Narrative reconstruction therapy for prolonged grief disorder - a pilot study.

Background: Prolonged grief disorder (PGD) is a chronic and disabling condition that affects approximately 10% of non-traumatically bereaved people. Narrative reconstruction (NR), originally designed for the treatment of posttraumatic stress disorder (PTSD), is a time-limited integrative therapy consisting of exposure to the loss memory, detailed written reconstruction of the loss memory narrative, and an elaboration of the personal significance of that memory for the bereaved. Objective: This pilot study examined the efficacy of NR therapy in reducing symptoms in bereaved people diagnosed with PGD. Method: Ten PGD patients participated in the study and were treated with 16 weekly sessions of NR. PGD, PTSD, and depression symptoms, as well as levels of loss integration, were assessed at pre-treatment, post-treatment, and at a 3-month follow-up.
Results: Following NR, participants showed significant reductions in PGD, depression, and PTSD symptoms, and elevated levels of trauma integration. Symptoms showed further improvement at the three-month follow-up. Conclusions: These findings provide preliminary evidence for the feasibility and efficacy of NR in treating PGD. Narrative reconstruction therapy requires further evaluation in randomized controlled trials.

Prolonged grief disorder (PGD) was included as a formal diagnosis for the first time in the ICD-11 (World Health Organization [WHO], 2018) and was recently approved as a formal diagnosis by the DSM-5 steering committee (APA, 2020). It is estimated to occur among 10% of non-traumatically bereaved individuals (Lundorff, Holmgren, Zachariae, Farver-Vestergaard, & O’Connor, 2017). This disorder is characterized by a persistent and pervasive longing or a persistent preoccupation with the deceased, accompanied by intense emotional pain following the death of a significant other. In recent decades, several therapies have been evaluated for the treatment of PGD (Boelen, de Keijser, van den Hout, & van den Bout, 2007; Rosner, Rimane, Vogel, Rau, & Hagl, 2018; Shear et al., 2016). Most of these interventions drew on cognitive behavioural therapy (CBT) approaches and included psychoeducation about normal and prolonged grief processes, exposure to internal and external stimuli connected with the most painful aspects of the loss, and cognitive restructuring. These elements were found previously to contribute to the effectiveness of therapies that targeted posttraumatic stress disorder (PTSD), a condition that shares certain common elements with PGD (Haagen, Smid, Knipscheer, & Kleber, 2015).

The symptoms of intrusion in the form of intrusive thoughts and memories related to the loss or the lost person, avoidance, and negative emotional responses following a loss resemble the symptoms of PTSD. Although PGD and PTSD differ, with yearning being a key emotion in PGD and anxiety being more central to PTSD and intrusive memories being hallmark symptoms of PTSD but not PGD; still, however, elevated PGD has been found to be associated with increased experiences of intrusive recollections connected with the circumstances of the loss (Boelen & Huntjens, 2008). In addition, as with PTSD, it has been suggested that impaired integration of the loss memory into the autobiographical memory is central to the development of the disorder (Boelen, van den Hout, & van den Bout, 2006; Ehlers, 2006). The lack of integration may itself lead to a tendency to retrieve general level memories and thus impede the processing of specific aspects of the death and further inhibit the integration of the loss (Maccallum & Bryant, 2013).

Although CBT-based therapy methods have shown efficacy in the treatment of PGD, the effect size was shown to be rather small (Hedge’s g below 0.5, Johannsen et al., 2019), and a significant number of patients do not initiate therapy or do not complete the full range of therapy sessions (Shear et al., 2016). As such, one might conclude that there is a need for new and creative interventions that will enhance existing interventions.

