone sulfate (DHEA-S; a functional HPA axis antagonist) (34), urinary cortisol excretion (an integrated measure of 12-hour HPA axis activity); and urinary norepinephrine and epinephrine excretion levels (integrated indices of 12-hour SNS activity). Protocols described below for the older cohort were replicated for the WLS cohort with the exception that all assays were performed at the University of Wisconsin–Madison General Clinical Research Center rather than Nichols Laboratory.
Systolic and diastolic BP were calculated as the average of the second and third of three seated BP readings (35). Waist/hip ratio was calculated on the basis of waist circumference (measured at its narrowest point between the ribs and iliac crest) and hip circumference (measured at the maximal point of the buttocks) (36). Blood samples for assays of HDL cholesterol, total cholesterol, glycosylated hemoglobin, and DHEA-S were obtained by phlebotomists who visited subjects’ homes the morning after their home interview to collect the blood sample and an overnight urine sample (see below). Although subjects were not required to fast, most blood samples were taken early in the morning before subjects had eaten. Sera and heparinized blood were sent to Nichols Laboratory for measurements of HDL and total cholesterol, DHEA-S, and glycosylated hemoglobin (HbA1c).
Subjects completed an overnight urine collection from 8 PM on the evening after their home interview to 8 AM the next morning. Individual differences in overnight excretion of cortisol as well as norepinephrine and epinephrine may well reflect differences in "steady-state" operating levels of the HPA axis and SNS (ie, estimates of "basal," nonstimulated levels of activity) because subjects generally spent this time at home (and much of that time in bed). Urine samples were sent to Nichols Institute immediately after collection for assays of norepinephrine, epinephrine, and cortisol. Determinations were made by high-pressure liquid chromatography (37, 38). Results for each of the three outcomes are reported as micrograms (of norepinephrine, epinephrine, or cortisol) per gram of creatinine to adjust for body size.
Operationalization of AL was based on the algorithm initially developed by Seeman et al. (25). For each of the 10 biological parameters outlined above, subjects were classified into quartiles based on the distribution of scores in the older cohort. AL was measured by summing the number of parameters for which the subject fell into the highest-risk quartile (ie, top quartile for all parameters except HDL cholesterol and DHEA-S, for which membership in the lowest quartile corresponds to highest risk). Table 1 presents the actual criterion cutoff points used for each component of AL. Use of the "top/bottom quartile" criterion to define contributions to higher AL from these various biological parameters reflects a data-driven partitioning of the sample. Membership in the upper or lower quartile represents a quantitative way of classifying those with more extreme levels of system activity relative to the rest of the sample. Alternative algorithms for summarizing these data (eg, alternative cutoff points, averaging Z scores) have been examined and yield comparable findings with respect to health risks (26).
|
Analyses reported here examined AL in two ways. The first used the full range of AL scores. The second focused on relationships between social factors and risks for high AL. Classification of "high" AL for the younger cohort was based on scores of 3 or greater (27). Because the older cohort had more individuals with higher scores, a cutoff point of 5 or more was used to define high AL.
Social Relationship Measures
WLS relationship measures.
In the younger cohort, caring, supportive, and affectionate relationships between parents and children were hypothesized to be important components of cumulative health advantages. Conversely, the experience of uncaring and even abusive interactions with one or both parents was a defining feature of a negative social relationship pathway predicted to be associated with physiological risk. Twelve "caring" items in the Parental Bonding Scale (39) were used to discriminate between those who experienced genuine caring and warmth in the parent-child bond, as opposed to indifference and rejection. In addition, four aspects of connection to a spouse or significant other (ie, midlife social experience) were hypothesized to contribute to cumulative relationship pathways contributing to later life indicators of physiological wear and tear. Different aspects of intimacy were assessed from four subscales of the Personal Assessment of Intimacy Relationships (PAIR) Inventory (40). The emotional and sexual subscales were included because of their focus on the most intimate forms of connection between two people. The intellectual and recreational subscales emphasize mutually enjoyed experience, companionship, and the scope of shared communication. Details about these scales are contained in Singer and Ryff (27).
