Abstract
Although complicated grief (CG) following the death of a loved one has traditionally been viewed as avoidance-driven, recent research implicates approach behavior. Researchers have investigated how coping styles predict CG; however, emotional closeness remains unexamined. This study investigated the differential relationship of approach- and avoidance-focused coping on CG and depressive symptoms. Bereaved adults (n = 340) completed questionnaires about their loss, coping styles, CG, and depression. Results suggest that approach-, but not avoidance-, focused coping mediates the relationship of emotional closeness on CG, while emotional closeness of the relationship had no direct impact on depressive symptoms. This suggests closeness of the relationship plays a unique role in CG.
Complicated grief (CG) is a maladaptive response to a loss characterized by excessive yearning or longing for the loved one, emotional pain, difficulty accepting the loss, and feelings of being stunned or shocked by the loss (Horowitz et al., 2003; Papa et al., 2014; Prigerson et al., 1996; Shear et al., 2011). Although most people who experience a loss are able to adjust and continue to lead a functional life, those with CG experience a variety of functional impairments and see a substantial decrease in their engagement in life (Prigerson et al., 1995a). Despite sharing some symptom overlap, CG is distinct from bereavement-related depression based on symptom presentation (Bonanno et al., 2007; Papa et al., 2014; Prigerson et al., 1995a,1995b). A number of studies examining the distinction between the two conditions have utilized statistical methods such as exploratory and confirmatory factor analysis (Boelen & van den Bout, 2005; Papa et al., 2014; Prigerson et al., 1995a,1995b), latent class analysis (Djelantik et al., 2019), and growth modeling (Lenferink et al., 2018); however, minimal research examines associated processes and behavioral tendencies (e.g., coping styles and appraisal styles) that may be associated with one condition rather than the other (Eisma et al., 2013).
Attachment theory informs traditional views of CG (Bowlby, 1982), proposing that after the loss of an attachment figure, an individual avoids reminders of the deceased individual, leading to a restricted life and the development of CG (Eisma et al., 2013; Shear et al., 2007). Although this attachment theory of CG, based on an avoidance-driven model, has gained some prominence (Archer, 2008), other researchers suggest that avoidance alone may not explain CG (Maccallum & Bryant, 2013, 2019; O’Connor & Arizmendi, 2015; Stroebe et al., 2010). For example, people experiencing CG were more likely to pull loss reminders toward them rather than push them away in a computer-based experimental paradigm (Maccallum et al., 2015). Such a finding suggests that, beyond avoidance-driven phenomena, some approach responses may be implicated. Further findings from Maccallum and Bryant (2019) illustrate that individuals who meet criteria for CG were likely to pull closer names of the deceased, whereas those without CG did not demonstrate this tendency. In addition, neuroscience shows continued activation of reward pathways in the brains of individuals with CG when viewing images or words related to the deceased (Arizmendi et al., 2016; O’Connor et al., 2008; O’Connor & Arizmendi, 2015). This finding held only in individuals with CG rather than those who had experienced bereavement without CG and may demonstrate an underlying tendency to approach these reminders, suggesting some rewarding or reinforcing aspect, despite the emotional pain they cause (O’Connor & Arizmendi, 2015).
In addition, various coping styles can impact the outcomes in CG (Schnider et al., 2007). Some studies provide evidence supporting an attachment theory of CG, where global avoidance coping predicts CG (Eisma et al., 2013; Schnider et al., 2007); however, selected coping measures in studies of CG speak of coping styles in a more global fashion and are not specific to coping with reminders of the loss, which is the driving mechanism under attachment theory (Shear et al., 2007). Although coping styles relate more broadly to adjustment and maladjustment in the face of adversity (Ebata & Moos, 1991), specific assessment of avoidance-focused coping regarding reminders of the loss could help illuminate the specific relation between avoidance and CG. Given the importance of avoiding reminders of the loss to the attachment theory of CG, a study examining approach and avoidance coping in response to reminders is needed to examine whether findings from behavioral (Maccallum et al., 2015) and clinical neuroscience research (Arizmendi et al., 2016; O’Connor et al., 2008; O’Connor & Arizmendi, 2015) also apply when specific reminders of the loss are confronted.