The current study was a pilot study that followed a preliminary case demonstration examining the efficacy of narrative reconstruction (NR) for PGD (Peri, Hasson-Ohayon, Garber, Tuval-Mashiach, & Boelen, 2016). NR is a time-limited integrative intervention that has shown efficacy for PTSD (Gofman et al., 2020) and has been adjusted for PGD patients. NR includes exposure to the loss memory together with meaning-making. The process helps integrate the loss memories with past memories and with personal significance to re-establish a coherent self-narrative (Barbosa, Sá, & Carlos Rocha, 2014; Neimeyer, 2019). The patient and therapist reconstruct together the most disturbing memory associated with the loss or any other significant memory of the deceased that is painfully haunting the patient’s thoughts in an intrusive manner (see Peri et al., 2016). The therapist and the patient reconstruct – along a timeline – a detailed, written, and well-organized narrative of the chosen intrusive memory. The personal meaning of the loss in the patient’s life is revealed in the process, and the memory is then related and actively integrated into the patient’s autobiographical memory. The reconstruction is conducted only in the presence of the therapist. The patients narrate their deeds, their cognitions, and their emotions moment by moment in a very detailed manner, while the therapists type the narrative on a laptop word by word. In each session the narrative is reread to the point reached in the previous session and then updated while the therapist asks for associated memories and a discussion of their meaning. This method provides the patient with more time to process the anxiety-provoking parts of the story, allowing the patient self-focus and emotional involvement. It also slows down the pace of exposure, which helps to improve emotional regulation (for a detailed description see Gofman et al., 2020). Through this process, avoidance is overcome, catastrophic thoughts are given a more realistic evaluation, and emotional control is regained.

In this paper, we report the results of a pilot trial of PGD patients who were treated with NR over the course of 16 sessions. We expected to find a reduction in PGD symptoms as well as in intrusion, avoidance, and depression symptoms following NR. We also expected to find that the change had been retained at the three-month follow-up evaluation.

Methods

Participants

Participants were recruited via the use of community outreach, including advertising. Patients were included in the study if the time since their loss exceeded half a year and their PGD diagnosis was validated by a clinical interview that followed the (at that time) proposed criteria for PGD (Prigerson et al., 2009), as well as a score higher than 30 on the PG-13 (Prigerson, personal communication; In fact, all of our participants had a score above 35, a score that was proposed as clinical cut-off by Pohlkamp, Kreicbergs, Prigerson, & Sveen, 2018). Exclusion criteria were brain injury, psychosis, severe depression with suicidal ideation posing imminent danger, and drug or alcohol abuse as reported by participants in a clinical interview or assessed via the structured Clinical Interview for the DSM-IV-TR (SCID; Blake et al., 1995; Shalev et al., 1997, Hebrew version). Three participants from the final sample (N = 10) continued their antidepressant medication use during therapy.

Twenty-five people were screened for the study. Of them, 12 were excluded because they did not meet the inclusion criteria (PG-13 > 30), two refused to participate in the suggested treatment, and one dropped out of treatment. The final pilot sample comprised 10 bereaved patients (7 women, 3 men). Participants’ ages ranged from 24 to 63 (M = 41.5; SD = 14.3). Time since the loss varied considerably, ranging from one to seven years (M = 2.7, SD = 1.7). Seven patients had lost a parent (six of them as a result of chronic illness, one from a sudden illness), two had lost a child (one as a result of suicide, one from an accident) and one lost a spouse (after a chronic illness).

Measures

All participants were evaluated before the initiation of treatment and after the completion of treatment by an MA-level clinical psychologist masked to treatment condition. The following clinician-administered and self-report measures had been used:

Structured Clinical Interview for DSM-IV Hebrew version (SCID; Shalev, Freedman, Peri, Brandes, & Sahar, 1997)

The SCID was used to assess possible comorbid Axis I Disorders.

Prolonged grief disorder-13 (PG-13; Prigerson et al., 2009)

The PG-13 contains 11 items assessing severity of symptoms of yearning, cognitive difficulties, and emotional difficulties on a 5-point scale ranging from 1 (not at all) to 5 (several times per day). Two additional items are yes/no questions which were not included in the total score. The PG-13 was translated into Hebrew and back-translated into English to evaluate the accuracy of the Hebrew translation (Brislin, 1980). Differences were then reconciled by comparing the original translation and the back-translation. The total severity score, ranging from 11 to 55, was used in the current study. Internal reliability was high in the current study (Cronbach’s α = .88).