A summary measure of "relationship pathways" was created on the basis of data from both the parental bonding measure and the PAIR data. WLS cohort members were defined as positive (+) or negative (-) on the mother caring or father caring component of the Parental Bonding Scale if they had a score above or below the median, respectively. Thus, there are four categories of parent-child relationships based on both mother caring and father caring, designated + +, + -, - +, and - -. From the PAIR Inventory, emotional (E) and sexual (S) items, probing the most personal and intimate aspects of spousal ties, were combined by summing scores on these subscales. Intellectual (I) and recreational (R) items, probing more companionate and cognitive forms of spousal connection, were also combined. Individuals were classified as positive (+) or negative (-) on each of the E+S and I+R measures on the basis of whether their scores were above or below the median on the respective subscales. The resulting combinations from the two-way classification of these two measures (E+S category, I+R category) were designated as + +, + -, - +, and - -.
Putting early parental ties and adult spousal ties together, we defined an individual to be on the negative pathway if she or he scored - - on the parental bonding classification and/or - - on the concatenated subscales of the PAIR Inventory. These individuals thus experienced negative relationships with both parents and/or negative interaction with a spouse on both combined aspects of intimacy described above. We defined an individual to be on a positive pathway if he or she had at least one + on the mother caring or father caring categories (ie, + +, + -, or - +) and at least one + on the E+S or I+R categories. Thus, the positive path required some positive relational experience with one or both parents in childhood and at least one of the two combined forms of intimacy in adulthood. This pathway underscores the cumulative nature of positive emotional experiences with significant others in childhood and adulthood.
MAC measures of social integration and support.
The baseline survey included assessments of respondents’ perceptions of their social network, including the presence of particular types of ties, as well as the extent to which those ties provide social support and/or are sources of demands or criticism for the respondent. Social integration was measured by summing the number of reported ties with children, close relatives, and close friends; marital status was considered separately. The summary measure of social integration was examined both as a continuous measure as well as a categorical measure, with the latter providing assessment of possible nonlinear threshold effects. An initial categorical analysis examined a three-level variable to compare isolates with those who had a few and those who had more social ties (ie, 0, 1–2, 3 or more). Because there were only a small number with 0 ties (N = 8), we also examined analyses using a bottom category that combined those with few or no ties (ie, 0–2). Because these latter analyses showed comparable results, the measure used in the analyses reported here was based on a four-level variable (ie, 0–2, 3–6, 7–11, 12 or more ties). To test for possible nonlinear threshold effects, indicator variables were created; these variables represented each of the top three groups with the bottom group serving as the reference category.
Measures of social support available from the MacArthur survey included assessments of emotional and instrumental support as well as a measure of negative interactions reflecting criticism and/or excessive demands from others. Emotional support was measured on the basis of responses to a set of six items (reflecting pairs of items asked separately with respect to one’s spouse, one’s children, and one’s close friends and relatives). The first three items asked, "How often does/do your [spouse/children/close friends, relatives] make you feel loved and cared for?" The remaining three items asked, "How often is/are your [spouse/children/close friends, relatives] willing to listen when you need to talk about your worries or problems?" A summary measure of emotional support was constructed on the basis of the average reported frequency of such interactions across network ties; responses ranged from 0 (never) to 3 (frequently). A parallel measure of instrumental support was based on averaged responses to a second set of six items. The first three items asked, "How often does/do your [spouse/children/friends, relatives] help with daily tasks like shopping, giving you a ride, or helping you with household tasks?" The remaining three items asked, "How often does/do your [spouse/children/friends, relatives] give you advice or information about medical, financial, or family problems?" Responses ranged from 0 (never) to 3 (frequently). To assess negative aspects of social relationships, subjects were asked a parallel set of items regarding how often their spouse, children, and friends and relatives "make too many demands on you" or "are critical of what you do." A summary measure of demands/criticism was constructed on the basis of the average reported frequency of such interactions across network ties; responses ranged from 0 (never) to 3 (frequently).
