Journal of Affective Disorders 167 (2014) 5663

Contents lists available at ScienceDirect

Journal of Affective Disorders

journal homepage: www.elsevier.com/locate/jad

Research report

Efcacy of an outpatient treatment for prolonged grief

disorder: A randomized controlled clinical trial
Rita Rosner a,b,n, Gabriele Pfoh a, Michaela Kotouov a,1, Maria Hagl a,1
a
b

Department of Psychology, Ludwig-Maximilian-University of Munich, Germany

Catholic University Eichsttt-Ingolstadt, Ostenstr. 25, D-85071 Eichsttt, Germany

art ic l e i nf o

a b s t r a c t

Article history:
Received 26 March 2014
Received in revised form
20 May 2014
Accepted 22 May 2014
Available online 2 June 2014

Background: Abnormal forms of grief, currently referred to as complicated grief or prolonged grief
disorder, have been discussed extensively in recent years. While the diagnostic criteria are still debated,
there is no doubt that prolonged grief is disabling and may require treatment. To date, few interventions
have demonstrated efcacy.
Methods: We investigated whether outpatients suffering from prolonged grief disorder (PGD) benet
from a newly developed integrative cognitive behavioural therapy for prolonged grief (PG-CBT). A total of
51 patients were randomized into two groups, stratied by the type of death and their relationship to the
deceased; 24 patients composed the treatment group and 27 patients composed the wait list control
group (WG). Treatment consisted of 2025 sessions. Main outcome was change in grief severity;
secondary outcomes were reductions in general psychological distress and in comorbidity.
Results: Patients on average had 2.5 comorbid diagnoses in addition to PGD. Between group effect sizes
were large for the improvement of grief symptoms in treatment completers (Cohen's d1.61) and in the
intent-to-treat analysis (d 1.32). Comorbid depressive symptoms also improved in PG-CBT compared to
WG. The completion rate was 79% in PG-CBT and 89% in WG.
Limitations: The major limitations of this study were a small sample size and that PG-CBT took longer
than the waiting time.
Conclusions: PG-CBT was found to be effective with an acceptable dropout rate. Given the number of
bereaved people who suffer from PGD, the results are of high clinical relevance.
& 2014 Elsevier B.V. All rights reserved.

Keywords:
Bereavement
Prolonged grief disorder
Complicated grief
Persistent complex bereavement-related
disorder
Randomized controlled trial
Psychotherapy

1. Introduction
Prolonged grief disorder is being proposed as part of the stress
related disorders category in the ICD-11; in the DSM-5, persistent
complex bereavement disorder is classied as a condition for
further study (American Psychiatric Association, 2013). Despite
this discrepancy, many papers on severely impairing forms of
abnormal grief have been published in the last few decades. The
term complicated grief was coined by Horowitz and his colleagues (Horowitz et al., 1997; Prigerson et al., 1995), while traumatic grief was used by Prigerson and her colleagues (Jacobs,
1999; Prigerson et al., 1997; Shear et al., 2001). As the concepts
evolved, the criteria and denitions for grief varied (for an overview and a comparison of the terminology, see Shear et al. (2011)).
However, core symptoms overlap between denitions: for example, intense yearning and preoccupation with the loss, reactive
n
Corresponding author at: Catholic University Eichstaett-Ingolstadt, Ostenstr. 25,
D-85071 Eichsttt, Germany. Tel.: 49 8421 93 1581; fax: 49 8421 93 2033.
E-mail address: rita.rosner@ku.de (R. Rosner).
1
MK and MH are no longer afliated with the Ludwig-Maximilian-University

http://dx.doi.org/10.1016/j.jad.2014.05.035
0165-0327/& 2014 Elsevier B.V. All rights reserved.

distress symptoms, such as avoidance of memories of the deceased

person and emotional numbing, as well as social/identity disruption, such as feeling detached or having difculties trusting others.
In this manuscript, we are referring to the respective grief
condition as prolonged grief disorder (PGD). Estimates regarding
the prevalence of PGD vary between studies. While a study from
the USA has reported that approximately 10% of the bereaved
show symptoms that cause impairment in everyday life (Bonanno
et al., 2008; Ott, 2003), European studies have reported prevalence
values of 4.2% in a sample of older Swiss (Maercker et al., 2008)
and 4% in a German sample (Kersting et al., 2011). In the Netherlands (Newson et al., 2011), a prevalence of 4.8% was found in a
sample of 5741 adults aged 55 years and older. For those
participants who had experienced a loss, the prevalence of PGD
was 25.4%.
PGD has been found to be associated with deteriorated health
(Stroebe et al., 2007), increased depression, and suicidality (Boelen
and Prigerson, 2007; Latham and Prigerson, 2004). Having a
diagnosis of PGD six months after a loss correlated with an
increased risk for heart disease, high blood pressure, cancer, and
altered eating habits (Prigerson et al., 2008). Furthermore, PGD has