Clinician-administered PTSD scale for DSM-IV (CAPS-IV; Blake et al., 1995)

The validated Hebrew CAPS-IV was used to assess intrusion and avoidance symptoms following the loss (Shalev et al., 1997).

Beck depression inventory-II (BDI-II; Beck, Steer, & Brown, 1996)

The validated Hebrew version was used in the study to evaluate depression symptoms (Levav, 2009). The scale’s internal reliability was high (Cronbach’s α = .91)

The Integration of Stressful Life Experiences Scale (ISLES; Holland, Currier, Coleman, & Neimeyer, 2010)

The scale assesses the extent to which survivors of highly distressing life events or losses are able to integrate these events into their overall framework of meaning (Holland et al., 2010) It includes 16 items (e.g. ‘This event is incomprehensible to me’; ‘Since this event happened, I don’t know where to go next in my life’) to which the respondent indicates agreement or disagreement on a 5-point scale. We implemented the validated Hebrew version (Hasson‐Ohayon, Peri, Rotschild, & Tuval‐Mashiach, 2017). The scale’s internal reliability was high in the current study (Cronbach’s α = .91)

Procedure

The study took place at the community clinic of Bar-Ilan University and was approved by its institutional review board (IRB). Participants were recruited through advertisements and links posted on psychology web pages. Participants underwent telephone screening and were invited for a clinical interview and evaluation with self-report tools. Written informed consent was obtained from all participants before baseline assessment. Consecutive candidates who met the inclusion criteria for the study were assigned to receive the NR treatment and went on for further evaluation with self-report tools. Treatments were provided by PhD-level clinical psychology residents who were trained to work with the NR protocol by the authors (GE, TP, or IHO). All clinicians participated in weekly group supervision meetings throughout the treatments. Treatment began within two weeks of the initial evaluation and lasted for 16 weeks (i.e. 16 weekly NR sessions of 60-minutes duration each). Self-report and clinician-based diagnostic measures were administered again after treatment and at a 3-month follow-up. Evaluators conducted their part masked to the participants’ stage in the study. All assessments and treatments took place at the Community clinic at the Psychology Department of Bar-Ilan University.

Therapy

Following psychoeducation about normal and prolonged grief processes and explanations about the rationale underlying the use of NR, patients were directed to choose the memory related to the loss which was repeatedly haunting them (for many of them this was the memory of the moment of the death, but it could also have been other memories of encounters with the deceased). Difficulties in deciding on a single memory were discussed in the first meeting with the therapist to collaboratively identify the specific memory that was the most painful and disturbing for the patient.

During the meetings, patients created a minute-by-minute description of the events that unfolded during the time of the loss memory, by describing what they were doing, seeing, hearing, thinking, and feeling at the time. During the sessions, the therapist transcribed onto the laptop everything that was said by the patient, such that a very detailed written narrative was created. The narrative was reread at each session up to the point that had been reached in the previous session, and then the patient continued from there. Gaps in the narrative were recorded and, as the treatment continued, the therapist returned to these gaps to recover those memories. Gaps were also filled in with missing information that the patient proceeded to recall between sessions. The therapist could ask questions or try to clarify facts and uncover memories associated with those facts. As the story unfolded and patients expressed their thoughts and emotions, the therapist and patient elaborated on the personal meaning of the loss in the context of associated memories that were recalled.

NR enabled significant moments to be identified and revealed dissociated feelings and facts, while actively connecting the memory of the loss to the patient’s memories from earlier in life. At the final therapy session, the therapist provided the patient with a printed copy of the final version of their narrative, which they then reviewed together one last time. The session ended by highlighting the changes attained and connecting them to the patient’s emotional state (for more details, see Peri et al., 2016; therapy protocol will be provided upon request).

