An appraisal-based coping model of attachment and adjustment to arthritis

Introduction

A major source of disability worldwide (Arthritis Society, 2010), arthritis is one ofthe most common chronic pain conditions. Although arthritis can take many different forms, joint pain, inflammation, and functional limitations are common challenges for people living with arthritis. The functional limitations and unpredictable nature of arthritis symptoms can also contribute to stress by making daily planning difficult and creating dependency upon others for routine daily activities (Gignac et al., 2000). Together these disease-related and psychosocial stressors can take a toll on psychological well-being (e.g. Murphy et al., 2012). Given that stress has been implicated in the etiology, maintenance, and exacerbation of rheumatic diseases (Cohen et al., 2012; Everset al., 2013) and that adaptive coping is linked to better adjustment to arthritis over time (Pinto-Gouveia et al., 2013; Sirois and Hirsch, 2013), understanding the factors and processes that impact how people with arthritis perceive and respond to disease-related stressors can have important implications for disease management and adjustment.

How stressors are appraised and the factors that affect these appraisals are key for understanding how people successfully cope with arthritis. The role of appraisals in the perception of and adjustment to stressors is perhaps best explained by Lazarus and Folkman’s (1984) cognitive transaction model of stress. In this model, appraisals play a central role in both the initiation and continuance of the stress response. The presence of a stressor alone is not sufficient to understand its impact; rather how that stressor is appraised initially and in light of perceived coping resources will determine the extent to which it is threatening. In the context of arthritis, primary appraisals of challenging events such as pain or functional limitations are followed by a secondary appraisal of the potential coping resources that may be available to help manage the stressor.

Because functional limitations are a common stressor for people with arthritis, social support is one coping resource that may be particularly important. Low levels of social support have been implicated in long-term functional disability and pain (Evers et al., 2003a), psychological distress shortly after diagnosis (Evers et al., 1997), and longitudinally over several years (Benka et al., 2012; Evers et al., 2003b; Gafvels et al., 2014) in people with arthritis. However, social support may have to be first perceived and then used to facilitate healthy coping and adjustment (Curtis et al., 2004; Kool et al., 2013).

One individual difference that may play an important role in whether social resources are perceived and successfully used by people with arthritis is attachment style. Briefly, attachment was conceptualized by Bowlby (1969) to describe relationship patterns between children and their caregivers that evolve into stable expectations and patterns of relating to significant others in adulthood. In contemporary research, attachment is typically measured by dimensions related to anxiety and avoidance (Brennan et al., 1998; Ravitz et al., 2010). Individuals high on anxiety tend to be dependent on significant others and fear rejection by them, while those high on avoidance are self-reliant and uncomfortable with close relationships. In contrast, the securely attached are low on both anxious and avoidant dimensions and seek out close relationships for comfort during times of stress. From this perspective, secure attachment can be considered an inner resource or resilience factor that can foster successful coping with stress, whereas insecure attachment is a risk factor that may interfere with successful coping (Mikulincer and Florian, 1998).

Central to attachment theory is that the attachment system is activated during the experience of a threat (Bowlby, 1988). This threat, however, can include symptoms of illness or pain. In the Attachment-Diathesis Model of Chronic Pain (ADMoCP; Meredith et al., 2008), appraisals of pain, as well as of the ability to cope with it and the availability of support, influence responses to appraisals and adjustment outcomes. Studies have shown that individuals with chronic pain who were insecurely attached appraised pain as more threatening (Meredith et al., 2005) and described themselves as less capable of coping with pain (Meredith et al., 2008). Mikulincer and Shaver’s (2007) general model of attachment activation and functioning posits that initial threat appraisal is followed by the appraisal of social resources in the form of availability and responsiveness of an attachment figure as a means of coping with the threat. Hyperactivating strategies (e.g. hyper vigilance and rumination over both the threat and proximity seeking) characteristic of anxious attachment may ensue as a means of coping when proximity seeking is possible, but the attachment figure is perceived as unavailable. Deactivating strategies such as distraction and distancing from the threat and from proximity seeking that are common to avoidant attachment may also result. Hyperactivation and deactivation can each lead to poor adjustment over time (Mikulincer and Shaver, 2007) and can influence threat appraisal through feedback loops and reappraisal. According to this model, attachment factors are prominently in the primary and secondary appraisals of stress and therefore can influence perceptions of how well one is coping with illness.