R. Rosner et al. / Journal of Affective Disorders 167 (2014) 5663

been effectively differentiated from depression and posttraumatic

stress disorder (PTSD) (Boelen and van den Bout, 2005; Boelen et
al., 2008). However, PGD is often comorbid with these disorders
(Maercker and Znoj, 2010; Simon et al., 2007).
Despite discussions about the precise criteria for PGD, there is
an impressive amount of literature on treatment outcomes for
bereavement associated problems. In general, meta-analyses evaluating the efcacy of treatments for grief give small (Kato and
Mann, 1999: d 0.11; Fortner, 2000: d 0.13; Rosner et al., 2005:
d 0.20; Currier et al., 2008: d 0.16) to medium (Allumbaugh and
Hoyt, 1999: d 0.43) effect size (ES) calculations at best. However,
studies including only patients with severe grief symptoms
showed evidence of larger effect sizes than studies with subjects
that did not have substantial grief symptoms: Currier et al. (2008)
reported an ES of d 0.51, and Rosner et al. (2005) reported an ES
of d 0.27. The most recent meta-analysis compared preventive
and treatment interventions for PGD and found an ES of 0.03 for
prevention and 0.53 for treatment interventions (Wittouck et al.,
2011).
Studies based exclusively on patients meeting the criteria for
PGD are still rare. There are three successful randomized controlled trials investigating individual treatment for PGD in terms of
overall ES: (1) a trial with two active conditions comparing
complicated grief treatment with interpersonal therapy (Shear et
al., 2005); (2) a trial comparing an internet-based intervention
with an untreated control (Wagner et al., 2006); and (3) a trial
comparing cognitive behavioural therapy (CBT) with supportive
counselling (Boelen et al., 2007). Because two of the studies
compared active treatments (Boelen et al., 2007; Shear et al.,
2005), it is reasonable to use the pre- to post-treatment effect
sizes to compare the respective treatment protocols. Boelen et al.
(2007) reported an ES of 1.36 for the combination of cognitive
restructuring followed by exposure and an ES of 1.80 for the
combination of exposure followed by cognitive restructuring.
Shear et al. (2005) reported an ES of 1.63, and Wagner et al.
(2006) reported an ES of 1.41 (the combined result for posttraumatic symptoms and symptoms of dysfunctional adaptation to
grief). All reported ES values are based on completer analyses.
In all treatment protocols, patients with severe comorbid
disorders were excluded. Boelen et al. (2007) excluded severely
depressed patients, as well as patients with substance use disorders, and provided no further information on additional diagnoses. Shear et al. (2005) excluded patients with substance abuse,
psychosis, and bipolar disorder. However, 45% of patients in their
complicated grief treatment group met criteria for a current
depressive episode, and 49% of patients met criteria for PTSD. An
effectiveness study based on patients seeking inpatient treatment
showed an even wider range and higher number of comorbid
disorders (Rosner et al., 2011b). Therefore, under clinical conditions, a high number of comorbid diagnoses should be expected.
The reduction of comorbid symptoms has not yet been covered in
depth. In regards to the studies mentioned above, Boelen et al.
(2007) reported an ES of 1.18 measuring overall psychological
distress with the Dutch version of the SCL-90-R (Derogatis, 1977)
for the cognitive restructuring followed by exposure condition and
an ES of 1.15 for exposure followed by cognitive restructuring.
Shear et al. (2005) reported an ES of 1.22 for depression and 0.82
for anxiety.
To develop a successful intervention, we reviewed studies that
reported positive outcomes regarding their specic therapeutic
interventions. We also performed a meta-analysis on therapeutic
interventions by correlating them with symptom severity (Rosner
et al., 2005; Rosner and Hagl, 2007) to estimate the contribution of
PGD status to outcome. The most promising treatment strategies
were the following: psycho-education about normal and prolonged grief processes, exposure elements relating to the most

57

painful aspects of the loss, and transformation of the loss to enable

change. The study by Boelen et al. (2007) was not published at the
time our manual was developed. Furthermore, we identied other
promising interventions in our literature review, such as grief
resolution in a publication by Melges and DeMaso (1980) and
Rando's (1993) description of Gestalt and psychodrama interventions. We decided to use exposure and cognitive interventions
similar to those described in two PTSD interventions: Ehlers's
manual on the treatment of PTSD in adults (Ehlers, 1999) and
Cohen and coworkers' manual (2006) on the treatment of PTSD
and grief in children. We included these elements in our newly
developed intervention for inpatients. The resulting structure and
selected interventions were then adapted for different settings. An
inpatient group treatment based on our manual showed a large
pre- to post-treatment ES of 1.21 for inpatients with comorbid
complicated grief (Rosner et al., 2011b).
Hence, the primary goal of this study was to evaluate the
efcacy of a specic individual outpatient treatment manual for
PGD, named integrative cognitive behaviour therapy for prolonged
grief (PG-CBT), in terms of improving patients' grief severity
compared to a wait list control group. Secondary goals were to
test whether PG-CBT is more effective in improving general
distress symptoms and comorbid symptoms compared with a
wait list control group.

