Social relations and depressive symptoms in older adults with knee osteoarthritis

Abstract

Depressive symptoms often occur as a comorbid condition in the context of chronic illnesses such as arthritis. However, the role of both social support and social strain in relation to depressive symptoms has not been adequately explored. This study investigates the association of support and strain with depressive symptoms among a sample of older men and women in the USA (N=298, mean age 71 years) who have knee osteoarthritis (OA). Data were collected from a survey mailed to residents who had previously participated in the Osteoarthritis Study in Seniors (OASIS), a longitudinal observational study of OA progression (survey response rate was 77%). Hierarchical regression analyses revealed significant associations of both support and strain with depressive symptoms, while statistically controlling for a variety of demographic, psychosocial and disease-related variables. In addition, social support significantly buffered the relation between social strain and depressive symptoms. The interaction effect was not significantly different for women and men, nor were the individual associations of support and strain with depressive symptoms conditioned by participant gender. The results add to the ongoing discussion regarding gender and social relations as well as highlight the role of both positive (social support) and negative (social strain) aspects of social interactions in relation to the psychological functioning of older adults coping with a chronic illness.

Introduction

Chronic illness has replaced infectious disease as the biggest public health burden in the US over the past century. Adapting to the multiple effects of long-term illness can challenge even strong coping skills, and increases the risk of depression, helplessness, and other negative health outcomes (Revenson, Schiaffino, Majerovitz, & Gibofsky, 1991). For example, Penninx et al. (1997) reported that individuals with arthritis (either rheumatoid or osteoarthritis) reported elevated depressive symptoms when compared to those without arthritis, and the level of depressive symptoms was higher for individuals with more severe arthritis.

The considerable diversity in how individuals adapt to the challenges of chronic illnesses is not fully explained by medical factors (i.e., disease severity, length of illness). This variation has prompted investigators to consider the role of social relations in promoting positive adaptation to illness or in exacerbating the negative effects of illness (Berkman, 1995). A great deal of both cross-sectional and longitudinal work indicates that higher quality (i.e. more supportive) relationships are related to positive psychological and physical health outcomes (see reviews by Berkman, Glass, Brissette, & Seeman, 2000; Heller & Rook, 2001; Seeman, 2000).

The mechanisms by which social relations function in relation to well-being appear to vary. Social relationships can influence happiness and self-esteem and provide support and companionship. Anxiety, depression, loneliness, and poor physical health are associated with a lack of high quality social relationships (Berkman et al., 2000; Rowe & Kahn, 1998). Relationship quality may influence health behaviors and adherence to medical recommendations (e.g., medications, diet, exercise) (Daschner, 1986; Doherty, Schrott, Metcalf, & Iasiello-Vailas, 1983; McMahon, Miller, Wikoff, Garrett, & Ringel, 1986; Schafer, McCaul, & Glasgow, 1986).

The connection between social relationships and well-being has also been linked to the stress-reducing role of social support (see Cohen & Wills, 1985, for a review). Social ties, particularly their health-related functions (Berkman, 1995; Berkman et al., 2000) are likely to be especially important in later adulthood because of the often stressful life transitions that can accompany aging (e.g., widowhood, forced relocation, chronic or acute illness). In addition, several theoretical perspectives point to the importance of social factors in adulthood and to the loss of social ties as an additional stressor that may result in accelerated aging (Berkman et al., 2000; George, 1996).

Considerable research efforts have focused on evaluating two models of the relation between stress, social support, and psychological and physical well-being, with mixed results. Cobb (1979) first proposed that the benefit of social support was primarily seen under conditions of stress, when support would serve a protective function by promoting positive coping efforts. This buffering hypothesis predicts an interactive effect of support at different levels of stress. In contrast, the direct effects hypothesis posits that supportive relationships promote well-being even in the absence of stress. Support for both models has been reported in samples of both older and younger adults (see Russell & Cutrona, 1991).

While social relationships are generally assumed to have beneficial effects on health, negative health behavior (e.g. smoking) can also be reinforced by social models. In addition, research has shown that negative aspects of social relations are also related to psychological well-being (Abbey, Abramis, & Caplan, 1985; Burg & Seeman, 1994; Ingersoll-Dayton, Morgan, & Antonucci, 1997; Rook, 1990). These negative aspects have been variously referred to as “social conflict” (Abbey et al., 1985; MaloneBeach & Zarit, 1995), “negative social interactions” (Lakey, Tardiff, & Drew, 1994; Rauktis, Koeske, & Tereshko, 1995), “negative social support” (Revenson, 1990) and “social strain” (Rook, 1992, Walen & Lachman, 2000). Unlike the protective effect of support, however, social strain appears to function as a risk factor for impaired functioning, particularly depression. Social strain may exacerbate the negative effects of other life stressors. Some authors report that strain is more strongly related to depressive symptoms (Manne, Taylor, Dougherty, & Kemeny, 1997; Rook, 1984) and physical health outcomes (Davis & Swan, 1999) than is support. Others have found comparable effects of both support and strain (Schuster, Kessler, & Aseltine, 1990), particularly when the source(s) of strain and support are considered (Walen & Lachman, 2000). In addition, Okun and Keith (1998) report that positive interactions with three different sources of support were more strongly related to depressive symptoms than were negative interactions. Finch, Okun, Pool, and Ruehlman (1999) suggest that measurement choices may inflate the impact of strain (relative to support) on psychological well-being.

