Effectiveness of a Cancer Bereavement Therapeutic Group.

This study examined the effectiveness of a novel cancer bereavement group. Twenty-seven participants attended a sixsession cancer bereavement therapeutic group. Data were collected at baseline, intervention completion, and three-month follow-up. Grief intensity and symptoms of posttraumatic stress disorder (PTSD), depression, and anxiety were reduced postintervention, and self-compassion increased. At follow-up, improvement remained for grief, PTSD, and depression. A small quasi-experimental waiting-list comparison group showed no change on any measure between baseline and waiting-list end. This study provides preliminary evidence that a brief therapeutic group is an effective intervention for cancer bereavement.

The loss of a loved one to cancer can be distinguished from other types of bereavement (MacKinnon et al., 2013). Between cancer diagnosis and cancer-related death, family and friends are often exposed to additional sources of psychological distress, such as prolonged periods of uncertainty regarding prognosis, observing changes in the physical appearance of their loved one (e.g., cachexia), and witnessing traumatic events (e.g., emergency hospital admissions). Although most people adjust well to bereavement by cancer, some experience negative outcomes such as higher levels of depression, anxiety, and sleep difficulties (Jonasson et al., 2009).

Therapeutic bereavement groups specific to a type of loss (e.g., HIV-related death) have been shown to be an effective intervention for bereavement (Sikkema, Hansen, Kochman, Tate, & Difranceisco, 2004). In group therapy generally, homogeneity of group members is associated with increased group cohesion and better outcomes (Yalom & Leszcz, 2005), and when individuals experience the same type of bereavement there may be similar themes or symptoms in their grief reactions (Houck, 2007), which specific therapeutic support groups are well placed to address. However, despite the clinical rationale for cancer-specific therapeutic bereavement groups, and some preliminary evidence that participants find them beneficial (Souter & Moore, 1990), little research has examined their effectiveness.

The present study evaluated a new cancer bereavement therapeutic group intervention. The intervention incorporated aspects of compassion-focused therapy (CFT), cognitive-behavioral therapy (CBT), cognitive therapy for posttraumatic stress disorder (CT-PTSD; Ehlers & Clark, 2000), and selfhelp groups. The present study aimed to test whether the intervention was associated with changes in grief responses, psychological symptoms, and self-compassion, at intervention completion and at three-month follow-up.

Method

Study design

This was a longitudinal cohort study using a pre–post intervention design. Data were collected via online questionnaires at baseline, at intervention completion, and three months after intervention completion. In addition, a small quasi-experimental waiting-list group was used to estimate changes over time in the absence of an intervention.

Participants and procedure

The research was conducted in collaboration with a UK charity (The Loss Foundation) that provides support to adults who have had a cancer-related bereavement, via free, open, professionally facilitated self-help groups and other supportive events.

The eligibility criteria for participants were (a) aged 18 or over, (b) experience of bereavement by cancer, (c) time since bereavement greater than six months, so as not to interfere with a natural recovery process (Schut & Stroebe, 2010), and (d) self-referral to the charity or via a related organization. The exclusion criterion was significant substance or alcohol misuse, which would interfere with ability to participate.

The study was publicized via the charity’s website, social media, and e-mail list. Individuals registered interest using an online form and were then provided with the participant information sheet. Interested individuals received a screening phone call that used a protocol to assess eligibility, explain the study, and answer questions. A total of 58 individuals expressed interest; 57 received a screening phone call and one was not contactable.

Following screening, 49 individuals met the eligibility criteria and provided written informed consent. They were allocated to one of three therapeutic groups, based on their time preferences or time since bereavement in order to meet eligibility criteria. Three groups were run to ensure clinically appropriate group sizes and maximize therapeutic benefits. The groups were identical in content but began at different points in the year. Groups 1 and 2 (the main focus of this study) were allocated 33 participants (17 in the first group and 16 in the second); prior to the intervention, 6 withdrew, leaving a total of 27. Sixteen participants were allocated to group 3; prior to the intervention, 5 withdrew, leaving 11 participants who formed a quasi-experimental waiting-list comparison group.