Complementary to attachment theory, the identity continuity theory of CG (Papa & Lancaster, 2016) implicates additional factors that may help predict CG responses following a loss. Within this theory, disruption to important aspects of an individual’s sense of self could contribute to maladjustment after a loss. Research indicates that such disruption is particularly likely to lead to higher levels of CG when the lost role is highly central to the person’s sense of self, as with the loss of emotionally close individuals (Papa et al., 2014). Although research highlights self and identity factors that may contribute to or buffer against CG symptoms after a loss (e.g., role centrality, self-construal balance, self-complexity; Papa & Lancaster, 2016; Papa et al., 2014), to our knowledge an integration of these factors with coping and avoidance does not exist. Given both the paucity of research and disparate results regarding coping, identity, and CG, synthesizing these areas of the literature could enhance our understanding of how these factors are related.
This study examined the relation between approach- and avoidance-focused coping and CG symptomology. In addition, in integrating the coping and identity continuity literatures, we examined whether these coping styles mediated the relationship between emotional closeness (a proxy variable for role centrality) and CG. We hypothesized (a) a positive relation would exist between approach- and avoidance-focused coping and CG when controlling for relevant covariates, (b) coping style would mediate the relationship between emotional closeness and CG symptomology, and (c) the pattern of relations between coping strategies and levels of depression would differ from that observed with levels of CG.
Method
Procedure
Participants were recruited from a variety of social media outlets via electronic advertisements specifying the rationale, eligibility criteria, compensation information, and link to the online survey for this study. Online bereavement support groups and forums, professional organization listservs, and social networking sites (e.g., Facebook and Reddit) were utilized. After providing electronic informed consent on the first page of the link, participants completed the online survey. The measures included in this study are a subset of those administered during a larger study of factors associated with psychopathology following bereavement. Assessment domains included demographics, coping styles, CG, and depression. Following the survey, participants received national mental health services information and crisis hotline phone numbers. All participants were also entered into a raffle for several Amazon gift cards upon completion of the survey. This study and its materials received institutional review board approval.
Participants
Participants were 340 U. S. adults (mean age = 32.6, SD = 12.5, age range = 18–72) who lost a loved one within 5 years of the study (mean time since loss in months = 28.74, SD = 34.53, range = 0.1–60). We selected this timeframe for the loss to increase the homogeneity of the sample and ensure participants could recall adequate information regarding the loss. Most participants lost someone to illness or disease (58%). They were predominantly White (89%), women (78%), and 60% reported having at least a college degree.
Of those who responded to study advertisements, 458 individuals completed all or part of the questions; however, 118 participants were excluded from the analyses (83 for providing invalid birthdates/zip codes, 23 for failing to report the type of death, and 12 because they lost loved ones more than 5 years ago), leaving a final sample of 340.
Measures
Participants provided information about themselves and their recently deceased loved one. Questions about the deceased included the cause of death and time since loss. As a proxy for role centrality—critical to the identity continuity theory—we assessed emotional closeness via two items (“How emotionally close were you with the person who died?” and “What was the nature of your relationship?”) in which participants responded on scales ranging from 1 (not at all close and never met the person, respectively) to 4 (intimately close and one of the closest relationships in my life, respectively). We used the mean of these two items in study analyses to represent closeness of the relationship.
The Coping Inventory for Stressful Situations Short Form (Endler & Parker, 1994) consists of 21 questions about responses to stressful scenarios rated on a 5-point Likert-type scale (from 1 = not at all to 5 = very much). The Coping Inventory for Stressful Situations Short Form allows for a specified stressful scenario. In this study, the specified stressful scenario was, “Indicate how much you engaged in these types of situations when you thought about your loved one’s death.” This questionnaire includes three subscales that measure task-oriented, emotion-oriented, and avoidance-oriented coping styles (Endler & Parker, 1994). As no hypotheses were made with regard to the construct of task-focused coping given the previous literature on CG, this scale was not utilized in the current analyses. Internal consistency coefficients are adequate for emotion-oriented (α = .78–.87) and avoidance-oriented (α = .70–.80) subscales (Endler & Parker, 1994, 1999; Endler et al., 2000; McWilliams et al., 2003). In this sample, we achieved acceptable to good internal consistency for the emotion-oriented (α = .89) and avoidance-oriented (α = .78) subscales in this sample. For this study, we conceptualized emotion-oriented coping as approach-focused, as the items included coping behaviors and cognitive strategies that involved approaching reminders of the loss and dwelling on the subsequent emotions.
The Inventory of Complicated Grief (Prigerson et al., 1995b) is a 19-item self-report questionnaire that evaluates symptoms of CG. Items are rated on a 5-point scale (from 0 = never to 4 = always). This questionnaire has high internal consistency (α = .94) and high criterion and convergent validity (Prigerson, et al., 1995a,1995b). Internal reliability in this sample was excellent (α = .94).
