All people lose significant others to death as they traverse their lifespan. The death of close others is a major life stressor (Stroebe et al., 2013) that prompts emotional, social, and practical changes in the life of the bereaved. Individuals must continue with their daily activities but in the absence of a person who was an integral part of their life. They are faced with countless practical tasks related to the death. Their roles may change, and their identity may be challenged (Eckholdt et al., 2018; Harris et al., 2021). Though loss is stressful, most individuals cope. Some, however, experience prolonged, negative reactions, that is, symptoms of psychopathology (O'Connor, 2010). This may include symptoms of depression, post-traumatic stress disorder (PTSD), anxiety, and prolonged grief disorder (PGD). Affected individuals experience painful yearning for the deceased, lack of motivation and reduced positive affect, overwhelming worries, and/or intrusive images related to the loss (Boelen, 2021; Komischke-Konnerup et al., 2021; Maccallum & Bryant, 2011).
Cognitive theories suggest that severe negative reactions to loss derive from poor integration of the loss into autobiographical memory, an identity characterized by merging with the deceased, negative self-beliefs, and interpreting symptoms in maladaptive ways (Boelen et al., 2006; Maccallum & Bryant, 2013). In the present paper, we take a narrative approach to illuminating the processes involved in negative reactions to loss. This is grounded in literature on the key role of narratives in those processes theorized to be involved in negative reactions, including autobiographical memory, identity, and emotional processing (Bluck & Habermas, 2000; Conway et al., 2004; Habermas, 2019; McAdams, 2001; McLean et al., 2007).
According to the narrative approach, individuals attempt to adapt to the loss through constructing a narrative that makes sense of the loss and its impact on their life and identity (Baddeley & Singer, 2010; Bluck & Mroz, 2018; Neimeyer et al., 2014). Creating a personal narrative helps individuals organize events in time and make attributions about causes and consequences (Bruner, 1990). Narratives of autobiographical events are intertwined with identity as they represent what the person did to affect the course of events and how these events in turn impacted the self, e.g. narrative identity (McAdams & McLean, 2013; Pasupathi et al., 2007). As such, loss narratives constitute a phenomenological perspective on identity involved in adaptation to loss. The importance of narrative in dealing with loss is also evident in narrative-focused interventions developed to support individuals with prolonged negative reactions (Barbosa et al., 2014; Neimeyer et al., 2008).
Prior studies have focused on investigating whether characteristics of loss narratives predict mental health over time (Bauer & Bonanno, 2001; Capps & Bonanno, 2000; Thomsen et al., 2018). The implicit assumption in this research is that narrative identity processes influence how individuals react to the loss. This perspective is consistent with theoretical frameworks on PGD that emphasize identity as contributing to mental health after loss (Boelen et al., 2006; Maccallum & Bryant, 2013). We provide a new perspective on the relation between mental health and narrative identity by examining whether symptomology after a loss predicts individuals’ construction of loss narratives. To address this issue, we followed 507 bereaved spouses and adult children over 16 months and examined whether levels of psychopathology symptoms over time predicted themes and emotional qualities of loss narratives.
Characteristics of Adaptive Loss Narratives
Studies on narrative identity demonstrate that various characteristics of narratives are related to better mental health (Adler et al., 2016; Bauer, 2021). We drew from that literature to identify four characteristics of loss narratives to investigate in the current research: Agency and communion themes, emotional tone, and self-event connections. Together, these characteristics capture central human needs, affective orientation, and integrating the loss into identity in adaptive ways.
Agency and communion refer to two recurring patterns of human intentions that emerge as themes in narratives (McAdams, 1993). Agency themes include striving for power, autonomy, and mastery, whereas communion themes comprise striving for closeness, being a part of a larger whole, and nurturing. Both agency and communion may be challenged by the death of close others. Death is the ultimate reminder of our inability to prevent the passing away of our loved ones, a clear challenge to personal agency. Though we cannot prevent death itself, narrating loss with some modicum of agency is possible by highlighting those parts where we were able to play a directive role before, during or after the death. Similarly, communion with the loved one is challenged by death but narratives of loss can still emphasize caring and loving in how relationships to the lost one and with other people are depicted (Mroz et al., 2020). Studies have confirmed that higher agency and communion themes in loss narratives relate to better outcomes (Habermas, 2021; Huang & Habermas, 2019). Agentic language in relationship narratives shared by bereaved spouses predicted better mental health over time when controlling for baseline measures of mental health (Bauer & Bonanno, 2001; Capps & Bonanno, 2000). We take a different perspective on the temporal relationship between loss narratives and mental health, and test whether higher initial symptoms levels predict lower agency and communion themes in loss narratives shared 16 months later.