Power analysis

The study aimed to measure the changes that occurred pre–post intervention. Using G*Power (Faul, Erdfelder, Lang, & Buchner, 2007), it was calculated that to detect a large effect size of d = 0.80 with an alpha of 0.05, a sample size of 15 would be required. To detect a medium effect size of d = 0.50 with an alpha of 0.05, a sample size of 35 would be required. With an alpha of 0.05, the achieved sample of 27 gave a power of 97% to detect a large effect size of d = 0.80, and a power of 68% to detect a medium effect size of d = 0.50.

Ethical approval

The study received ethical approval from University College London’s Research Ethics Committee (project ID: CEHP/2015/530).

Therapeutic group intervention

The intervention, developed by the second and last authors, was a six-session cancer bereavement therapeutic group based on CFT, CBT, and CT-PTSD theories of distress (Beck, 2011; Ehlers & Clark, 2000; Gilbert, 2009). It covered three main components of bereavement support: psychoeducation, development of self-compassion, and grief cognitions. The first component, psychoeducation, is based on evidence that learning about the grief process is beneficial (Goldstein, Alter, & Axelrod, 1996). The second component aimed to develop self-compassion through compassionate mind training. Losing a loved one to cancer may result in a heightened state of threat detection, where anxiety, shame, and self-criticism are experienced. Developing self-compassion can help a person move from a state of threat to developing self-soothing and social safeness systems (Gilbert, 2009). The third component used principles from CBT and CT-PTSD to work with unpleasant memories, negative grief cognitions, and maladaptive coping behaviors. CBT for bereavement is effective in alleviating distress and evidence suggests that it be integrated in other approaches to ensure that the multidimensional nature of grief is considered (Currier, Holland, & Neimeyer, 2010).

Each session had specific content:

Session 1: psychoeducation about the grief experience (e.g., sleep difficulties and anxiety)

Session 2: self-care and daily routine

Session 3: self-compassion

Session 4: unhelpful cognitions

Session 5: developing resilience to unhelpful thoughts or memories through exposure

Session 6: reflections and endings

Sessions lasted two hours and the intervention covered a period of eight weeks. It was delivered by the second, third, fourth, and last authors, all clinical psychologists.

The outline of a sample session is given in the box. The full curriculum is available from the last author (e-mail: erin@thelossfoundation.org).

Measures

Patient-reported outcomes

All measures were collected via secure online questionnaires at baseline, intervention completion and three months after the final session.

The Grief Intensity Scale (GIS: Prigerson & Maciejewski, 2015) is a 12-item self-report questionnaire that measures time from loss, grief intensity, and functional impairment. For the present study the analysis excluded the time since loss and functional impairment items to create a 10-item measure of grief intensity with scores ranging from 0 to 40. Cronbach’s alpha for the intensity items in the present study was 0.84.

The PTSD Checklist for DSM-5 (PCL-5: Weathers et al., 2013) is a 20-item self-report questionnaire that measures symptoms of posttraumatic stress disorder (PTSD). Scores range from 0 to 80. Cronbach’s alpha was 0.86.

The Patient Health Questionniare-9 (PHQ-9: Kroenke, Spitzer, & Williams, 2001) and the Generalized Anxiety Disorder-7 (GAD-7: Spitzer, Kroenke, Williams, & Löwe, 2006) were used to assess symptoms of depression and anxiety respectively. Cronbach’s alpha was 0.83 for the PHQ-9 and 0.90 for the GAD-7.

The Self-Compassion Scale–Short Form (SCS-SF: Raes, Pommier, Neff, & Van Gucht, 2011) is a 12-item self-report questionnaire. Scores range from 12 to 60. Cronbach’s alpha was 0.81.