The Center for Epidemiologic Studies Short Depression Scale (Andresen et al., 1994) is a 10-item shortened self-report questionnaire about depression symptoms. Items are rated on a 4-point scale (from 0 = rarely or none of the time to 3 = all of the time). The Center for Epidemiologic Studies Short Depression Scale has adequate internal consistent (Cronbach’s α = .78; Boey, 1999) and sufficient factorial validity (Cheng et al., 2006). Internal reliability in this sample was good (α = .84).
Data Analytic Strategy
All assumptions regarding normality, homoscedasticity, and no multicollinearity were assessed before conducting study analyses. Missing data were handled with listwise deletion for each analysis. Univariate and multivariate outlier analyses were conducted for each regression, with multivariate outliers being removed due to undue effects on distance and leverage statistics. After initial bivariate correlations were calculated, multiple linear regressions were utilized to examine relationships between coping styles and CG while controlling for relevant covariates. An additional multiple linear regression was conducted, regressing depressive symptoms on these same variables to assess for differential relations.
To further analyze these relationships, we examined a mediation model utilizing the PROCESS macro for SPSS (Hayes, 2017), wherein coping styles mediated the relationship between emotional closeness and CG symptoms. A similar mediation model was conducted with depressive symptoms as the outcome variable for comparison. Both models controlled for age and time since loss at each stage. In both models, we utilized nonparametric bootstrapping of 5,000 subsamples to generate 95% confidence intervals for indirect effects (Hayes, 2017).
Results
Initial bivariate correlations, means, and SDs are presented in Table 1. CG correlated significantly with approach- and avoidance-focused coping, and depression (r values ranging from .34 to .72, p < .01, moderate to strong for all). Emotional closeness correlated significantly with both coping styles and CG (r values ranging from .25 to .41, p < .01 for all). In addition, age was significantly negatively correlated with approach-focused coping (r = –.14, p < .05), and time since loss was significantly negatively correlated with depression (r = –.22, p < .01). As a result, these two variables were included as covariates in all future analyses to allow for model consistency.
Correlations Between Complicated Grief Symptoms, Independent Variables, and Covariates With Means and Standard Deviations.
Note. *p < .05. **p < .01. ***p < .001.
To test the first hypothesis—a positive relation would exist between approach- and avoidance-focused coping and CG when controlling for time since loss and age—CG symptoms were regressed on approach-focused coping and avoidance-focused coping, with emotional closeness, time since loss, and age. No univariate outliers were removed, and three multivariate outliers were removed.
Regression results are displayed in Table 2. Omnibus test of the model whole was significant, F(5,193) = 56.09, p < .001, R2 = .59. Approach-focused coping, b = 1.40, t(200) = 12.76, p < .001, and emotional closeness with the deceased, b = 5.70, t(200) = 5.07, p < .001, were associated with higher levels of CG, though time since loss, b = –.12, t(200) = –2.90, p < .01, was associated with lower levels of CG.
Multiple Regression Analysis of Factors Related to Psychopathology Symptoms.
Note. b = unstandardized coefficient; ß = standardized coefficient.
*p < .05. **p < .01. ***p < .001.
To test the second hypothesis—that coping style would mediate relations between emotional closeness and CG—we tested a mediation model utilizing the PROCESS macro for SPSS (Hayes, 2017), wherein the two types of coping would mediate the relationship between closeness to the deceased and CG symptoms after controlling for age and time since loss. The model and unstandardized coefficients are presented in Figure 1, and the direct and indirect effects of emotional closeness through parallel paths for coping are presented in Table 3. We found a significant indirect effect for emotional closeness via approach-focused coping (ab = 2.25, 95% confidence interval (CI) = [1.120, 3.57]). The direct effect of relationship closeness on CG relationship remained significant (c’ = 6.76, 95% CI = [4.55, 9.31]), indicating partial mediation.
Mediation Analysis of Emotional Closeness on Complicated Grief Symptoms When Mediated by Coping Styles. *p < .05. **p < .01. ***p < .001.
Direct Effects and Indirect Effects in Mediation Models.
Note. IE 95% CI is the 95% confidence interval based upon 1,000 nonparametric bootstrapping. DE = direct effect; IE = indirect effect; TE = total effect.
*p < .05. **p < .01. ***p < .001.
To examine the third hypothesis—that this pattern of relations would be different in depression rather than CG—we examined a multiple regression model with depressive symptoms as an outcome, both coping styles, emotional closeness, time since the loss, and age in the model. Multivariate analyses revealed two outliers which were removed. The results for the multiple regression analyses are displayed in Table 2. Omnibus test of the model whole was significant, F(6,197) = 30.17, p < .001, R2 = .44. Approach-focused coping, b = .66, t(205) = 11.23, p < .001, and age, b = .06, t(205) = 2.00, p = .047, were significantly associated with higher levels of depressive symptom. However, neither avoidance-focused coping nor emotional closeness was significant.