Results

Table 1 and Figure 1 present the means and standard deviations of all of the study variables at the three assessment points (i.e. pre-treatment, post-treatment, and 3-month follow-up or T1, T2,T3). One patient (#8) failed to complete the 3-month follow-up assessment; as such, the patient’s post-treatment scores were carried forward. To test our hypotheses, we estimated a series of repeated-measures ANOVAs, followed by post hoc paired t-tests. Effect sizes were estimated using Cohen’s d for a repeated-measures design formula.

Table 1.

Means and standard deviations for all outcome variables at each phase

	Pre-treatment	Post-treatment	Follow-up
1. Prolonged Grief (PG-13)	41.70 (2.95)	28.20 (5.90) ***	24.80 (7.94) ***
2. BDI-II	24.20 (12.58)	19.10 (8.08) *	15.60 (10.54) ***
3. CAPS-Total	42.60 (27.23)	25.60 (15.33)	18.20 (15.75) *
4. CAPS-Intrusion	11.00 (9.36)	5.30 (6.50)	3.50 (4.38) *
5.ISLES	34.60 (9.49)	50.70 (7.53) ***	54.50 (11.87)

Note. P< .05 = *; P < .001***.

Figure 1.

Patients’ measures and means of the variables at pre-treatment (T1), post-treatment (T2), and follow-up (T3)

Changes in PGD symptoms

A one-way repeated-measures ANOVA with time as the independent variable and PG as the dependent variable revealed a significant effect for time, F(2,18) = 50.42, p < .001. Post hoc paired t-tests showed that in comparison to the PG scores at the pre-treatment assessment, the PG scores at the post-treatment assessment (T1 compared to T2), t(9) = 8.62, p< .0001, Cohen’s d= 2.73, and at the follow-up assessment (T3 compared to T1), t(9) = 7.52, p< .0001, Cohen’s d= 2.38, were significantly lower. In addition, the PG scores at the follow-up assessment were significantly lower than at the post-treatment assessment (T3 compared to T2), t(9) = 2.40, p= .004, Cohen’s d= 0.76.

Changes in depressive symptoms

A one-way repeated-measures ANOVA with the BDI-II as the dependent variable revealed a significant effect for time, F(2,18) = 5.47, p = .014. Post hoc paired t-tests showed that in comparison to the BDI-II scores at the pre-treatment assessment, the scores at the follow-up assessment (T1 compared to T3), t(9) = 3.67, p= .005, Cohen’s d= 1.16, were significantly lower. The difference between the BDI-II scores at the pre-treatment assessment (T1) and post-treatment assessment (T1 compared to T2) was not significant, but was medium in size, t(9) = 2.03, p= .073, Cohen’s d= 0.64. Finally, the difference between the BDI-II scores at the post-treatment assessment and the follow-up assessment (T2 compared to T3) was not significant, t(9) = 1.19, p= .27, Cohen’s d= 0.38.

Changes in PTSD-related symptoms: CAPS – total score

A one-way repeated-measures ANOVA with the CAPS total score as the dependent variable revealed a significant effect for time, F(2,18) = 5.05, p = .018. Post hoc paired t-tests showed that in comparison to the CAPS scores at the pre-treatment assessment, the scores at the follow-up assessment (T1 compared to T3), t(9) = 2.78, p= .021, Cohen’s d= 0.88, were significantly lower. No significant difference was found between the CAPS scores at the post-treatment assessment and either the pre-treatment assessment, t(9) = 1.90, p= .090, Cohen’s d= 0.60, or the follow-up assessment, t(9) = 1.38, p= .20, Cohen’s d= 0.435.