Another important characteristic of loss narratives is their emotional tone. Clearly narratives of illness and death may involve negative affect, including anger, sadness, and confusion. For some individuals, negative reactions may dominate the narrative. While the death of a close other is stressful, the period of the loss, or certain memories from that time, may still be narrated positively (Bluck & Mroz, 2018; Wolf et al., 2023) with the feeling that the loved one had a good ending to their life (Generous & Keeley, 2022). Spouses and adult children may construct a narrative that emphasizes positive aspects. They may reflect on how their caregiving and interactions brought comfort to the loved one and highlight positive emotions such as love and gratefulness (Lowers et al., 2020).
Loss narratives may involve linking the loss to positive and negative stability and change in the individual’s identity, also termed self-event connections (Liao & Bluck, 2022; Pasupathi et al., 2007). That is, the narrator engages in reasoning about their own stable characteristics (self-stability connections, both negative and positive) or changes (self-change connections, both negative and positive) in relation to the loss. Individuals may interpret themselves negatively, for example, considering how they have always been poor at coping (negative self-stability connection) or have become more vulnerable after the loss (negative self-change connection). However, they could also narrate themselves as bringing their strengths into play during the period of the loss and as undergoing positive changes from coping with loss-related challenges (Mackay & Bluck, 2010; Mroz et al., 2020). Research has demonstrated that more positive and less negative loss narratives are associated with better mental health, including lower psychopathology (Bauer & Bonanno, 2001; Huang et al., 2020; Maccallum & Bryant, 2008; Mroz et al., 2020; Thomsen et al., 2018). However, higher levels of initial symptoms of psychopathology may be an obstacle to forming loss narratives that emphasize positive emotions and self-event connections.
As shown, previous research indicates that how individuals narrate their loss may impact their level of psychopathology symptoms. Likewise, theories suggest that identity is causally involved in the development of PGD (Maccallum & Bryant, 2013). However, we suggest that the reverse may also be true: That negative reactions, as captured in high symptomatology, may interfere with the construction of an adaptive narrative identity (i.e., one with relatively high agency and communion, lower negative self-event connections, and greater positive self-event connections) after the loss. Having a maladaptive narrative could contribute to maintaining high symptom levels. This idea is consistent with literature suggesting that psychopathology compromises identity in ways that undercut adaptive coping (Thomsen et al., 2023). Individuals who react to the loss with more symptoms of psychopathology may shy away from exploring the loss through narrating: they may struggle to narrate themselves as agentic and communal, finding it hard to construct positive self-event connections. That is, those bereaved individuals who find themselves burdened with constant yearning for their close other, uncontrollable worries and intrusive images, difficulties motivating themselves to engage in activities and feeling a lack of pleasure in life, may create loss narratives that highlight a sense of their own lack of ability to take charge of their life, their loneliness and isolation, and a view of themselves as defective or having been damaged by their loss. Such loss narratives and the accompanying negative identity implications may undermine healthy coping efforts and hinder identity adaptation to the bereaved person’s altered life circumstances, thereby maintaining higher symptom levels.
The Present Study and Hypotheses
We followed 507 bereaved adults (loss of spouse or parent) over 16 months. The data derived from the Aarhus Bereavement Study (Harris et al., 2021; Lundorff et al., 2020). This multi-wave study encompasses various measurements, including several indicators of psychopathology symptoms (depression, generalized anxiety, PTSD, and PGD), assessed at 2 (T1), 6 (T2), 11 (T3), and 18 (T4) months after the loss. At the latest time point (i.e., only at 18 months post loss) participants wrote a narrative of their loss. We asked them to write not just about the death itself, but about the period concerning the loss of their spouse/parent, because the experience of loss extends over time rather than being limited to the moments of death. We examined whether more symptoms of psychopathology at T1, T2, T3, and T4 related to and predicted characteristics of loss narratives written 18 months after the loss (T4).
We hypothesized that having greater symptoms of psychopathology at T1, T2, T3, and T4 would predict T4 loss narratives with: (1) lower agency themes, (2) lower communion themes, (3) less positive and more negative emotional tone, and (4) less positive and more negative self-event connections. We assessed agency and communion themes in the narratives with standard, reliable content-coding methods (Adler et al., 2017). Emotional tone and self-event connections were assessed using self-report questions developed in prior studies (Jensen et al., 2020).
To examine whether symptoms of psychopathology predicted characteristics of loss narratives over and above other correlates of reactions to loss, we included measures that may relate to narrative characteristics and symptoms of psychopathology. These included neuroticism, attachment, type of relationship (i.e., loss of spouse or parent), gender, age, and type of death (Boelen et al., 2006). In our statistical analyses, we then control for any of these variables that were found to relate to characteristics of the loss narratives.
Method
Hypotheses 1 and 2 and related analyses concerning agency and communion themes were pre-registered (https://aspredicted.org/gg538.pdf). Hypotheses 3 and 4 concerning emotional tone and self-event connections were not pre-registered, as initial analyses of these variables were conducted prior to registration. The data is not being made publicly available due to national data sharing rules. Analyses of the narratives are unique to this paper, and other articles relying on this dataset do not overlap with the current study (Harris et al., 2021; Johannsen et al., 2022; Komischke-Konnerup et al., 2021; Lundorff et al., 2020, 2021; Vang et al., 2022).