Statistical Analysis
Linear regression and analysis of variance (ANOVA) techniques were used to examine mean differences in overall AL; cross-tabular and logistic regression analyses were used to examine associations with the dichotomous measure of high vs. low AL. The first set of analyses focus on comparisons of AL scores between the younger and older cohorts. The second set of analyses examine hypothesized relationships between measures of social relationship experiences (relationship pathways for the younger WLS cohort; social integration and support for the older MAC cohort) and AL. Because of the differences in measures of social experiences for the two cohorts, separate analyses are presented for each cohort to assess consistency in findings across diverse relational indicators. Also, because previous findings suggest possible gender differences in association of the social environment with biological parameters (see Ref. 12 for review), these latter analyses also probed possible gender differences in the patterns of association. SAS version 6.12 was used for all analyses (41).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
|
|
Gender Differences in Social Experience and Allostatic Load
In each cohort, men tended to have higher AL load scores than women, although the differences were not significantly different (MAC: mean = 2.68 and SD = 1.58 for men, mean = 2.47 and SD = 1.46 for women; WLS: mean = 2.61 and SD = 1.46 for men, mean = 2.04 and SD = 1.17 for women). Specific components of AL that most frequently contributed to AL scores for men and women revealed a consistent pattern across the two cohorts. In each case, the cardiovascular components (eg, BP, cholesterol, waist/hip ratio) contributed most frequently for men, whereas the neuroendocrine components (urinary cortisol, catecholamines) contributed most frequently for women (data not shown). Examination of gender-specific patterns of co-occurring pairs (and triplets) of biological dysregulation revealed patterns similar to what had been seen for the cohorts as a whole, namely multiple patterns occurring with relatively comparable frequency. Thus, for both men and women in the two cohorts, there seem to be a variety of pathways to higher AL.
Comparison of gender-specific distributions for the social experience measures revealed that men in the younger WLS cohort tended to score higher on each of the parental and spouse relationship measures (see Table 3), though men and women were classified into the positive pathway with equal frequency (47%). In the older MAC cohort, there were no major differences in overall levels of reported social integration (ie, number of ties) or levels of emotional, instrumental, or negative support (see Table 3).
|
WLS relationship pathways.
A first set of analyses of variance examined mean differences in AL across the four specific relational measures that contributed to the cumulative measure of relationship pathways. As shown in Table 4, the general pattern of means suggests that those low on each of the social relational measures tend to have higher mean AL, though few of these relationships achieve statistical significance. Only "intellectual/recreational" interactions were associated with significant AL differences for women, whereas low vs. high "mother caring" was marginally significant for men. However, when the separate childhood and adult social experience measures are used to create a classification that reflects the cumulative relationship experiences, individuals classified as having experienced more positive relationship pathways (ie, more positive parental and spousal relationship experiences) had lower mean AL scores. This relationship was significant for women (p < .05) and marginally significant for men (p < .08). As also shown in Table 4, logistic regression analyses focusing on risks for high AL (ie, scores of 3 or more) revealed that among men, those with positive relationship pathways were only 25% as likely to have high AL scores, compared with men with negative relationship pathways (p < .05; see Table 4). Analyses among women revealed a similarly significant protective effect of having experienced a positive vs. negative relationship pathway, with women on the positive pathways being only 22% as likely to have a high AL score (p < .01; see Table 4).
|
|
Logistic regression analyses were examined next to assess relationships between these same social integration and support measures and risk of a high AL score (ie, score of 5+ = top 10% of AL scores). As shown in Table 5, among both men and women, those with higher levels of social integration were significantly less likely to have AL scores of 5 or more. For men, those in the top level of social integration (ie, 12+ ties) were only 25% as likely to have AL scores of 5 or more as compared with men reporting 0 to 2 ties (p < .05). Among women, those reporting few or no ties were significantly more likely to have AL scores of 5 or more as compared with each higher level of reported social integration (all odds ratios [ORs] 
0.20 comparing higher social integration to "few or no ties" group, p < .05; see Table 5). Again, separate analyses of marital status did not reveal significant associations for either men or women.
Analyses of the support measures examined the relative odds of exhibiting a high AL score for those who reported receiving the specified type of support "frequently" vs. those reporting that such support was received only "sometimes," "rarely," or "never." As shown in Table 5, men reporting more frequent receipt of emotional support were only 44% as likely to have high AL scores (p < .05) as men reporting infrequent or no emotional support; this relationship was not significant for women. Neither instrumental support nor criticism/demands from others were found to predict high AL scores. However, as in the earlier linear regression analyses, there was some indication that negative interactions with family (spouse and/or children) were associated with increased likelihood of a high AL score. For both men and women, those reporting frequent criticism/demands from their children were found to be more than three times as likely to have a high AL score as those reporting less frequent negative interactions with children (OR = 3.24, p < .10 for men; OR = 3.69, p < .10 for women). For men, frequent criticism/demands from one’s spouse was also found to associated with significantly increased likelihood of having a high AL score (OR = 5.04, p < .01). Though the observed gender differences are suggestive, it should be noted that formal statistical tests for gender interactions with social integration and support measures again revealed no significant differences. This is clearly a potentially important area for further research.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
Comparison of AL across the two cohorts revealed several strikingly consistent findings. First, in both there seemed to be multiple pathways of relatively comparable frequency through which individuals achieved higher AL scores. This is true both for pairs of co-occurring elevations in biological parameters as well as for triplets of co-occurring elevations. This finding is consistent with the idea that a cumulative measure of biological dysregulation that includes assessments across a range of regulatory systems may offer advantages in predicting health risks because there seem to be multiple diverse pathways to higher risk. Thus, a more comprehensive assessment of biological risks across multiple systems may provide a more accurate picture of health risk than assessments based on single parameters or even multiple parameters within individual physiological systems. The fact that we are able to demonstrate comparable patterns of findings across two independent cohorts provides further support for, and confidence in, the overall set of findings.