Changes in CAPS – intrusion score

A one-way repeated-measures ANOVA with the CAPS intrusion score as the dependent variable revealed a significant effect for time, F(2,18) = 4.17, p = .033. Post hoc paired t-tests showed that in comparison to the CAPS scores at the pre-treatment assessment, the scores at the follow-up assessment (T1 compared to T3), t(9) = 2.79, p= .021, Cohen’s d= 0.88, were significantly lower. No significant difference was found between the CAPS scores at the post-treatment assessment and either the pre-treatment assessment, t(9) = 1.70, p= .12, Cohen’s d= 0.54, or the follow-up assessment, t(9) = 0.95, p= .37, Cohen’s d= 0.30.

Changes in level of integration scale

A one-way repeated-measures ANOVA with the ISLES total score as the dependent variable revealed a significant effect for time, F(2,8) = 27.05, p < .001. Post hoc paired t-tests showed that in comparison to the ISLES scores at the pre-treatment assessment, both the ISLES scores at the post-treatment assessment, t(9) = −7.46, p< .0001, Cohen’s d= 2.36, and at the follow-up assessment (T3 compared to T1), t(9) = −5.10, p< .0001, Cohen’s d= 1.61, were significantly higher. The ISLES scores at the follow-up assessment were not significantly different than at the post-treatment assessment, t(9) = −3.80, p= .15, Cohen’s d= 0.51.

Discussion

Results of this pilot study provide preliminary evidence for the feasibility and potential efficacy of NR in treating PGD. Participants showed significant improvements in PGD symptomatology, depression, and PTSD symptoms following NR, and at the three-month follow-up they showed evidence of having continued to improve. The effect sizes of the reduction in symptoms compared favourably with those reported in a recent meta-analysis on treatments for PGD (Johannsen et al., 2019). For our primary outcome measure (i.e. PGD symptoms; PG-13) we found a large pre-treatment to post-treatment effect size of the treatment (Cohen’s d = 2.73). PGD symptoms continued to decrease significantly at follow-up (Cohen’s d = 0.76, follow-up vs. post-treatment). This effect was encouraging considering the average effects reported in the Johannsen et al. meta-analysis. Similar to the meta-analyses of interventions with PGD, we found at follow-up that the improvement at post-treatment had continued, with further reductions of symptoms from one timepoint to the next (Wittouck, van Autreve, de Jaegere, Portzky, & van Heeringen, 2011). Yet comparisons of our findings with those reported in these meta-analyses should be made with caution until effects of NR relative to other treatments are available from controlled trials.

Our finding of increased integration of the loss memory following NR indicates that NR is effective in enhancing PGD patients’ memory integration and meaning-making. This short communication does not allow us to present a detailed analysis of the relationship between the ISLES and symptom measures. Yet our findings may be considered as showing preliminary support for the notion that NR’s efficacy is related to the integration of the loss memory in the self-narrative.

The reduction in depression symptoms following NR is in line with findings from a recent meta-analysis (Johannsen et al., 2019). This reduction may be related specifically to the working through of the loss, or alternatively, it may be related to a wider enhancement in affect regulation following the intervention. To address this question, further research of the specific mechanisms of change involved is needed.

Observed individual patterns of change (Figure 1) indicated that almost all patients followed a similar pattern of improvement at post-treatment with continuous improvement or no deterioration from post-treatment to follow-up vis a vis all of the measures. Only patient no. 10 showed improvement at post-treatment but then deteriorated and showed an intensification of symptoms at follow-up. This increase in symptomatology of patient no. 10 was accompanied by a reduction in the ISLES, a finding that further supports the possible relationship between grief symptoms and the lack of integration of the loss memory mentioned earlier.

The study’s limitations include its small number of patients and the lack of a control group. In addition, it uses a new translation into Hebrew of the PG-13, a measure that is not fully in line with the current PGD formal diagnosis in the ICD-11 or DSM-5. However, building on the case study mentioned earlier (Peri et al., 2016), the current study provides further evidence that, as a time-limited, structured intervention, NR can successfully alleviate long-lasting emotional pain following loss. Of note, based on the encouraging findings of this pilot trial, a larger trial is currently taking place with the use of a waitlist-controlled study of NR for PGD.