Participants
The participants were a subsample from the Aarhus Bereavement Study. National registers in Denmark were used to identify individuals between 25 and 85 years living within the Aarhus area who lost their spouse between January 2017 and March 2018. Identified individuals received an invitation to participate, approximately 1 month following the death of their spouse and, if so agreed, responded to the first questionnaire 2 months post loss. To recruit adult child participants, spouse participants were invited to share information about the research with their adult children.
We included those participants who at T4, 18 months after their loss, responded to the loss narrative prompt. This included 507 adults with 331 having lost their spouse and 163 having lost their parent (13 missing responses for this item). There were 340 women (67%) and 154 men (30%; 13 missing responses), and the mean age was 61.85 years (SD = 16.24, range 18–86). The sample for the present study did not differ significantly from the full sample responding at T1 on the psychopathology symptom measures, ts(1185) < 1.60.
The sample was relatively well-educated with 12% having a university degree at the MA level, 26% having the Danish equivalent of an undergraduate college degree, and the remaining 58% reporting other education, manual training, high school, or secondary school as their highest level of education (20 missing responses). Finally, cause of death of their lost one was reported as from different conditions, though largely cancer.
Materials and Measures
Loss narrative prompt. Participants were given the following instruction with an open text field to write their narrative: “In this part of the study, we ask you to please describe the period or the chapter that concerns the loss of your spouse/parent. Please include descriptions of your relationship with your spouse/parent, what happened during the time before the loss and the following months, what you thought and felt during the course of events, what the period says about who you are as a person and how the period may have changed you”. We asked for a period/chapter because loss is best conceived as an extended process that would be mentally represented as a life story chapter (McAdams, 2001; Thomsen, 2015).
Emotional tone and self-event connections. After narrating their loss chapter participants rated positive/negative emotional tone, positive/negative self-change connections, and positive/negative self-stability connections in relation to the loss using six questions. Questions were rated on 5-point scales anchored with 1 = not at all and 5 = to a very high degree. Similar questions have been used in previous studies and their construct validity is supported by correlations with personality traits and symptoms of psychopathology (Holm & Thomsen, 2018; Jensen et al., 2020). The six questions were (positive/negative indicates two separate questions):” Try to think about how you experienced the period when it happened. To what degree would you describe what happened during this period as something positive/negative?” [positive/negative emotional tone]; “Try to think about how you experience that period today. Does the period emphasize some positive/negative attributes that characterize you today?” [positive/negative self-stability connections]; and “Try to think about how you experience the period today. Has the period changed you as a person in a positive/negative way?” [positive/negative self-change connections].
The participants completed psychopathology symptom measures at T1–T4. In addition, they completed measures of neuroticism and attachment at T1, which are used as control variables in the present study. For all scales missing values were replaced using the Expectation–Maximization algorithm (Twala, 2009) if less than 50% of items were missing. The measures include those listed below.
Symptoms of PGD were measured using the Prolonged Grief-13 (Prigerson et al., 2009). The scale is validated in Danish and includes 11 items assessing symptoms within the last month, including yearning, shock, and numbness as well as two items measuring functional impairment and symptom duration, which were not included in the sum score for the scale (Vang et al., 2022). Symptom items were all rated on 1–5 points scales with higher scores indicating more severe symptoms.
Symptoms of anxiety were measured using the Generalized Anxiety Disorder-7 (Spitzer et al., 2006). The scale includes seven items assessing generalized anxiety disorder symptoms within the last two weeks, including worrying, difficulties relaxing, and feelings of nervousness and anxiety. Items were rated on 0–3 points scales with higher scores indicating higher symptom levels.
Symptoms of PTSD were measured using the Posttraumatic Stress Disorder Checklist for DSM-5 (PCL-5) (Blevins et al., 2015; Forkus et al., 2022). The list includes 20 symptoms within the domains of re-experiencing, avoidance and hyperarousal that are all rated on 0–4 points scales with respect to the selected event. For the present study, the scale was adapted to invite participants to complete the items with respect to the death of their close other and instructions asked them to consider their experiences within the last month. Higher scores indicate more severe symptoms.
Symptoms of depression were measured using the Center for Epidemiologic Studies Short Depression Scale (Radloff, 1977; Weiss et al., 2015). The scale includes 10 items measuring symptoms such as poor sleep, lack of positive affect, and feelings of depression all rated on 0–3 points scales with respect to frequency within the last week. Higher scores reflect more severe symptom load.
Neuroticism was measured with the NEO Personality Inventory-Revised neuroticism subscale (Costa & McCrae, 1992) which includes 12 items rated on 1–5 points scales. Higher scores indicate more neuroticism.