2. Methods
2.1. Procedure
The study was approved by the university's ethics committee
and has been registered with Clinical Trials (Identier:
NCT01433653). Pilot patients were seen in 2005. Randomization
began in October 2006. The last patient nished therapy in June
2011. The study design is a stratied randomized controlled trial.
We stratied our sample according to the patient's relationship to
the deceased, namely, a child or other form of kinship, and
according to the type of death, namely, a natural or non-natural
death. A stratied randomization list was electronically produced
and provided by the university's Department of Statistics; then it
was transferred into four groups of envelopes (according to type of
death  type of kinship) that contained group allocation. Allocation was not disclosed to patients or project workers before the
end of baseline assessment.
The control condition was a wait list. The waiting period was
set at four months or longer. This shortest possible waiting period
of four months was chosen for ethical reasons to avoid unnecessary suffering on the patients' part as well as for practical purposes
to ensure treatment adherence. Once a month during the waiting
period, patients met with the diagnostician, who had assessed the
patients' baseline, for ethical and safety reasons. These interim
sessions did not include any treatment; rather, these sessions
consisted of informal safety check-ups, such as inquiring about
possible intentions of self-injurious behaviour or suicidal ideations. None of the participants had to be excluded from wait list
because of respective clinical reasons. Treatment was offered as
soon as a therapist was available, but not before the minimum
waiting period of four months. On average, patients in the wait list
control group were re-assessed by their respective diagnostician
after six months of waiting (M 6.04; SD 1.36). Patients in the
treatment group were assessed at baseline by a diagnostician,
while later assessment was conducted by their respective therapist. During the treatment, but not during the wait list period,
symptom questionnaires (PG-13 and SCL-90-R, see Measures
section below) were administered three times (along with additional process measures): (1) between sessions 5 and 7,

58

R. Rosner et al. / Journal of Affective Disorders 167 (2014) 5663

(2) between sessions 10 and 12, and (3) between sessions 15

and 17.
2.2. Participants
Participants were recruited through word-of-mouth, from
referrals after personal meetings with other health care professionals or clergy, after presentations and letters to institutions and
counselling centres, as well as directly via newspaper articles,
magazine articles, and radio interviews. Prospective patients called
the outpatient clinic at the Department of Psychology (LudwigMaximilian-University of Munich, Germany) and were preliminarily assigned to the study diagnostician if they mentioned that
they were calling because of a loss or if they answered positively to
the intake question: Did you lose a loved one, and is the loss still
disturbing you? A study diagnostician then invited patients to an
assessment. Patients had to meet the following inclusion criteria:
age 18 years or older, uent German language skills and literacy,
reasonable distance between the caller's residence/workplace and
the outpatient clinic (one hour maximum travelling time), and a
clinical diagnosis of PGD, that is the prevalence of clinically
signicant grief symptoms related to a death that occurred six
months prior at least. Clinical signicance was judged by the
diagnostician's impression concerning distress and impairment
and in addition to information obtained from an interview on grief
symptoms (see Measures section). Exclusion criteria were acute

psychosis, severe substance dependence, suicidality, unstable use

of psychotropic drugs (i.e., medication had to be stable for at least
6 weeks), and concurrent psychotherapeutic treatment.
Altogether, 107 people called the outpatient clinic during the
designated trial period between 2006 and 2010. A total of 56
persons were not enroled, 20 of them because they rejected the
offer for treatment and 36 because they did not meet criteria; 27
of those who did not meet the study criteria did not receive a
clinical diagnosis of PGD (see Fig. 1 for participant ow and further
reasons for exclusion). As a result, 51 persons were randomized
into the study (intent-to-treat sample; ITT); 24 were randomized
into PG-CBT and 27 into the wait list (WG).
A detailed description of the participants can be found in
Table 1. On average, participants were middle aged, ranging from
20 to 78 years. Most were female (86%) and lived without a
partner (67%). A comparison of treatment and wait list group
yielded signicant differences concerning only age and time since
loss (p o0.05). The WG patients were signicantly older than
patients in PG-CBT, t(49)  2.15, p 0.037. The average time since
loss was marginally longer for participants in PG-CBT than in WG,
t(27.41) 2.03, p 0.052 (unequal variances assumed) which is
probably due to the wide range in PG-CBT (from a half year to 37
years). Most participants (86%) fullled diagnostic criteria for at
least one other disorder in addition to PGD when assessed for
symptoms within the last twelve months. On average, patients had
2.5 comorbid disorders (range 07), determined by a standardized

Assessed for eligibility (n = 107)

Excluded (n = 56)
Not meeting inclusion criteria (n = 36)
- No PGD diagnosis (n = 27)
- Insufficient language skills (n = 3)
- Currently in psychotherapy (n = 2)
- Acute substance dependence (n = 3)
- Acute psychosis (n = 1)
Declined to participate (n = 20)