In addition to individual direct effects, strain and support can also work interactively in their association with psychological well-being. A review of such studies by Okun and Keith (1998) showed somewhat mixed results, in that about half the studies reviewed indicated no significant interaction of strain and support, while in the rest, social support or positive interactions buffered the negative effect of conflict or social strain. For example, Schuster et al. (1990) reported that the interaction of supportive and negative contact with relatives was associated with depressive symptoms for both women and men. The supportive contact somewhat reduced the distressing effect of the negative contact with relatives. On the other hand, they found no significant interactions for positive and negative contact with spouses or friends. Examining data from the Americans’ Changing Lives (ACL) survey, Okun and Keith report that the impact of negative exchanges on depressive symptoms was buffered by positive interactions, although the patterns of interaction by source of support and strain were somewhat different for older adults compared to younger. They also report that gender did not moderate the effect of any of the interactions. Walen and Lachman (2000) reported that interactions of support and strain were significantly related to psychological functioning, particularly for women's relationships with partners and friends. For example, the negative relation of partner strain to women's life satisfaction and positive mood was buffered by women's friend support. They report no such interaction effect for men. Taken together, these findings suggest that social support may alleviate some of the effect of negative social contact on well-being for healthy adults, but indicate some inconsistent findings regarding whether patterns of support-strain interaction are similar for men and women.

Positive and negative social exchanges may be even more strongly related to psychological well-being in the context of a chronic illness, which represents additional life stress. For example, Zautra, Burleson, Matt, Roth, and Burrows (1994) found that the number of positive interpersonal events was related to lower levels of depression for women with both rheumatoid arthritis (RA) and osteoarthritis (OA) while the number of conflict events was related to higher depression, especially for the women with RA. Zautra et al., did not test the interaction of conflict and positive events in their cross-sectional analyses.

Manne et al. (1997) investigated the positive and negative responses in spousal relationships in the context of treatment for cancer. They found that supportive spouse interactions did not buffer negative spouse responses in relation to psychological distress or well-being for either men or women (mean age 56 years). Rather, supportive and negative spouse behaviors contributed independently to distress and well-being. In addition, gender did not interact with spouse negative interactions. In contrast, supportive spousal responses had different effects for men and women. While supportive behaviors were not related to outcomes for men, more support was related to lower distress and higher well-being for women.

Revenson et al. (1991) reported that “positive support” and “problematic support,” as well as the interaction of the two support types, were all strongly related to depressive symptoms for men and women with newly diagnosed rheumatoid arthritis. Their results are similar to that of Schuster et al. (1990) and Walen and Lachman (2000) in that individuals with RA who experienced high levels of positive support simultaneously with high levels of problematic support had lower levels of depressive symptoms than those with low positive support and high problematic support. The unique effect of the interaction term was substantial, accounting for 9% of the variance in depressive symptoms. Thus, interactive effects of social support and social strain in samples of both healthy individuals and those with RA suggest that concurrent levels of social support can temper the negative impact of strain in much the same way that the original buffering hypothesis holds for the effect of social support on stressful life events.

However, there are some limitations of the Revenson et al. (1991) study. First, their measurement of “positive” and “problematic support” was specific to rheumatic arthritis, and asked about recent behaviors of other people during a recent pain episode. While the illness-specific measurement approach has several advantages, the measures of positive and problematic support are not generalizable to other illness conditions. In particular, the measures may not be applicable to those with osteoarthritis (OA) because adults with OA may not suffer the unpredictable pain flare-ups associated with RA, but rather experience on-going pain in the affected joint (Rejeski et al., 1995; Zautra et al., 1994). For example, Dieppe, Cushnaghan, and Shepstone (1997) reported no significant changes in reported pain over three years in the Bristol “OA500” study, but also noted that there was considerable individual variation in reported pain. It seems likely that coping with a “recent pain episode” is less applicable to a sample of individuals with OA than those experiencing RA. Since OA is the most common form of arthritis (Cicuttini & Spector, 1995) with up to 85% of those over the age of 65 experiencing OA (Brandt, 1995), it seems appropriate to extend the study of support and strain to a sample of individuals with osteoarthritis.

Furthermore, Revenson et al. (1991) did not include important demographic characteristics that may be related to depressive symptoms, such as gender, ethnicity, or marital status. Gender is particularly important to consider when the outcome variable is depressive symptoms because of the well-known gender differences in prevalence and severity in presentation of depression (Culbertson, 1997; Nolen-Hoeksema, 1987) and the reported gender differences in support and strain (Manne et al., 1997; Walen & Lachman, 2000). In addition, Revenson et al. did not control for pain, a highly salient feature of osteoarthritis (McAlindon, Cooper, Lirwan, & Dieppe, 1992; Rejeski et al., 1995), nor did they control for health perceptions or life satisfaction, both of which may also be related to depressive symptoms.

The present paper expands on the Revenson et al. (1991) approach in several areas. For example, more demographic, physical, and psychological health status variables have been included as controls. We also used measures of support and strain that applied to both healthy and chronically ill individuals. In addition, we considered the possibility of gender-specific patterns of strain and support. Further, the sample consists of older adults with knee OA, which is the most common type of arthritis, and the most common site of symptomatic OA (Brandt, 1993; Ling & Bathon, 1998). The following hypotheses were tested:

• Strain and support will each contribute significantly and independently to the variance in depressive symptoms, after accounting for demographic characteristics, disease severity, perceived pain, health perceptions, and life satisfaction.

• The interaction of strain and support will be associated with unique variance in depressive symptoms after all other variables are included in the model.

• The relation of strain and support to depressive symptoms will be different for men and women.