In the context of a painful chronic illness such as arthritis, insecure attachment may be particularly problematic. The heightened stress appraisals and perceptions of unavailable social support resources associated with insecure attachment may foster feelings of being unable to cope successfully with illness-related stressors and contribute to poor adjustment. For example, coping efficacy, the perceived ability to cope with stressors, has been linked to arthritis adjustment outcomes (Katz, 1998; McKnight et al., 2010), and to perceptions of psychological growth over time (Sirois and Hirsch, 2013). Although, attachment theory posits that secure attachment is associated with greater self-efficacy (Mikulincer and Florian, 1998), the inter-relations between insecure attachment, primary and secondary illness-related stress appraisals, and coping efficacy have not been previously examined in the context of arthritis.

This study

Bringing together this theoretical and empirical literature, we propose an appraisal-based coping model of attachment and adjustment to arthritis (see Figure 1). Consistent with coping theory (Lazarus and Folkman, 1984), models of attachment activation and pain-related stress (Meredith et al., 2008; Mikulincer and Shaver, 2007), we posit the following: Insecure attachment will be associated with poor coping efficacy, and this link may be explained by primary appraisals of stressors related to arthritis that include but are not limited to pain and by secondary appraisals of social resources or social support assessed as perceived social support. Coping efficacy was chosen as the indicator of adjustment to arthritis as previous research has demonstrated that perceiving oneself as coping well with illness-related stressors is linked to a variety of other important adjustment outcomes (Katz, 1998; McKnight et al., 2010; Sirois and Hirsch, 2013).

OPEN IN VIEWER

Figure 1.

Proposed appraisal-based attachment model of coping and adjustment to arthritis. Dashed line indicates constrained path to be tested.

Using structural equation modeling (SEM), we tested the proposed model with arthritis coping efficacy as the outcome variable (see Figure 1). Given that theory and research indicate that attachment may have both direct and indirect effects on primary stress appraisals and secondary appraisals of perceived social support (Meredith et al., 2005, 2008; Mikulincer and Shaver, 2007), we included both direct and indirect paths from attachment to appraisals and coping efficacy and tested alternative models to examine the possible mediating effects of appraisals on coping efficacy. We also allowed primary and secondary appraisals to covary to simulate the reappraisal process, with less perceived social support increasing appraisals of arthritis-related stressors and vice versa. To test the possible differential associations of anxious and avoidant attachment with the appraisal and coping variables, we chose to test the model separately for anxious and avoidant attachment.

Method

Measures

Table 1 presents the means and Cronbach’s alphas for all the scales.

OPEN IN VIEWER

Table 1.

Bivariate correlations among the attachment and appraisal model variables.

Variable	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
1. AV attachment	–
2. AV parcel 1	0.91**	–
3. AV parcel 2	0.88**	0.72**	–
4. AV parcel 3	0.72**	0.50**	0.46**	–
5. ANX attachment	0.33**	0.33**	0.25**	0.28**	–
6. ANX parcel 1	0.29**	0.28**	0.19**	0.28**	0.89**	–
7. ANX parcel 2	0.30**	0.30**	0.23**	0.22**	0.89**	0.58**	–
8. Primary appraisals	0.12*	0.11*	0.13**	0.08	0.24**	0.26**	0.15**	–
9. Pain/symptoms	0.07	0.07	0.07	0.05	0.19**	0.20**	0.13*	0.78**	–
10. Physical limitations	0.08	0.07	0.09	0.05	0.13*	0.15**	0.07	0.86**	0.61**	–
11. Dependence on others	0.11*	0.09	0.11*	.07	0.23**	0.26**	0.14**	0.86**	0.49**	0.66**	–
12. Daily impact	0.14**	0.13*	0.15**	.08	0.22**	0.24**	0.15**	0.86**	0.61**	0.65**	0.63**	–
13. Social support	−0.28**	−0.26**	−0.19**	−.29**	−0.46**	−0.38**	−0.44**	−0.21**	−0.17**	−0.13*	−0.19**	−0.20**	–
14. Coping efficacy	−0.26**	−0.22**	−0.23**	−.20**	−0.32**	−0.30**	−0.27**	−0.43**	−0.32**	−0.35**	−0.36**	−0.42**	0.30**	–
15. Functioning	0.02	0.06	0.07	.06	0.00	−0.03	0.01	0.41**	0.33**	0.37**	0.33**	0.37**	−0.03	−0.28**	–
16. Pain severity	0.07	0.02	0.01	.03	0.10	0.11*	0.04	0.56**	0.34**	0.46**	0.56**	0.48**	−0.08	−0.29**	0.48**	–
Mean	3.64	3.50	4.12	3.12	2.95	3.30	2.56	4.37	4.64	4.60	4.07	4.17	3.79	3.28	1.85	3.04
Standard deviation	1.20	1.52	1.30	1.40	1.23	1.34	1.38	1.10	1.08	1.18	1.65	1.33	1.04	1.01	0.52	0.77
Cronbach’s alpha	0.84	–	–	–	0.65	–	–	0.86	–	–	–	–	0.92	0.89	0.94	–