Enrollment

Randomized (n= 51)

Allocation
Allocated to PG-CBT (n = 24)
Started treatment (n = 22)

Started/completed wait list (n= 24)

Did not begin treatment (n = 2)

Dropped out after allocation (n= 3)

Allocated to wait list (n = 27)

Post-treatment
assessment
Lost to post assessment (n = 1)

Discontinued intervention (because of

relocation and other reasons) (n = 3)

Discontinued waiting period (n = 0)

Analysis
Analyzed in ITT sample (n = 24)
Excluded from ITT-analysis (n = 0)

Excluded from ITT-analysis (n = 0)

Analyzed in ITT sample (n = 27)

Analyzed in completer sample (n = 19)

Analyzed in completer sample (n = 23)

Fig. 1. Participant ow. Note. PG-CBT integrative cognitive behavioural therapy for prolonged grief; ITT intent-to-treat.

R. Rosner et al. / Journal of Affective Disorders 167 (2014) 5663

clinical interview. In this study, we report only the disorders that

are often associated with PGD: major depressive disorder, PTSD,
generalized anxiety disorder, panic disorder, and two somatoform
disorders (somatization disorder and pain disorder) (see Table 1).
2.3. Measures
The primary outcome measure was the severity score of the
Interview for Prolonged Grief-13 (PG-13; Prigerson and
Maciejewski, n.d.; Prigerson et al., 2009), which was obtained by
calculating the sum of the 11 symptom item scores (range: 1155
points). The 13 items of the PG-13 roughly correspond to the
consensus criteria for PGD proposed by Prigerson et al. (2009):
criterion B (separation distress), criterion C (cognitive, emotional,
and behavioural symptoms), criterion D (duration), and criterion E
(impairment). However, criterion F (medical exclusion) and criterion G (relation to other mental disorders) were not explored by the
instrument. The German version of the PG-13 was translated by
(Pfoh 2007; Rosner et al., in press). The PG-13's psychometric
evaluation showed adequate reliability in the literature; a Cronbach's alpha value of.76 was reported in a study by Schaal et al.
(2009). However, the PG-13 was less reliable in our sample, with
0.53 (N 51), which might be explained by the fact that we
studied an especially homogenous sample with rather severe
symptoms. However, the post-assessment reliability was adequate, with 0.89 (n 42). For the three assessment points
during treatment, a self-report version of the PG-13 was administered (Prigerson and Maciejewski, n.d.).
Improvement in general psychopathology was assessed using
the Symptom Checklist-90-Revised (SCL-90-R; Derogatis, 1977;
German version: Franke, 2002), a widely used self-report questionnaire measuring psychological distress during the last seven
days. In addition to the SCL-90-R's nine subscales, a global severity
index (GSI) can be calculated. The GSI measures the intensity of
global mental health distress and was used as a secondary outcome measure, along with the subscale scores covering depression, anxiety, phobic anxiety, and somatization. These subscales
were selected because they correspond to comorbid diagnoses
frequently found in our sample. The SCL-90-R was administered
pre-treatment, post-treatment, and three times during treatment.
To assess comorbidity at baseline, the 12-month prevalence
rates of disorders commonly associated with PGD were assessed
using the respective sections of the Computerized Diagnostic
Interview for DSM-IV (DIA-X; Wittchen and Pster, 1997). The
DIA-X is a standardized interview used to obtain DSM-IV diagnoses. The respective sections of the DIA-X were repeated after the
treatment or waiting period. Further measures were also used for
explorative reasons (e.g., posttraumatic growth) and as process
measures (e.g., concerning therapeutic alliance); however, they
will be covered elsewhere.
2.4. Intervention
As described above, PG-CBT is based on the results of a
literature review. PG-CBT has been adapted to the standard health
insurance coverage for outpatient mental health treatments in
Germany, which allows for 25 sessions. Of those 25 sessions, ve
are optional and are directed towards special situations or critical
occasions, such as anniversaries, holidays, family sessions, or legal
proceedings/court days. The remaining 20 sessions compose the
standard treatment for all patients and are divided into three
parts. Part one consists of seven sessions that focus on stabilizing
and motivating the patient as well as on further exploration of the
patients' grief situation; part two consists of nine sessions, rst
teaching relaxation, then focusing on confrontation and reinterpretation of the cognitions and perceptions related to patients