Sociodemographic and bereavement data

Sociodemographic information including age, gender, ethnicity, marital status, education, and psychological treatment was collected at baseline. Participants were also asked about the characteristics of their bereavement, for example, the length and nature of their relationship with the deceased.

Results

Data screening

There were no missing data at baseline (n = 27); however, postintervention there was 85% data completion (n = 23) and at three-month follow-up 93% data completion (n = 25). Five variables showed a deviation from normality; therefore the data were analyzed using two-tailed Wilcoxon signed-rank tests.

Participant characteristics

Participants were predominantly women (n = 22, 82%) and White British (n = 24, 89%); half were college graduates (n = 14, 51%); ages ranged from 26 to71, with a mean of 49 (SD = 15). Most had lost a spouse or partner (n = 11, 41%), parent (n = 7, 26%), or sibling (n = 6, 22%). The length of their relationship with the deceased ranged from 7 to 69 years (M = 32.5, SD = 12). The mean time since the bereavement was 26 months (SD = 25; range 6 to 103). Most (n = 19, 70%) saw their loved one daily in the three months preceding the death and were present at the death (n = 20, 74%).

Table 1 shows participants’ psychological status over the three time points. Preintervention, the sample had moderate levels of depression, anxiety, and PTSD symptoms; self-compassion scores were lower than in general population samples (Antúnez, Navarro, & Adan, 2015), and grief intensity was lower than that of bereaved parents (Lichtenthal et al., 2015). Two thirds (n = 18, 67%) had previously or currently received psychological treatment postbereavement.

Table 1

Outcome measures at baseline to intervention completion (session 6), and baseline to follow-up (3 months after completion).

	Baseline (n = 27)		Completion (n = 23)		Wilcoxon Signed Ranks			Follow-Up (n = 25)		Wilcoxon Signed Ranks
Measure	M	(SD)	M	(SD)	z	p	Cohen’s d	M	(SD)	z	p	Cohen’s d
GIS	22.56	(7.45)	11.30	(3.62)	4.17	<.001	1.87	17.08	(6.18)	3.70	<.001	0.79
PCL-5	35.70	(12.91)	28.52	(13.44)	2.45	.014	0.54	25.52	(9.46)	4.08	<.001	0.90
PHQ-9	10.85	(5.61)	7.65	(5.01)	3.07	.002	0.60	8.04	(3.50)	2.40	.017	0.60
GAD-7	9.59	(5.40)	7.17	(5.20)	3.13	.002	0.46	7.52	(4.72)	1.76	.079	0.41
SCS-SF	31.63	(7.86)	36.69	(7.29)	2.68	.007	0.67	35.45	(7.33)	1.68	.094	0.50

Note. GIS: The Grief Intensity Scale; PCL-5: The PTSD Checklist for DSM-5; PHQ-9: The Patient Health Questionnaire-9; GAD-7: The Generalized Anxiety Disorder-7; SCS-SF: The Self-Compassion Scale-Short Form.

Pre- and Postintervention comparisons

Attendance at group sessions was high: 74% (n = 20) of participants attended all six sessions, 19% (n = 5) attended three or four, and 7% (n = 2) attended one or two.

Baseline to intervention completion

Following the intervention, scores decreased on symptoms of depression (PHQ-9, z = 3.07, p = .002) and anxiety (GAD-7, z = 3.13, p = .002; Table 1). PHQ-9 scores decreased by 3.2 points, indicating a moderate effect size (d = .60); GAD-7 scores decreased by 2.4 points, indicating a small to moderate effect size (d = .46).

Self-compassion scores on the SCS-SF increased postintervention (z = 2.68, p = .007) by 5.1 points, with a moderate to large effect size (d = .67). Grief scores on the GIS (z = 4.17, p = <.001) and PTSD scores on the PCL-5 (z = 2.45, p = .014) were lower at intervention completion, by 11.3 points on the GIS, with a large effect size (d = 1.87), and by 7.2 points on the PCL-5, with a moderate effect size (d = .54). Intention-to-treat analyses were also conducted using last observation carried forward, yielding substantially the same results.