ANX = anxious; AV = avoidant.

*

p < .05; **p < .01; N = 365.

Arthritis pain and functioning

Two questions addressed when participants had received a formal diagnosis of arthritis and the specific type of arthritis. Severity of arthritis pain experienced within the past month was measured with one item from the Arthritis Impact Measurement Scales 2 (Meenan et al., 1992). Participants rated their usual pain severity on a 4-point Likert-type scale ranging from severe to none, and the item was reverse scored so that higher scores reflected greater pain. Day-to-day functioning was assessed with the 20-item functioning subscale which includes statements about different daily activities such as climbing five steps, doing chores, and running errands, rated on a 4-point Likert-type scale ranging from 1 (without any difficulty) to 4 (unable to do). Items are averaged to form a mean score with higher scores reflecting lower levels of functioning.

Attachment style

A 12-item measure of attachment style (Simpson et al., 1992) assessed anxious (four items) and avoidant (eight items) attachment. Higher scores on the avoidant subscale (e.g. “I’m somewhat uncomfortable being too close to others”) reflect greater avoidant attachment, whereas lower scores on this scale reflect more secure attachment. Higher scores on the anxious subscale (e.g. “I often worry that my partner(s) don’t really love me”) reflect greater anxious attachment. Responses are scored on a 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree).

Primary stress appraisals

The appraisal of arthritis-related stressors was assessed with four questions designed for this study that became the indicators for this latent variable. The questions were based on previous empirical work on the key areas of stress for people with arthritis: (1) pain- and symptom-related problems including symptom severity, unpredictability, and symptoms related to medication; (2) physical limitations and functional impairment; (3) dependence on others, including feeling that they are a burden to others; and (4) impact of arthritis on day-to-day life, including social limitations and difficulties planning activities (Gignac et al., 2000; Katz, 1998). Participants rated how stressful each of these four arthritis-specific concerns had been for them over the previous 6 months on a 6-point Likert-type scale ranging from 1 (did not experience/not stressful) to 6 (extremely stressful).

Perceived social support

The Duke-UNC Functional Social Support Questionnaire (Broadhead et al., 1988) assessed perceived social support that served as a proxy secondary stress appraisals. Eight items are rated on a 5-point Likert-type scale with responses ranging from 1 (much less than I would like) to 5 (as much as I would like). Items include both emotional and practical aspects of social support with higher scores reflecting greater perceived social support.

Coping efficacy

The three-item coping efficacy scale developed by Gignac et al. (2000) assessed appraisals of efficacy in coping with the chronic stressors associated with arthritis. The scale focuses on three particular challenges associated with arthritis: symptoms, emotional aspects, and day-to-day problems. Items such as “I am successfully coping with the symptoms of my arthritis” are scored on a 5-point Likert-type scale with responses ranging from 1 (strongly disagree) to 5 (strongly agree) with higher scores reflecting greater coping efficacy.