59

themselves, their deceased loved ones and the circumstances of

death; part three consists of four sessions, focusing on future
prospects while maintaining a healthy bond to the deceased.
Sessions were prescribed weekly lasting 50 min each, with the
exception of two 90 min sessions. Therefore, the treatment can be
completed within ve to six months; however, in most cases, the
duration of treatment was nine months (mode value) or longer,
M11.53 (SD 2.89). Treatment frequency was adapted to the
personal circumstances of the participants; the usual reasons for
treatment prolongation were illness, vacation, or new losses. As a
result, the time between pre- and post-treatment assessment
differed signicantly between PB-CBT and WG, t(24.54) 7.59,
po 0.001.
Therapists were two master's degree level psychologists (G.P.
and M.K.); these therapists were trained in the application of the
manual by an experienced CBT psychotherapist (R.R.), who also
had seen some of the pilot patients. During the pilot phase, the
manual was rened, adapted, and made available as a written
protocol afterwards. Therapists received biweekly supervision to
ensure treatment delity. Most therapy sessions were recorded on
audiotape (depending on patients' permission), and selected
recordings were discussed during supervision. Both therapists
were female, had an average age of 40 years, and worked as
therapists for an average of 10 years. At that time they were not
yet certied according to German regulations. A short description
of the treatment can be found in Rosner et al. (2011a).
2.5. Data analyses
Possible differences concerning demographic data and baseline
scores between the conditions, as well as between treatment
completers and non-completers, were examined with chi-square
tests and independent t-tests. We performed both intent-to-treat
(ITT) and completer analyses. For the ITT analyses, we carried the
last available observation forward in cases of missing data. When
analyzing treatment length, the actual length between the pretreatment assessment and the last observation was used (i.e., a
score of 0 was obtained in cases where a missing post-treatment
score was substituted with the pre-treatment score because of
immediate dropout). Between-group changes were analyzed using
univariate ANCOVAs, with pre-treatment scores as covariate.
Within-group changes were analyzed with paired-sample t-tests.
Effect sizes for within- and between-group changes were calculated using Cohen's d. Partial eta-squared (2) for each ANCOVA
was also calculated, reecting the portion of variance explained
by the treatment factor after controlling for pre-treatment scores.
For 2, results larger than 0.14 are considered large effects.
Clinical signicance was evaluated following Jacobson and
Truax (1991), combining their cut-off A with the reliable change
index (RCI). To calculate the RCI, the Cronbach's alpha value from
baseline was used. Following that denition, subjects who reliably
improved and were over the calculated cut-off for signicant
change at post-treatment (i.e., an improvement more than two
standard deviations above the average score at pre-treatment)
were considered recovered.

3. Results
3.1. Completion and baseline differences
Three out of the 27 participants dropped out of WG (11%)
immediately after allocation; another participant completed the
waiting time, but parts of the data at post-treatment assessment
are missing. Five out of the 24 allocated participants in PG-CBT
dropped out during treatment (21%): two did not start treatment,

60

R. Rosner et al. / Journal of Affective Disorders 167 (2014) 5663

Table 1
Demographic and clinical characteristics of the intent-to-treat sample.
Total sample
(N 51)
Female gender
Age in years, M (SD)

44 (86%)
47.53 (14.72)

PG-CBT
(n 24)
20 (83%)
43.0 (13.0)

Table 2
Intent-to-treat analyses: means (SD) at pre- and post-treatment for PG-CBT (n 24)
and wait list (n 27), controlled comparison.

WG
(n27)
24 (89%)
51.56
(15.22)

Marital status
Single
Married/partnered
Divorced/separated
Widowed

17 (33%)
17 (33%)
7 (14%)
10 (20%)

10 (42%)
8 (33%)
4 (17%)
2 (8%)

7 (26%)
9 (33%)
3 (11%)
8 (30%)

Type of relationship to deceased

Child
Other

15 (29%)
36 (71%)

7 (29%)
17 (71%)

8 (30%)
19 (70%)

31 (61%)
20 (39%)
4.93 (7.08)

17 (71%)
7 (29%)
7.10 (9.43)

14 (52%)
13 (48%)
3.01
(3.10)

Type of death
Natural
Non-natural
Years since loss, M (SD)
Comorbid disorders a
At least one other disorder
Number of comorbid disorders,
M (SD)
Major depressive disorder
Posttraumatic stress disorder
Generalized anxiety disorder
Panic disorder
Somatoform disorders

86%
2.52 (1.73)
31 (62%)
11 (22%)
7 (14%)
12 (24%)
27 (54%)

83%
2.54 (1.89)
15
6
5
8
12

(62%)
(25%)
(21%)
(33%)
(50%)

88%
2.50
(1.61)
16 (62%)
5 (19%)
2 (8%)
4 (15%)
15 (58%)

Note. Comorbid disorders are 12-months prevalences. PG-CBT Integrative cognitive behavioural therapy for prolonged grief; WG wait list group;
a
n 50 in the total sample and n 26 in WG (one person in WG was not
assessed formally for comorbid disorders).