The second area of noteworthy commonality across the cohorts was the consistency of findings with respect to gender. In both cohorts, men tended to have higher AL scores. Men and women were also seen to obtain higher scores through somewhat different patterns of biological dysregulation. In both cohorts, men tended to exhibit dysregulation in the cardiovascular parameters (eg, BP, cholesterol), whereas women were more likely to exhibit high levels of the neuroendocrine parameters (eg, urinary cortisol, catecholamines). For men and women, there were multiple profiles of co-occurring dysregulation, indicating multiple pathways to higher AL with contributions to profiles of risk from multiple systems.
Analyses presented here also provide consistent support, from each of the two cohorts, for our key hypothesis—that different social relationship experiences would be associated with variations in cumulative physiological dysregulation (ie, AL). The degree of consistency is all the more striking in light of the fact that the measures of social experience available for the two cohorts were not identical. For the younger WLS cohort, a summary measure of relationship pathways, representing a cumulative index of both early experiences with parental caring and later life experience with one’s spouse or significant other, was found to be related to levels of AL. For both men and women, those on the more positive relationship pathway were significantly less likely to exhibit high AL scores. These cumulative relationship profiles gave particular emphasis to emotional ties with significant others, dimensions that may be particularly consequential for biological sequelae (21).
For the older MAC cohort, similar protective effects of positive relationship experiences were obtained, though the pattern of association was stronger for men. In this older cohort, men reporting greater social integration were found to be significantly less likely to exhibit high AL. Examination of qualitative aspects of these social ties indicated that men reporting greater frequency of emotional support from their social network were significantly less likely to exhibit high AL. Neither of these associations was significant for women, though tests for gender interactions did not reveal any significant differences. One possible reason for the generally weaker pattern of associations among the women is suggested by research indicating that women may have greater physiological reactivity to negative aspects of social relationships (see Refs. 12 and 42 for discussion of these data). If so, such negative aspects of social relationships may more strongly counteract any positive physiological benefits of either greater social integration and/or greater reported emotional support among women. Interestingly, in this older cohort, marital status did not differentiate those with higher and lower AL scores, a finding consistent with earlier results indicating that marital status is not associated with significant differences in mortality risks at older ages (43). The MAC data also provide some initial evidence regarding possible associations between negative aspects of social experience and biology, effects evident for both men and women. Although the overall summary measure of frequency of criticism and/or demands from others was not related to differences in AL, two specific sources of such criticism/demands did have greater impact. Specifically, more frequent criticism/demands from a spouse or children was associated with higher AL.