Results

SEM analyses of the anxious attachment–adjustment model

SEMs were estimated using AMOS 21.0. Several indices were used to assess model fit: (1) chi-square statistic, (2) root mean square error of approximation (RMSEA), (3) comparative fit index (CFI), and (4) incremental fit index (IFI). A model is considered to have a good fit to the data if the CFI and IFI are greater than 0.95, and the RMSEA is 0.05 or less (Schermelleh-Engel et al., 2003). Given the relatively large sample size and the known tendency of the chi-square test to become significant as sample size increases (Kline, 2005), more weight was placed on the results of the fit indices for evaluating model fit.

Latent variables were created for both attachment styles and for primary appraisals using parceling. Anxious attachment items were divided into two parcels, and avoidant/secure attachment items were divided into three parcels. Given that the primary appraisal items each tapped a different domain of potential illness-related stressors and, therefore, reflected distinct but related factors of this construct, each of these four items served as a single-item domain representative parcel (Little et al., 2013) for the latent variable reflecting primary appraisals. This approach was taken following Little et al.’s (2013) recommendations outlining the advantages of parceling for creating indicators with better measurement properties than scale scores, for reducing Type II errors, and for enhancing precision in measuring constructs.

The initial test of the overall model with anxious attachment revealed a good fit to the data, χ²(16, N = 365) = 31.51, p < 0.05; RMSEA = 0.05 (90% confidence interval (CI): 0.05–0.07); CFI = 0.98; and IFI = 0.98. The direct paths from anxious attachment to primary stress appraisals, social support, and coping efficacy were each significant in the expected directions (see Figure 2). However, the paths linking primary stress appraisals to social support, and social support to coping efficacy appraisals, were not significant and were therefore removed from the model. Accordingly, the nested mediation model was tested for significant indirect effects through primary stress appraisals by constraining the path from attachment to coping efficacy to zero following the recommendations of Holmbeck (1997). The fit of the nested model was generally good, although the fit indices changed in a direction reflecting slightly poorer fit, χ²(19, N = 365) = 57.80, p < 0.01; RMSEA = 0.07 (90% CI: 0.05–0.10); CFI = 0.96; and IFI = 0.96. The test comparing the chi-square value for the nested model with indirect effects through primary and secondary stress appraisals, to the full model, found significant difference in the chi-square values, Δχ²(1) = 26.25, p < 0.01 such that the addition of the constrained path was a significant decrement to the model fit. These findings suggested that the unconstrained, direct effects model should be retained and that test of the mediation model was not necessary. The total effect of anxious attachment on coping efficacy for the final model was −0.44, p < 0.001.

OPEN IN VIEWER

Figure 2.

Final appraisal-based models of coping and adjustment to arthritis for anxious and avoidant attachment—standardized estimates (*p < .05; **p < .01; N = 365).

Discussion

To the best of our knowledge, this is the first study to empirically test an appraisal-based model of attachment and coping in people with arthritis based on the general attachment activation and functioning model proposed by Mikulincer and Shaver (2007) and the ADMoCP (Meredith et al., 2008). Similar to other research on insecure attachment in chronic pain populations (Meredith et al., 2008), we found that both anxious and avoidant attachment were associated with appraisals of disease-related threat and less perceived social support to deal with this threat. However, a unique finding of this study was that both forms of insecure attachment were also associated with feeling less successful in being able to cope with the emotional, symptom-related, and day-to-day aspects of arthritis.

Contrary to our hypothesis, the primary appraisals and social support (secondary appraisals) did not explain the link between insecure attachment and coping efficacy. Instead, the direct effects of attachment on coping efficacy were present after considering the role of primary and secondary stress appraisals. This suggests that in the context of arthritis, people who are insecurely attached may perceive difficulty in managing their disease regardless of how threatened they feel by the illness-related stressors they experience or any perceived lack of social support. Similar to general self-efficacy (Bandura, 1977), coping efficacy is likely reinforced by past successes and failures in managing stressors. Thus, previous unsuccessful attempts to manage the demands of living with arthritis associated with insecure attachment could contribute to a general and enduring perception that one is not coping well with its challenges that is independent of appraisals of current stressors.