Attachment was measured with the Experiences in Close Relationship Scale-Short (Wei et al., 2007) that assesses attachment in romantic relationships. The scale includes 12 items that assess orientations of anxious attachment (6 items) and avoidant attachment (6 items), rated on 1–7 points scales with higher scores indicating more anxious/avoidant attachment. Participants who were spouses to the deceased were instructed to complete the items with respect to how they experienced their relationship with their spouse. For the adult child participants, the response was based on their experience of romantic relationships in general.
Procedure
Approximately 2 months after the death of their spouse or parent, individuals provided informed consent and received either an electronic link with the questionnaire or a paper copy of the questionnaire. They responded to questionnaires again at 6, 11, and 18 months after the death of their close other (the content differed across T1–T4 but always included the measures of symptoms described above). The loss narratives were only collected at T4, and after all other measures). In case of non-response, participants received up to two reminders.
Content-Coding: Agency and Communion
The loss narratives were coded for agency and communion themes using an adapted coding manual developed by Jonathan Adler (McLean et al., 2020). Agency themes were scored from 0 to 4, with 0 representing individuals who narrate themselves as powerless and at the mercy of external circumstances and 4 representing individuals who narrate themselves as able to affect their own lives, initiate change, and have some degree of influence over events in their lives. Scores of 1, 2, and 3 represent intermediate levels of agency themes. Communion themes were likewise scored from 0 to 4, with 0 indicating individuals who narrate themselves as completely disconnected, isolated or rejected and 4 indicating individuals who narrate themselves as highly connected to others. Scores of 1, 2, and 3 represent intermediate levels of communion themes. The first author adapted the coding manual to capture specific expressions of agency and communion themes in the context of loss narratives. These adaptions included examples of how agency and communion themes may emerge in the narrative, e.g., making decisions about how to cope with illness and death preparations and showing strength in coping with the loss [examples of agency themes] and describing a loving relationship with the deceased as well as support from others [examples of communion themes]. The adapted coding manual and examples of loss narratives high and low on agency and communion themes can be found in Appendices A and B.
The first author trained an MSc student in Psychology using the narratives from the current study. When interrater reliability was satisfactory, the student assistant coded all narratives independently. A reliability check was performed with the first author independently coding 20% of the narratives. Interrater reliability as calculated with intraclass correlations was good for agency (0.77) and for communion (0.73).
Results
We first report descriptive analyses on symptom measures and narrative characteristics, and then provide preliminary analyses to identify relevant control variables for the main analyses (e.g., demographic variables, neuroticism, and attachment). In the main analyses, we report the pre-registered correlations and multiple regression analyses concerning agency and communion themes (hypotheses 1 and 2) as well as a similar set of analyses for positive/negative emotional tone and self-event connections (hypotheses 3 and 4).
Descriptive Analyses
The means for all psychopathology symptom measures as well as the control variables, neuroticism and attachment orientations, are shown in Table 1. As would be expected, symptoms were higher at T1 and then dropped over time with the lowest mean scores at T4, Fs(3, 1401) > 29.93, ps < 0.001. At the same time, measures of symptoms at T1–T4 correlated highly: 0.75 to 0.84 for PGD symptoms; 0.63 to 0.72 for anxiety symptoms, 0.73 to 0.82 for PTSD symptoms, and 0.65 to 0.79 for symptoms of depression. Interrelations between all symptom measures were positive and moderate to high (rs from 0.48 to 0.80).
Table 1
Descriptive statistics for psychopathology symptom measures (T1–T4) and correlations with characteristics of the loss narrative at T4
M (SD) | Agency | Communion | Positive tone | Negative tone | Positive self-change | Negative self-change | Positive self-stability | Negative self-stability | |
|---|---|---|---|---|---|---|---|---|---|
T1 PGD | 25.34 (8.77) | − 0.34* | − 0.26* | − 0.20* | 0.30* | − 0.05 | 0.36* | − 0.11 | 0.42* |