In considering these findings, several limitations of the data should be acknowledged. First, the social relationship assessments available from the two cohort studies, though comparable at a general level, were not the same. In addition to the actual item differences, there was an important difference in the scope of the measures. Relationship data for the younger cohort covered not only their current experience but their childhood as well. By contrast, data for the older cohort were reports about only their current social experience. A second limitation is that the older cohort of persons 70 to 79 was selected to represent only the top third of that age group, having been subsampled from a larger cohort of older adults based on their relatively high scores on measures of cognitive and physical functioning. As noted earlier, this selection may have limited our ability to observe age-related differences in AL. In addition, the available data limit our ability to generalize our findings to all older adults in this age range. A third limitation is the fact that both cohorts were predominantly white. Thus, the question of whether these findings apply to other ethnic groups remains to be addressed. A final limitation is that data on AL were available at only one time point for each cohort and thus do not allow direct tests of the hypothesis that social experiences influence AL over time. Rather, the available data provide a snapshot of current levels of AL and their relationship to reported prior short-term and long-term social experiences. Current research from our group is focused on developing longitudinal data on AL and social experience to permit a more thorough test of the hypothesized direction of effects.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
These findings point to biological pathways through which social environment characteristics influence health risks. Our measure of AL can be viewed as an intermediate index of cumulative, biological "wear and tear," a measure that has previously been shown to predict increased morbidity and mortality in those with higher scores (25, 26). Use of this cumulative measure of biological risk may be particularly valuable when examining the health effects of more complex factors like social relationships. Such relationships likely affect a range of biological systems as cognitive and emotional qualities of social experiences are translated by the brain to downstream patterns of physiological activity. Evidence from our cohorts suggests that future research on the role of social experiences in health and aging should include more comprehensive and cumulative approaches to assessing biological risks across multiple regulatory systems. Such a pattern of more generalized biological effects is consistent with previous research indicating that social relationships have effects on a variety of health outcomes (1, 3). Also, our ability to demonstrate similar patterns of association across different age groups is consistent with previous research indicating effects of social experience on physiological regulation in younger adults and children (9, 12–14). The current findings highlight the fact that social environment effects on physiology are evident throughout the life course and may thereby represent a pathway for social environment effects on health and aging.
These findings suggest several valuable avenues for future research on aging and the life course. First, as indicated above, the WLS data point to the potential value of a life-course, cumulative view of social experience in predicting both physiological risk profiles at later ages and, perhaps most importantly, for predicting ultimate health risks. Second, the data on AL from both the WLS and MAC cohorts suggest the value of research that takes a more comprehensive and cumulative approach to the assessment of health risks across multiple physiological regulatory systems.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication August 7, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S.L. Szanton, J.K. Allen, C.L. Seplaki, K. Bandeen-Roche, and L.P. Fried Allostatic Load and Frailty in the Women's Health and Aging Studies Biol Res Nurs, January 1, 2009; 10(3): 248 - 256. [Abstract] [PDF]
|
|
|
|
 |
|
 J. D. Kasper, M. E. Ensminger, K. M. Green, K. E. Fothergill, H.-S. Juon, J. Robertson, and R. J. Thorpe Effects of Poverty and Family Stress Over Three Decades on the Functional Status of Older African American Women J. Gerontol. B. Psychol. Sci. Soc. Sci., July 1, 2008; 63(4): S201 - S210. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. W. Evans, E. Wethington, M. Coleman, M. Worms, and E. A. Frongillo Income Health Inequalities Among Older Persons: The Mediating Role of Multiple Risk Exposures J Aging Health, February 1, 2008; 20(1): 107 - 125. [Abstract] [PDF]
|
|
|
|
 |
|
 D. A. Glei, N. Goldman, Y.-L. Chuang, and M. Weinstein Do Chronic Stressors Lead to Physiological Dysregulation? Testing the Theory of Allostatic Load Psychosom Med, October 1, 2007; 69(8): 769 - 776. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. S. McEwen Physiology and Neurobiology of Stress and Adaptation: Central Role of the Brain Physiol Rev, July 1, 2007; 87(3): 873 - 904. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. M. Crimmins, J. K. Kim, D. E. Alley, A. Karlamangla, and T. Seeman Hispanic Paradox in Biological Risk Profiles Am J Public Health, July 1, 2007; 97(7): 1305 - 1310. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Maselko, L. Kubzansky, I. Kawachi, T. Seeman, and L. Berkman Religious Service Attendance and Allostatic Load Among High-Functioning Elderly Psychosom Med, June 1, 2007; 69(5): 464 - 472. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Levy-Storms and J. E. Lubben Network composition and health behaviors among older Samoan women. J Aging Health, December 1, 2006; 18(6): 814 - 836. [Abstract] [PDF]
|
|
|
|
 |
|
 C. F. Garfield, E. Clark-Kauffman, and M. M. Davis Fatherhood as a Component of Men's Health. JAMA, November 15, 2006; 296(19): 2365 - 2368. [Full Text] [PDF]
|
|
|
|
 |
|
 K. Avlund, T. Rantanen, and M. Schroll Tiredness and Subsequent Disability in Older Adults: The Role of Walking Limitations J. Gerontol. A Biol. Sci. Med. Sci., November 1, 2006; 61(11): 1201 - 1205. [Abstract] [Full Text] [PDF]
|
|
|
|