Consistent with cognitive transactional models of stress and coping (Lazarus and Folkman, 1984), our model includes a dynamic element that allows for reciprocally determined transitions between primary and secondary appraisals of arthritis-related threats and the perceived social resources available to deal with those threats. However, our findings suggest that there may be differences in the way that avoidant and anxious attachment shape this process. Although the test of the model with avoidant attachment supported a dynamic reappraisal process between primary and secondary appraisals similar to the feedback loops between the hyperactivating and deactivating strategies proposed by Mikulincer and Shaver (2007), this was not found when testing the model with anxious attachment. Instead, arthritis-related threat appraisals were unrelated to perceptions of available social support, despite the fact that they were negatively correlated in the bivariate analyses. For avoidant individuals, arthritis-related stressors may lead these individuals to react with heightened emotional distress not only because they feel helpless to deal with this chronic stress but also because they may need to rely on others, given their functional limitations and yet, unlike the anxiously attached, are uncomfortable with having to do so due to their perception of others as being untrustworthy (Mikulincer and Florian, 1998). In contrast, our findings suggest that like the avoidantly attached, anxiously attached individuals also perceive low social support and higher stress associated with arthritis, but these primary and secondary appraisals are unrelated when considered in the context of coping efficacy. Given that anxious attachment was more strongly related to perceived social support than avoidant attachment, it is possible that this result also reflects preoccupation with the unavailability of the attachment figure. Clearly, more work is needed to replicate and probe this intriguing result.

Although unique, the current findings should be considered in the context of certain limitations and strengths. The diagnosis of arthritis was self-reported, and although carefully screened by requesting that participants report any medications they were currently taking to manage their arthritis, it may not have been as reliable as a formal diagnosis from a rheumatologist. Despite this, the use of a large community-based sample of people self-reporting a variety of major rheumatic diseases is a potential strength. It could be argued that because the sample was heterogeneous with respect to arthritis type, the results may not be generalizable to specific forms of arthritis. However, there is some evidence that coping and adjustment-related outcomes do not vary significantly across major forms of arthritis such as rheumatoid and osteoarthritis (Sirois and Hirsch, 2013). Research replicating the proposed models, here, with individual arthritis samples would nonetheless clarify this issue. Although the use of an Internet survey may appear to present certain limitations with respect to generalizability, there is significant evidence that samples obtained from the Internet tend to be larger and more heterogeneous than those recruited by more traditional means, while being of equal quality (Gosling et al., 2004). Another strength is that stress and coping efficacy were not measured generally but specific to the challenges that people with arthritis face. The use of cross-sectional data precludes strong statements regarding the direction of causality among the variables proposed by our model. However, the proposed pathways are based on extant theory and research lending support to our model. Future work focusing on the proposed processes as they occur in daily life or in controlled settings is necessary to confirm the directionality of the relationships.

Our findings have several potential implications for improving the coping of people with arthritis who are insecurely attached. Given the direct link between insecure attachment and coping efficacy found in this study, interventions aimed at increasing coping efficacy may be important for improving adjustment, especially given the associations with lower pain severity and better physical functioning noted in this study, and with lower depression over time found in other research (Sirois and Hirsch, 2013). For example, interventions such as cognitive behavioral therapy (CBT) to enhance perceptions of control over common arthritis-related stressors including pain, dependence on others, physical limitations, and daily functioning may be particularly effective (Halford and Brown, 2009; see also Astin et al. (2002) for a meta-analysis indicating the effectiveness of psychological interventions, including CBT). Encouraging a focus on small coping successes might increase patients’ feelings of efficacy for coping with the daily demands of arthritis. Our findings further suggest that applying aspects of interventions focused on relationships (e.g. interpersonal therapy) or attachment theory (e.g. Johnson et al., 1999) may be an important way to maximize the effectiveness of psychological interventions for arthritis patients with insecure attachment (Sharpe, 2013). As suggested by Meredith et al. (2008), making health care providers aware of the stress and coping appraisals associated with insecure attachment and tailoring treatments accordingly may be an additional way to facilitate translational application of our findings into practice.

Conclusion

Overall, the findings from this study provide new and important insights into how insecure attachment is implicated in the coping appraisals and responses to disease-related stressors in people with arthritis. Although the findings support the notion that both anxious and avoidant attachment can be considered as risk factors for coping with the stressful challenges of living with arthritis, our findings also indicate that there may be differences in how these two forms of insecure attachment contour the coping process. Nonetheless, finding ways to enhance confidence in one’s ability to cope with arthritis-related stress may be an important priority for improving adjustment in individuals who are insecurely attached.