T2 PGD | 23.59 (8.34) | − 0.36* | − 0.26* | − 0.18* | 0.32* | − 0.09 | 0.41* | − 0.17* | 0.42* |
T3 PGD | 21.43 (7.63) | − 0.41* | − 0.28* | − 0.18* | 0.28* | − 0.09 | 0.43* | − 0.19* | 0.46* |
T4 PGD | 20.78 (7.67) | − 0.42* | − 0.32* | − 0.18* | 0.30* | − 0.10 | 0.47* | − 0.17* | 0.48* |
T1 anxiety | 4.25 (4.39) | − 0.30* | − 0.28* | − 0.17* | 0.22* | − 0.00 | 0.39* | − 0.13 | 0.38* |
T2 anxiety | 3.93 (4.35) | − 0.32* | − 0.24* | − 0.11 | 0.23* | − 0.08 | 0.43* | − 0.14 | 0.42* |
T3 anxiety | 3.20 (3.96) | − 0.30* | − 0.25* | − 0.15 | 0.23* | − 0.03 | 0.46* | − 0.17* | 0.40* |
T4 anxiety | 3.00 (3.88) | − 0.32* | − 0.25* | − 0.19* | 0.21* | − 0.09 | 0.39* | − 0.19* | 0.39* |
T1 PTSD | 12.68 (11.39) | − 0.34* | − 0.31* | − 0.19* | 0.30* | − 0.03 | 0.43* | − 0.13 | 0.46* |
T2 PTSD | 11.41 (11.29) | − 0.37* | − 0.29* | − 0.14 | 0.28* | − 0.11 | 0.47* | − 0.19* | 0.43* |
T3 PTSD | 9.72 (9.86) | − 0.37* | − 0.30* | − 0.18* | 0.29* | − 0.07 | 0.50* | − 0.18* | 0.45* |
T4 PTSD | 8.97 (9.63) | − 0.42* | − 0.33* | − 0.18* | 0.29* | − 0.10 | 0.51* | − 0.16* | 0.51* |
T1 depression | 8.84 (5.71) | − 0.36* | − 0.29* | − 0.18* | 0.23* | − 0.07 | 0.39* | − 0.14 | 0.41* |
T2 depression | 8.13 (5.75) | − 0.37* | − 0.26* | − 0.14 | 0.23* | − 0.10 | 0.43* | − 0.19* | 0.41* |
T3 depression | 7.25 (5.21) | − 0.36* | − 0.28* | − 0.14 | 0.21* | − 0.10 | 0.44* | − 0.17* | 0.37* |
T4 depression | 6.92 (5.04) | − 0.43* | − 0.30* | − 0.20* | 0.22* | − 0.17 | 0.42* | − 0.22* | 0.39* |
T1 neuroticism | 29.66 (8.67) | − 0.29* | − 0.26* | − 0.08 | 0.12 | − 0.02 | 0.32* | − 0.14 | 0.41* |
T1 attachment avoidance | 14.03 (5.58) | − 0.18* | − 0.25* | − 0.04 | 0.03 | − 0.04 | 0.13 | − 0.09 | 0.12 |
T1 attachment anxiety | 18.44 (6.01) | − 0.14 | − 0.18* | − 0.01 | − 0.01 | 0.09 | 0.16 | − 0.05 | 0.21* |
The mean scores for the characteristics of loss narratives were as follows: Agency themes (M = 2.02; SD = 0.89); communion themes (M = 2.14; SD = 0.77); positive emotional tone (M = 2.34; SD = 1.17); negative emotional tone (M = 3.25; SD = 1.24); positive self-change connections (M = 2.57; SD = 1.10); negative self-change connections (M = 1.51; SD = 0.80); positive self-stability connections (M = 2.90; SD = 1.07), and negative self-stability connections (M = 1.91; SD = 0.96). In general, the sample rated the loss narratives as more negative than positive in emotional tone but with substantial variation t(506) = 9.51, p < 0.001, d = 0.42. As a group, participants endorsed positive self-event connections to a higher extent than negative self-event connections (self-change: t(506) = 16.75, p < 0.001, d = 0.74 and self-stability: t(506) = 14.87, p < 0.001, d = 0.66), indicating that positive identity implications are common even amidst the stresses of the loss.
Preliminary Analyses
To identify control variables for the main analyses, we conducted a series of tests to examine correlates of the loss narratives characteristics. The mean number of words in the loss narratives was 155.58 (SD = 189.50). Number of words correlated positively and significantly (all ps < 0.05) but with small effect sizes with negative emotional tone, r(505) = 0.10, negative self-stability connections, r(505) = 0.09, and negative self-change connections, r(505) = 0.13. None of the other correlations reached significance, rs(505) < 0.08. Based on these analyses, number of words does not appear to be a major explanatory factor with respect to characteristics of loss narratives and was not included as a covariate in any analyses.
We examined associations between characteristics of loss narratives and demographic variables, type of relationship with the deceased, and type of death. There were no significant differences by gender (ts < 1.80), type of relationship (spouse vs. parent; ts < 1.45), or type of death (F < 1.90). Age correlated with small effect sizes but significantly (all ps < 0.05) and positively with communion themes, r(499) = 0.10, and positive emotional tone, r(499) = 0.09, and negatively with negative emotional tone, r(499) = − 0.19, negative self-stability connections, r(499) = − 0.13, negative self-change connections, r(499) = − 0.20, and positive self-change connections, r(499) = − 0.11. With the exception of the last finding on positive self-change, the age-related findings are consistent with studies showing that older age is related to positivity in memory and life stories (Carstensen & Mikels, 2005; Jensen et al., 2020). It is also possible that these relations reflect that loss is more normative, more expected, in older age. Based on these preliminary findings, it appears that individual differences in loss narratives are associated with age, so we include age in the multiple regressions reported below.