Baseline to follow-up (3 months after intervention completion)

At follow-up, symptoms of depression remained reduced (z = 2.4, p = .017) compared to baseline: PHQ-9 scores decreased by 2.8 points with a moderate effect size (d = .60; Table 1). Similarly, grief intensity and PTSD symptoms remained reduced. There was a decrease of 5.5 points on the GIS (z = 3.7, p = < .001), with a large effect size (d = .79), and a decrease of 10.2 points on the PCL-5 (z = 4.08, p = < .001), with a large effect size (d = .90). There was no change between pre- and follow-up scores on anxiety (GAD-7, z = 1.76, p = .079) and self-compassion (SCS-SF, z = 1.68, p = .094).

Intention-to-treat analyses conducted using last observation carried forward showed substantially the same results, except for outcomes of self-compassion, on which there was a continued increase between pre- and follow-up scores on the SCS-SF but a smaller effect size (z = 2.02, p = .043, d = .37).

Reliable change analysis

Reliable change (Jacobson & Truax, 1991) in outcomes between baseline and intervention completion and follow-up were calculated using Cronbach’s alpha as the reliability estimate for each measure (Table 2).

Table 2

Reliable change in outcomes at baseline to intervention completion (Session 6), and baseline to follow-up (3 months after completion).

	Baseline–Intervention Completion (n = 23)			Baseline–Follow-Up (n = 25)
Measure	% Improvement (n)	% No change (n)	% Deterioration (n)	% Improvement (n)	% No change (n)	% Deterioration (n)
GIS	70 (16)	30 (7)	0 (0)	36 (9)	64 (16)	0 (0)
PCL-5	22 (5)	74 (17)	4 (1)	28 (7)	72 (18)	0 (0)
PHQ-9	22 (5)	78 (18)	0 (0)	24 (6)	76 (19)	0 (0)
GAD-7	13 (3)	87 (20)	0 (0)	20 (5)	72 (18)	8 (2)
SCS-SF	26 (6)	74 (17)	0 (0)	12 (3)	88 (22)	0 (0)

Note. GIS: The Grief Intensity Scale; PCL-5: The PTSD Checklist for DSM-5; PHQ-9: The Patient Health Questionnaire-9; GAD-7: The Generalized Anxiety Disorder-7; SCS-SF: The Self-Compassion Scale-Short Form.

Between baseline and intervention completion, most participants experienced a reliable improvement in grief intensity (n = 16, 70%), and approximately one quarter experienced reliable improvements in self-compassion (n = 6, 26%), symptoms of PTSD (n = 5, 22%), and depression (n = 5, 22%). One participant experienced a reliable deterioration in PTSD symptoms (and had no reliable change on any other outcome measure).

At follow-up, reliable improvement in symptoms of depression (n = 6, 24%), anxiety (n = 5, 20%), and PTSD (n = 7, 28%) increased compared to outcomes at intervention completion. However, compared to outcomes at intervention completion, reliable improvement decreased for self-compassion (n = 3, 12%) and grief (n = 9, 36%). Two participants (8%) experienced a reliable deterioration in symptoms of anxiety: one of these showed no reliable change on other outcome measures except for a reliable improvement in depression symptoms; the other reliably improved on grief and PTSD symptoms but experienced no reliable change in depression symptoms or self-compassion.

Quasi-experimental comparison group

Data from the 11 participants on the waiting list for the third group were used to estimate changes over time in the absence of the intervention. Participants who were allocated to this group waited for three months before the start of their intervention. They were generally less severely distressed, and thus able to wait longer for the intervention to start. For this reason, their outcome data were not comparable to groups 1 and 2, and group 3 data were used as a waiting-list comparison only.