one completed only one session, and two dropped out after seven
respectively nine sessions. Two patients did not complete posttreatment assessment, but were still considered completers: one
patient nished after 18 treatment sessions, and the other completed the full 25 sessions. Of the other treatment completers, one
patient nished after 22 sessions, and one nished after 24; all
others received 25 sessions. Therefore, on average, treatment
completers received 24.42 sessions (SD 1.71). Treatment completers and drop-outs were compared on all variables reported in
Table 1 as well as on the baseline data of the PG-13 and SCL-90-R.
The 19 completers and 5 drop-outs differed in only one variable;
much more time had passed between the loss and study intake in
completers (M8.67 years, SD 10.06) than in drop outs (M 1.17
years; SD 0.85), t(18.93)  3.21, p 0.005 (unequal variances
assumed). However, this might have been due to a large range in
completers (6 months to 36 years vs. 6 months to 2.5 years in
drop outs).
Concerning outcome measures there were no differences in
baseline scores according to group neither in the original ITT
sample nor the completer sample.
3.2. Main outcomes
For the ITT analysis, the univariate ANCOVA yielded a signicant and large effect (d 1.32) of PG-CBT in comparison with WG
on the primary outcome measure, grief assessed with the PG-13
(see Table 2). The same is true for the completer analysis (d 1.61,
see Table 3). In addition to the results provided in Tables 2 and 3,
we computed simple pre- to post-treatment comparisons. In PGCBT, results were highly signicant and showed large effects,
t(23) 5.22, p o0.001, d 1.26, for the ITT analysis and t(18)
6.35, p o0.001, d 1.65, for the completer analysis. PG-13 scores in
WG did not change signicantly, t(26) 0.65, p 0.520, d 0.10, for

PG-13
Pre
Post

PG-CBT

WG

Group time

Between
group ES

M (SD)

M (SD))

23.60

0.33

1.32

2.84

0.06

0.33

0.23

0.01

0.09

7.79nn

0.14

0.73

1.32

0.03

0.23

1.43

0.03

0.46

nnn

41.17 (5.75)
30.71 (10.19)

42.63 (5.05)
42.07 (6.58)

SCL-90-R
Global severity
index
Pre
Post

1.24 (0.60)
0.86 (0.58)

1.23 (0.77)
1.08 (0.75)

Somatization
Pre
Post

1.26 (0.75)
0.93 (0.73)

1.24 (0.97)
1.00 (0.86)

Depression
Pre
Post

1.68 (0.85)
1.18 (0.76)

1.89 (0.85)
1.78 (0.88)

Anxiety
Pre
Post

1.16 (0.69)
0.78 (0.77)

1.15 (1.00)
0.96 (0.78)

Phobic anxiety
Pre
Post

0.65 (0.70)
0.39 (0.52)

0.91 (0.96)
0.75 (0.97)

Note. PG-CBT Integrative cognitive behavioural therapy for prolonged grief;

WG wait list group; PG-13 Interview for Prolonged Grief-13; SCL-90-R
Symptom Checklist-90-Revised.
nn

p o 0.01.
po 0.001.

nnn

the ITT analysis and t(22) 0.65, p 0.522, d 0.12, for the completer analysis.

3.3. Secondary outcomes

Concerning the secondary outcome measures, the SCL-90-R
(GSI and subscales), the improvements pre- to post-treatment
were less pronounced. In the ITT analysis, global mental health
distress as assessed by the GSI still improved signicantly with
PG-CBT, t(23) 3.50, p 0.002, d 0.64, while no improvement
was seen in WG, t(26) 1.57, p 0.129, d 0.20. However, the
univariate ANCOVA showed no signicant effect (Table 2). The
same holds true for completers (Table 3). The within-group
Cohen's d was nearly the same as in the ITT analysis, t(17) 3.36,
p 0.004, d 0.68. Again, the pre- to post-treatment comparison in
WG was not signicant, with a small effect, t(22) 1.25, p 0.226,
d 0.17.
To get a closer look at what types of symptoms, other than
grief, improved for PG-CBT patients, we tested four SCL-90-R
subscales (somatization, depression, anxiety, and phobic anxiety)
in a series of univariate ANCOVAs. In both the ITT and completer
analyses, depression improved signicantly more in PG-CBT compared with WG, with moderate to large effects (see Tables 2 and
3). With the exception of phobic anxiety, within-group changes in
PG-CBT were signicant in the ITT analysis, at least with an alpha
level of po 0.05 (tests not shown). In the completer analysis, only
somatization and depression improved signicantly. Pre- to posttreatment effect sizes (Cohen's d) were all within the moderate
range (0.42 to 0.68). None of these subscales improved signicantly in the WG. Because four statistical tests were performed for
each of the comparisons (controlled vs. uncontrolled; ITT vs.
completer), alpha ination should be considered. Bonferroni

R. Rosner et al. / Journal of Affective Disorders 167 (2014) 5663

61

correction would result in setting the alpha level at p 0.013. Even

in this case, the effect is still seen for depressive symptoms.