As a further step in identifying relevant control variables, we examined whether characteristics of loss narratives were related to neuroticism and attachment (i.e., avoidance and anxiety) measured at T1, by running a planned series of correlations (see Table 1). Neuroticism at T1 predicted lower agency and communion themes as well as less positive and more negative loss narratives at T4. The highest correlation coefficients were found for neuroticism with agency themes (− 0.29) and negative self-event connections (self-change: 0.32 and self-stability: 0.41). These findings are in line with other research on neuroticism and narrative identity (Jensen et al., 2020; McAdams et al., 2004). The prospective relations between attachment avoidance and anxiety and characteristics of loss narratives were in the expected direction but were generally weak, only reaching significance for agency and communion themes (− 0.18 and − 0.25) and negative self-stability connections (0.21). Since the correlations with neuroticism had larger effects sizes and were consistent across characteristics of the loss narratives, we included neuroticism as a control variable in the multiple regressions reported below.
Main Analyses
We conducted a series of correlations before running regression analyses. To test our pre-registered hypotheses 1 and 2 that symptoms of psychopathology measured at T1, T2, T3, and T4 were associated with agency and communion themes in loss narratives at T4, we ran the planned series of correlations. Similarly, to test hypotheses 3 and 4 we correlated symptom measures with emotional tone and self-event connections in the narratives (see Table 1 for all correlations). Because we ran 128 correlations between eight characteristics of the loss narratives and 16 T1–T4 measures of psychopathology, we Bonferroni corrected the p-level by dividing 0.05 by 128 and derived a significance level of p < 0.0004. Using this p-level, having more severe symptoms of psychopathology was related to loss narratives characterized by lower agency and communion themes. This was true for all four time points (rs from small to medium effect sizes; − 0.24 to − 0.43). The pattern of correlations for symptoms across T1–T4 with negative emotional tone and self-event connections (T4) was similar (rs from 0.21 to 0.51) but with lower effect sizes and more non-significant correlations for positive emotional tone and positive self-event connections (rs from − 0.00 to − 0.22). Notably, the correlation coefficients between symptoms and the loss narrative characteristics (T4) did not vary widely across measurement times (T1 to T4). There was a tendency, however, towards correlation coefficients for later measures of psychopathology symptoms to show stronger relations to the T4 narratives. We explored whether change between T1 and T4 for each of the symptom measures (i.e., change score = T1–T4 score), was related to any of the characteristics of loss narratives but did not find any significant correlations (rs < 0.095). This implies that there is no relationship between decrease or increase in symptom measures over time and loss narratives.
Following our pre-registered analyses, we ran two separate multiple regressions to test whether symptoms of psychopathology predicted agency and communion themes in loss narratives. Based on the preliminary analyses, we controlled for neuroticism and age. We ran the same set of multiple regressions for emotional tone and self-event connections variables. We had pre-registered that we would include the measurement of symptoms of psychopathology from only one time-point (i.e., T1–T4), the one with the numerically highest correlation to characteristics of the loss narrative. However, the correlations with narrative characteristics were very similar across T1–T4. As such, we decided to use T1 measures of symptomology to predict T4 narrative characteristics to take full advantage of this longitudinal dataset for testing relations prospectively over 16 months. Regression results can be seen in Table 2. Note that we do not report the regressions for positive self-change connections and positive self-stability connections as the overall models were not significant, Fs(6, 493) = 1.41 and 2.03 respectively. PGD symptoms at T1 predicted lower agency themes, higher negative emotional tone, and more negative self-stability connections in loss narratives at T4, after controlling for neuroticism and age. Other symptom measures did not relate significantly to characteristics of the loss narratives after controlling for neuroticism and age.
Table 2
Multiple regressions predicting characteristics of the loss narrative at T4 from T1 psychopathology symptom measures
Agency | Communion | Positive tone | Negative tone | Negative self-change | Negative self-stability | |
|---|---|---|---|---|---|---|
Step 1 | ||||||
Neuroticism | − 0.29* | − 0.25* | − 0.07 | 0.10* | 0.31* | 0.40* |
Age | − 0.00 | 0.08 | 0.08 | − 0.18* | − 0.16* | − 0.08 |
Step 2 | ||||||
Neuroticism | − 0.09 | − 0.07 | 0.10 | − 0.11 | 0.04 | 0.20* |
Age | 0.03 | 0.08 | 0.10* | − 0.19* | − 0.16* | − 0.09* |
PGD | − 0.15* | − 0.07 | − 0.12 | 0.23* | 0.10 | 0.20* |
Anxiety | 0.01 | − 0.03 | − 0.03 | − 0.03 | 0.05 | − 0.04 |
PTSD | − 0.06 | − 0.11 | − 0.06 | 0.13 | 0.15 | 0.16* |
Depression | − 0.16 | − 0.10 | − 0.09 | 0.06 | 0.15 | 0.06 |
Model summary | F(6, 493) = 14.54*, adj R2 = .14 | F(6, 493) = 10.62*, adj R2 = .11 | F(6, 493) = 4.88*, adj R2 = .05 | F(6, 493) = 13.05*, adj R2 = .13 | F(6, 493) = 23.69*, adj R2 = .22 | F(6, 493) = 27.98*, adj R2 = .25 |
The four measures of symptoms overlap in predicting narrative characteristics (Table 1 and descriptive analyses of relationships between symptoms measures). The analyses reported above indicated that PGD symptoms were most robust in predicting characteristics of loss narratives. Hence, we also explored whether symptoms of PGD predicted agency and communion themes, emotional tone, and self-event connections, after controlling for neuroticism and age, but excluding the three other measures of symptoms (e.g., depression, anxiety, and PTSD). This set of analyses were not pre-registered but followed from the above observations and help illuminate whether symptoms of PGD play a distinctive role in predicting characteristics of loss narratives, while avoiding potential problems of high covariation between predictor variables. Symptoms of PGD significantly predicted six of eight narrative characteristics (see Table 3, note that regressions for positive self-change and positive self-stability connections are not reported as PGD did not significantly predict these, βs = − 0.03 and − 0.06, ps > 0.05). Notably, higher initial levels of PGD predicted loss narratives that 16 months later were characterized by lower agency and communion themes (βs = − 0.27 and − 0.20), less positive tone (β = − 0.22), more negative tone (β = 0.33), and negative self-event connections (βs = 0.30 (self-change) and 0.31 (self-stability)).