Participant characteristics

The comparison group had similar sociodemographic characteristics to the intervention group: most were White British women (n = 10, 83%) and well-educated, having continued beyond secondary education (n = 10, 83%); ages ranged from 26 to 54 years, with a mean of 42.6 (SD = 10). The majority had experienced the bereavement of a spouse or partner (n = 7, 58%) or the loss of a parent (n = 4, 33%). The mean length of the relationship with the deceased was 26 years (SD = 12; range 8 to 45). The mean length of time since the bereavement was 16 months (SD = 9.9; range 4 to 42); this was a shorter period compared to participants in the intervention group. Most participants saw their loved one daily in the three months preceding their death (n = 9, 75%) and were present at the death (n = 8, 67%).

Participants in the comparison group had lower levels of depression and anxiety on the PHQ-9 (z = 2.09, p = .036) and GAD-7 (z = 2.87, p = .003) than the intervention group. They also had lower scores of grief intensity on the GIS (z = 2.56, p = .010) and lower levels of PTSD symptoms on the PCL-5 (z = 3.18, p = .001; Table 3).

Table 3

Waiting-list comparison group outcomes at initial assessment to end of waiting list and comparison with intervention group at initial assessment.

	Comparison Group (n = 11)						Intervention Group (n = 27)
	Initial Assessment		End of Waiting-List		Wilcoxon Signed Ranks		Initial Assessment		Wilcoxon Signed Ranks
Measure	M	(SD)	M	(SD)	z	p	M	(SD)	z	p
GIS	19.00	(6.68)	19.45	(6.28)	.060	.952	22.56	(7.45)	2.563	.010
PCL-5	22.83	(8.29)	21.00	(10.29)	.357	.721	35.70	(12.91)	3.182	.001
PHQ-9	7.17	(5.72)	7.45	(5.59)	.854	.393	10.85	(5.61)	2.09	.036
GAD-7	5.00	(4.51)	5.81	(3.97)	.850	.395	9.59	(5.40)	2.87	.003
SCS-SF	33.75	(9.23)	31.00	(6.36)	.479	.632	31.63	(7.86)	.548	.599

Note. GIS: The Grief Intensity Scale; PCL-5: The PTSD Checklist for DSM-5; PHQ-9: The Patient Health Questionnaire-9; GAD-7: The Generalized Anxiety Disorder-7; SCS-SF: The Self-Compassion Scale-Short Form.

Initial assessment to end of waiting-list comparisons

There were no differences in the comparison group scores between initial assessment to end of waiting-list (Table 3). This suggests that changes in outcomes did not occur spontaneously over three months and that improvements identified in the intervention group could be attributed to the intervention and not the passage of time. However, given that the participants in the comparison group started the intervention in less distress than the participants in the intervention group, any inferences must be made with caution.

Discussion

This study used an uncontrolled longitudinal cohort pre–post intervention design to examine the impact of a six-session therapeutic group intervention for cancer-bereaved adults. At intervention completion, grief intensity and symptoms of PTSD, depression, and anxiety were reduced; there was a particularly large change in grief intensity, with most participants experiencing a reliable improvement. Participants’ self-compassion also increased.

At three-month follow-up, the improvements remained for grief, PTSD, and depression with moderate and large effect sizes. However, the reduction in symptoms of anxiety and the increase in self-compassion were not maintained.

A small quasi-experimental waiting-list comparison group showed no change in any of the outcome measures in the three months between initial assessment and the end of the waiting list. This lends some weight to the interpretation that it was the group intervention that led to the improvements.

The findings suggest that, in the short term, the intervention was beneficial. The lack of similar studies into therapeutic groups for cancer bereavement means that the results are best compared with studies where the cause of bereavement was unspecified. They are consistent with studies that have assessed the effects of structured bereavement group interventions, which have found reductions in grief and symptoms of psychological distress (Goodkin et al., 1999) and stress (Kang & Yoo, 2007) at intervention completion.