for patients with strong therapy indications (Currier et al., 2008:

d 0.51; Rosner et al., 2005: d 0.27; Wittouck et al., 2011:
d 0.53), our intervention seems to be considerably more effective
than grief interventions in general. Furthermore, the results are
similar to other effective treatment protocols for PGD. As control
conditions did vary substantially between trials, we focus on preto post-treatment ES when comparing our study with other
randomized controlled trials. Regarding grief symptom scores,
Boelen et al. (2007) reported a pre- to post-treatment ES of 1.36
for the combination of cognitive restructuring followed by exposure and an ES of 1.80 for the combination of exposure followed by
cognitive restructuring. Shear et al. (2005) reported a pre- to posttreatment ES of 1.63, and Wagner et al. (2006) reported an ES of
1.41 (when results for posttraumatic symptoms and symptoms of
dysfunctional adaptation to grief were combined). With a pre- to
post-treatment ES of d 1.65 and a 42% clinically signicant
improvement in terms of the conservative two-step criterion of
Jacobson and Truax (1991), PG-CBT compares well to other
treatments for PGD. However, these studies achieved their results
in less time and with a smaller treatment dosage. Considering that
Jacobson and Truax's reliable change index requires a larger
improvement when the outcome instrument has low reliability
(as is the case with baseline scores of the PG-13), our results for
clinical signicant improvements are conservative indeed and
would have proted from using a more reliable measure.
Regarding self-reported comorbid symptoms, the pre- to posttreatment ES values for general mental health distress were
moderate and the controlled comparison was signicant only for
depressive symptoms. Overall, when comparing the ES values for
depression and anxiety with the results from Boelen et al. (2007),
improvements noted in our study seem less pronounced. However,
in our sample, variances in self-reported comorbid symptoms
were rather large, meaning that there were some patients who
had very low symptom scores initially, especially concerning
phobic anxiety. In some cases, there might not have been much
room for improvement.
In terms of completion rates, the results in our study are good,
with 79% completion in PG-CBT versus 73% in the study by Shear
et al. (2005) and 71% in the study by Boelen et al. (2007).

3.4. Clinically relevant improvement

4.1. Strengths and limitations

In terms of clinically relevant improvement in symptoms of

prolonged grief, the cut-off for clinically signicant improvement
at post-treatment assessment was set at a score of 31 on the PG13. Following Jacobson and Truax's (1991) two-step criterion, 42%
of PG-CBT completers (n 8) were considered to be recovered (i.e.,
they were below the cut-off and showed a statistically reliable
improvement) versus only 4% (n 1) in WG, 2 (1) 8.81, p 0.003.
In terms of reliable change, 53% (n 10) of the PG-CBT completers
improved versus only 4% (n 1) in WG, 2 (1)12.55, p o0.001.
Two participants showed a reliable improvement but did not meet
the cut-off value (i.e., they had a score above 31 at post-treatment
assessment). Likewise, there were two other participants who met
the cut-off criterion at post-treatment assessment, but had begun
with a rather low score, so that they could not achieve reliable
improvement. Participants did not deteriorate reliably in either the
PG-CBT group or WG.

The main strength of this trial is that it followed a randomized

controlled design with stratication according to two important
variables (relationship to the deceased and type of death).
Furthermore, it was conducted in a naturalistic clinical setting,
as the university's outpatient clinic is part of the local mental
health system. Study participants experienced conditions comparable to regular outpatient treatment, with the exception of
possibly being randomized into a wait list and a more thorough
diagnostic process. The study was not funded and therefore
followed the clinical practice conditions in Germany. Exclusion
criteria were not overly strict and followed clinical considerations:
acute psychosis, severe substance abuse, and/or suicidality would
preclude outpatient therapy with exposure elements in any case.
Apart from those conditions, patients showed high levels of
comorbidity according to a structured clinical interview. These
strengths regarding the generalizability to clinical practice are,
however, met with several weaknesses that limit the interpretation of the results. First and foremost, time between pre- and posttreatment assessments in the treatment group was nearly twice as
long compared to the wait list group. For ethical reasons, as well as
to prevent treatment drop-out, the waiting period had been set to
four months; however, the PG-CBT protocol could not be completed in less than ve or six months. In reality, PG-CBT often took
longer, nearly a year on average, while the waiting period lasted

Table 3
Completer analyses: Means (SD) at pre- and post-treatment for PG-CBT (n 19) and
waitlist (n 23), controlled comparison.

PG-13
Pre
Post
SCL-90-R

PG-CBT

WG

Group  time Between group

ES

M (SD)

M (SD)

0.47

1.61

3.63

0.09

0.33

0.42

0.01

0.13

7.44nn

0.16

0.60

1.26

0.03

0.29

1.06

0.03

0.33

34.03
41.21 (6.03)
28.05 (9.51)

nnn

41.78 (4.63)
41.13 (6.42)

Global
severity
index
Pre
Post

1.28 (0.64)
0.85 (0.63)

1.21 (0.81)
1.08 (0.75)

Somatization
Pre
Post

1.32 (0.74)
0.90 (0.71)

1.29 (1.02)
1.01 (0.91)

Depression
Pre
Post

1.74 (0.89)
1.15 (0.84)