Table 3
Multiple regressions predicting characteristics of the loss narrative at T4 from T1 PGD symptoms controlling for neuroticism and age
Agency | Communion | Positive tone | Negative tone | Negative self-change | Negative self-stability | |
|---|---|---|---|---|---|---|
Step 1 | ||||||
Neuroticism | − 0.29* | − 0.25* | − 0.07 | 0.10* | 0.31* | 0.40* |
Age | − 0.00 | 0.08 | 0.08 | − 0.18* | − 0.16* | − 0.08 |
Step 2 | ||||||
Neuroticism | − 0.17* | − 0.16* | 0.03 | 0.05 | 0.17* | 0.25* |
Age | 0.03 | 0.10* | 0.11* | − 0.21* | − 0.19* | − 0.11* |
PGD | − 0.27* | − 0.20* | − 0.22* | 0.33* | 0.30* | 0.31* |
Model summary | F(3, 490) = 26.70* adj R2 = 0.14 | F(3, 490) = 18.91* adj R2 = 0.10 | F(3, 490) = 8.53* adj R2 = 0.04 | F(3, 490) = 24.67* adj R2 = 0.13 | F(3, 490) = 40.61* adj R2 = .19 | F(3, 490) = 53.57* adj R2 = .24 |
Discussion
We examined whether higher levels of psychopathology symptoms predicted less adaptive loss narratives in 507 bereaved adults. Individuals who across time reacted to the loss with more symptoms of psychopathology, narrated the loss with lower agency and communion themes, with more negative emotional tone, and with more negative self-event connections. In narrative identity terms, they constructed the loss narrative in ways that represented them as powerless, disconnected, and as having changed negatively from this life stressor. It is particularly notable that content-coded agency and communion themes were predicted by symptom measures because these associations cannot reflect shared method variance as could be the case for the self-reported characteristics of emotional tone and self-event connections. Analyses controlling for neuroticism and age indicated that initial PGD symptoms were a robust predictor of less adaptive loss narratives 16 months later. This is consistent with viewing PGD as a distinct disorder tied to the loss of a close other (Maccallum & Bryant, 2013) and suggests that higher levels of PGD symptoms, even soon after the loss, may reflect a detrimental trajectory ahead.
Loss Narratives and Symptoms of PGD
By demonstrating that symptoms of PGD predict characteristics of loss narratives, our findings expand prior studies that have also documented associations between loss narratives and concurrent and later mental health (Bauer & Bonanno, 2001; Capps & Bonanno, 2000; Huang & Habermas, 2019; Maccallum & Bryant, 2008; Thomsen et al., 2018). Our findings show that bereaved individuals who react to loss with more PGD symptoms, including intense yearning for the deceased, shock and numbness, have difficulties constructing an adaptive loss narrative that could support adjustment. Neuroticism has been related to more severe grief reactions and to negative narrative identity (Jensen et al., 2020; Maccallum & Bryant, 2013; McAdams et al., 2004). Our analyses also showed that neuroticism at T1 was related to negative loss narrative characteristics 16 months later. Importantly, however, the relations between symptomology and loss narrative characteristics did not simply reflect neuroticism. Effects persisted when neuroticism was controlled in analyses. Notably, symptoms of PGD were the most robust predictor of less adaptive loss narratives. This may reflect that PGD symptoms arise in the context of the loss, whereas symptoms of depression and anxiety may not relate specifically to the loss. Some adaptive characteristics of the loss narrative, including positive emotional tone and positive self-event connections were not as strongly predicted by PGD measures. Future studies could illuminate factors involved in the construction of more positive loss narratives, including social support and extraversion.