Longer-term, the psychological improvements remained, but only for grief intensity, symptoms of PTSD, and depression. This finding has been replicated elsewhere. A study by Rheingold et al. (2015) of a manualized 10-session bereavement support group for adults who had lost a loved one to death by violence found that grief, symptoms of PTSD, and depression remained reduced at one-year follow-up. Overlap in the intervention components of relaxation techniques and commemorative imagery of the lost loved one may account for the similar findings. In the current study, large effect sizes for grief intensity were found at intervention completion (d = 1.87) and follow-up (d = 0.79). Although these effect sizes are much greater than those reported in a review of psychotherapeutic bereavement interventions (d = 0.51; Currier, Neimeyer, & Berman, 2008), they are similar to those observed in studies of interventions for prolonged grief. For example, in a study comparing CBT and supportive counseling, a pre–post effect size of d = 1.80 was found for the CBT condition (Boelen, de Keijser, van den Hout, & van den Bout, 2007). Similarly, in an RCT of CBT for prolonged grief (Rosner, Pfoh, Kotoučová, & Hagl, 2014) the effect size for grief symptoms at treatment completion between participants in the experimental and waiting-list conditions was d = 1.61.

Study limitations

The most substantial limitation was that the study lacked a randomized control group; therefore, attribution of change to the intervention must be made with caution. However, a randomized controlled trial (RCT) was not feasible within the logistics of the charity and may have been premature at this stage. MRC guidance (Campbell et al., 2007) suggests that, before RCTs are conducted, preliminary research should develop appropriate interventions and suitable evaluation procedures to provide a firm grounding for any subsequent rigorous trials.

Although the quasi-experimental waiting-list comparison group was valuable, its sample size was small and the participants differed on baseline measures compared to the intervention group. Therefore, this group may not have been appropriate to draw comparisons against, as participants were less anxious and depressed and had lower levels of grief intensity and PTSD symptoms.

The sample size in the intervention arm was also small and participants were typically White, middle-class women. It is therefore difficult to generalize the findings to other populations or to explore differences within the study sample, for example, gender or age differences.

Clinical implications and conclusion

The intervention used psychoeducation and techniques and exercises from CBT, CT-PTSD, and CFT. Without conducting a dismantling study, it is difficult to identify which components were effective. However, an understanding of which aspects of CBT have been found effective in treating prolonged grief may provide some insight. For example, identifying and changing negative beliefs and interpretations (session 4) can disrupt maintenance cycles, increasing behavioral activation and the development of more adaptive beliefs (Boelen, 2006). Exposure to unhelpful thoughts and memories (session 5) has also been identified as effective in facilitating the integration of the loss with existing knowledge (Boelen, 2006).

A possible mediator of the effectiveness of the intervention is the development of self-compassion. Self-compassion has been found to enhance coping and resilience when experiencing life stressors such as divorce (Sbarra, Smith, & Mehl, 2012), childhood trauma (Vettese, Dyer, Li, & Wekerle, 2011), and HIV diagnosis (Kemppainen et al., 2013). The development of self-compassion could increase the emotional resources and adaptive coping of those who have lost a loved one to cancer and who may have experienced the additional challenges of being a caregiver. It also complements the dual processing model of coping with bereavement, where individuals oscillate between confronting and accepting the pain and setting it aside. It may also be that when people are more self-compassionate they experience a decrease in uncompassionate responding (Germer & Neff, 2015).

This intervention was developed by a charity in response to requests from their service users to provide a structured therapeutic group intervention. Uptake was high and attrition was low, suggesting good acceptability of the intervention. Because it was manualized, it could be replicated elsewhere. Although further, more rigorous research needs be conducted before the intervention is potentially scaled up, these initial results indicate that it shows promise for helping people to cope with the life-changing, painful experience of losing a loved one to cancer.