To further examine process differences between CG and depressive symptoms, we conducted a mediation analysis with depressive symptom severity as an outcome variable retaining both coping styles, emotional closeness, time since loss, and age. The unstandardized coefficients for each path are specified in Figure 2, and a summary of the direct and indirect effects are presented in Table 3. Neither direct nor total effects for emotional closeness on depression were observed. However, indirect effects of the relationship of emotional closeness through approach-focused coping were significant (ab = 1.01, 95% CI = [.50, 1.55]).
Mediation Analysis of Emotional Closeness on Depressive Symptoms When Mediated by Coping Styles. *p < .05. **p < .01. ***p < .001.
Discussion
This study sought to examine the association of different coping styles with CG symptoms in a bereaved sample. We hypothesized that both approach-focused and avoidance-focused coping would be associated with CG. This hypothesis was partially supported, as approach-focused coping was significantly associated with CG, but avoidance-focused coping was not. In addition, we hypothesized that approach- and avoidance-focused coping would mediate the association between the emotional closeness—a determinant of that role’s centrality on an individual’s sense of identity—and CG. Again, this was partially supported, as approach-focused coping mediated this relationship, though avoidance-focused coping did not. We also conducted analyses to determine whether the results were maintained with depression as the outcome variable. Although we replicated the effect of approach-focused coping, the association of emotional closeness did not remain significant.
The results of this study provide evidence that CG symptomology may not be fully explained by avoidance-focused coping behavior after loss. In fact, approach-focused coping (i.e., approaching reminders of the loss by thinking about them more and then dwelling on the emotional response) was associated with higher levels of CG, and avoidance-focused coping was not associated with CG. This is inconsistent with previous predictions of the attachment theory of CG, though the approach-focused findings are consistent with basic (Maccallum et al., 2015) and clinical neuroscience research in CG (Arizmendi et al., 2016; O’Connor et al., 2008; O’Connor & Arizmendi, 2015).
Examination of this process in the development of depression demonstrated that approach-focused coping was again associated with higher depressive symptoms, though avoidance-focused coping was not significant, and emotional closeness had no direct effect on depressive symptoms. The different patterns of association and influence between coping styles, emotional closeness, and CG or depressive symptoms provide process-oriented evidence for a distinction between the two constructs, in that the mechanisms associated with higher symptoms of each may be distinct. The unique role of emotional closeness in CG provides a differentiation between this condition and depression. This may be due to the bereavement-focus of CG when compared with the more global attributions in depressive symptoms (Raps et al., 1982). Although CG has traditionally been viewed as resulting from excessive avoidance behavior (Shear et al., 2007), others have argued that approach may also be implicated (Maccallum & Bryant, 2019). Our results are consistent with this idea, given the absence of a significant association of avoidance-focused coping style on CG symptoms, and the significance of approach-focused coping with its emphasis on approaching and experiencing the negative emotions from the reminders of the loss.
Several study limitations warrant discussion. First, the cross-sectional design of the study makes causal direction questionable. Although the arrangement of this model is consistent with identity continuity theory and previous research (Boelen et al., 2003; Schnider et al., 2007), future studies should use a longitudinal design to test the alternative hypothesis that grief is disrupting coping strategies. Second, all measures utilized in this study were self-report and only one measure was used for each construct. Future studies should seek a behavioral outcome or biological marker. Furthermore, while the examination of coping was more specific to reminders of the loss, other reminders of the deceased may be relevant and produce different results (e.g., reminders of the deceased individual; Maccallum & Bryant, 2019). Third, as participants were recruited from online bereavement sample, they were self-selected and, as such, may not be representative of a broader bereaved sample. Particularly, participants in the study primarily experienced their loss 2 to 3 years previously, while other studies have examined people sooner after the loss. Such a difference may account for the higher degree of approach-focused rather than avoidance-focused coping, which may predominate in the early stages of loss.
Overall, this study contributes additional support for the hypothesis that avoidance is not the only maladaptive factor in the development of CG. It suggests that understanding CG through an avoidance framework may be insufficient, and other patterns of coping styles should be considered. Future studies should elaborate more fully how coping beyond avoidance may impact trajectories of adjustment in the case of CG, particularly other dimensions of coping beyond general maladaptive or adaptive coping (e.g., coping flexibility; Bonanno et al., 2004; Burton et al., 2012).
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