More broadly, the findings are consistent with the idea that psychopathology impacts narrative identity by fostering negative self-interpretations (Thomsen et al., 2023). This interpretation would have been more strongly supported in a design that included loss narratives at baseline so that we could have statistically controlled for baseline narrative characteristics in examining relations between symptomology and T4 narratives. Still, our findings suggest a bidirectional relation between narrative characteristics and psychopathological symptoms following loss (especially PGD). Theories tend to view narrative identity as affecting the emergence of symptoms (Maccallum & Bryant, 2013; Neimeyer, 2006), which is consistent with prior prospective studies on loss (Bauer & Bonnanno, 2001; Capps & Bonanno, 2000) and well as prospective studies on narratives and mental health more generally (Adler, 2012). Based on our findings (whether considering prospective or concurrent relations between symptoms and narrative characteristics), we suggest, however, further exploring bi-directional links between narrative identity and PGD symptoms. Individuals who are plagued with yearning and numbness may avoid using narrative to explore emotions and identity, which previous research has indicated can be beneficial (Pals, 2006). With fewer opportunities for explorative narrating, they may struggle to construct a loss narrative that emphasizes moments of agency and communion. The symptoms in themselves may become sources of negative self-interpretations (e.g., “I can’t cope” and “I am weak”) that over time are folded into less adaptive loss narratives. Carrying such narratives in memory as one moves forward in time may hinder healthy adaptation.
Clinical Implications
Our findings show that PGD symptoms are related to the construction of less adaptive loss narratives. We suggest that psychosocial interventions directly target loss narratives as a potential maintaining mechanism for prolonged grief. Existing narrative approaches emphasize the construction of narratives to make sense of the loss (Barbosa et al., 2014; Neimeyer et al., 2008). When doing so, therapists may invite individuals to focus narrative reconstruction on moments where agency and communion emerged most strongly during the loss period and in present life. This should be carefully balanced with narrative processing of negative emotions which may also be needed for healthy adjustment, and repeated narration of the loss may further aid grieving individuals in understanding and coping (Habermas, 2019). The grieving participants in our study found myriads of ways to exert agency and experience communion, including planning care of the deceased, asking for help and support, and engaging in social activities despite experiencing difficult emotions (see Appendices A and B for examples). Furthermore, our findings emphasize the need for an explicit focus on eliciting and encouraging reframing of negative self-interpretations that have become a part of the loss narrative, including those arising from intense symptom reactions (e.g., “There is something wrong with me”). This is consistent with existing cognitive behavioral approaches where identity and interpretations of symptoms are viewed as drivers in development and maintenance of disorders (Boelen et al., 2006; Maccallum & Bryant, 2013). We call attention to the key role of narratives in these processes and suggest integrating narrative methods into cognitive behavioral therapy. Narratives represent the loss and implicated processes of identity, appraisals, and memory in richly contextualized ways that help anchor interventions in individuals’ lived experience.
Limitations and Future Directions
The present study has several strengths including a relatively large sample, prospective assessments over 16 months, measuring a range of symptoms and control variables, and reliable content-coding of loss narratives for both agency and communion. However, there are also limitations. First, we cannot exclude the possibility that non-measured variables explain the prospective relations between PGD symptoms and characteristics of loss narratives. Specifically, we did not assess loss narratives in the early time points of the study. It is possible that these would directly predict later loss narratives eliminating effects of initial symptom levels and challenging the interpretation that symptoms can play a causal role in how loss narratives are constructed. Second, it is possible that aspects of the actual lived experience of the loss, determine both how individuals narrate it and the symptoms they experience. This would imply that loss narratives should be considered together with aspects of the actual loss in understanding grief reactions. Third, self-report scales, with their incumbent problems, were used as indicators of symptoms of psychopathology. Future studies could include clinical assessment. Fourth, the sample was recruited from a general population of bereaved individuals with only a minority reporting high symptom levels, thereby representing typical grief responses. Future studies could, however, include individuals at high risk of developing PGD to test whether our results generalize to clinical samples.
Conclusion
We conclude that high symptomatology after loss, especially PGD symptoms, predict less adaptive loss narratives, characterized by low agency and communion themes and more negative self-event connections. Less adaptive loss narratives may reflect narrative identity challenges after the death of a close other. Existing interventions, including cognitive behavioral approaches, could be tailored to target such narratives with the aim of supporting narrative reconstruction. Such work could focus on increasing agency and communion as well as self-event connections emphasizing how strengths, growth, and values emerged during this difficult life period.
Acknowledgements
We would like to thank Signe Lehn Brand for her assistance with content-coding. The study was supported by a grant to the last author from The Aarhus University Research Foundation.
Declarations
Conflict of interest
The authors report no conflict of interest.
Human or Animal Participants
The study involved human participants and was conducted in accordance with ethical standards for research, including asking participants to sign an informed consent form before participation.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