1.80 (0.83)
1.66 (0.86)

Anxiety
Pre
Post

1.16 (0.71)
0.76 (0.79)

1.21 (1.02)
0.99 (0.78)

Phobic
anxiety
Pre
Post

0.76 (0.75)
0.44 (0.58)

0.91 (0.99)
0.71 (0.99)

Note. PG-CBT integrative cognitive behavioural therapy for prolonged grief;

WG waitlist group; PG-13 Interview for Prolonged Grief-13; SCL-90R Symptom Checklist-90-Revised.
a
For the SCL-90-R in PG-CBT n at post-treatment is 18 because one person did
not complete the SCL-90-R (which means that ANCOVAs for Global severity index
and the SCL-90-R subscales are based on 41 cases).
nn
p o0.01.
nnn
p o 0.001.

4. Discussion
The present ndings suggest that our treatment for PGD is
highly effective in terms of reducing grief severity. The controlled
ES for grief symptom improvement in PG-CBT completers is
d 1.61; compared with the ES values reported in meta-analyses

62

R. Rosner et al. / Journal of Affective Disorders 167 (2014) 5663

for an average of six months. Therefore, one could argue that the
PG-CBT completers improved signicantly more than waiting list
participants only because they had more time to improve. However, given the rather long duration of time since loss, and
therefore the chronicity of their symptoms, such an argument
seems rather unlikely. Second, post-treatment assessment was not
blinded; for practical reasons, it was performed by the respective
therapist. Finally, we computed no power analysis in advance and
the sample size was small; furthermore only a few male patients
were treated. Therefore, these results should be regarded as being
preliminary.
4.2. Clinical implications and conclusions
We reported on a broader range of comorbid diagnoses than in
previous treatment trials; our results show that PGD patients have
a high degree of comorbidity. Although results on comorbidity are
not directly comparable between studies because the diagnostic
procedures varied, studies performed to date appear to show a
high comorbidity with depression and anxiety disorders as well as
with PTSD; for the latter diagnosis, we found a markedly lower
prevalence than Shear et al. (2005). In our study, somatoform
disorders were surprisingly prevalent. Somatization symptoms
should be assessed in future studies.
PG-CBT showed less pronounced improvement in comorbid
symptoms than the treatments in the study by Boelen et al.
(2007); this may be due to specic sample characteristics, such
as other comorbidity patterns. However, PG-CBT proved to be as
effective as other treatments in the designated target syndrome,
prolonged grief. Together with the results of Rosner et al. (2011b),
where this approach was evaluated in a very different setting
(group format with inpatients), these ndings are encouraging
enough to warrant further study. On a broader scope, this trial
once again showed that PGD is a useful diagnostic entity that can
be specically targeted, supporting the validity of PGD. Consequently, our results suggest that it is useful to screen for PGD and
provide a specic grief treatment protocol in clinical settings
where patients are admitted for various disorders.
Role of funding source
The actual therapy hours were compensated through the health insurance
system. The study itself has not been funded.

Conict of interest
None of the authors has any conict of interest.

Acknowledgement
None.

References
Allumbaugh, D.L., Hoyt, W.T., 1999. Effectiveness of grief therapy: a meta-analysis.
J. Couns. Psychol. 46, 370380.
American Psychiatric Association, 2013. Diagnostic and statistical manual of mental
disorders, fth ed. Am. Psychiatr. Pub., Arlington, VA.
Boelen, P.A., de Keijser, J., van den Hout, M.A., van den Bout, J., 2007. Treatment of
complicated grief: a comparison between cognitive-behavioral therapy and
supportive counselling. J. Consult. Clin. Psychol. 75, 277284.
Boelen, P.A., Prigerson, H.G., 2007. The inuence of prolonged grief disorder,
depression and anxiety on quality of life among bereaved adults. A prospective
study. Eur. Arch. Psychiatry Clin. Neurosci. 257, 444452.
Boelen, P.A., van den Bout, J., 2005. Complicated grief, depression, and anxiety as
distinct postloss syndromes: A conrmatory factor analysis study. Am. J.
Psychiatry. 162, 21752177.
Boelen, P.A., van den Hout, M.A., van den Bout, J., 2008. The factor structure of
posttraumatic stress disorder symptoms among bereaved individuals: a conrmatory factor analysis study. J. Anxiety Disord. 22, 13771383.
Bonanno, G.A., Boerner, K., Wortman, C.B., 2008. Trajectories of grieving. In:
Stroebe, W., Hansson, R.O., Schut, H., Stroebe, W. (Eds.), Handbook of

Bereavement Research and Practice. American Psychological Association,

Washington, DC, pp. 287307.
Cohen, J.A., Mannarino, A.P., Deblinger, E., 2006. Treating trauma and traumatic
grief. Guilford, NY.
Currier, J.M., Neimeyer, R.A., Berman, J.S., 2008. The Effectiveness of Psychotherapeutic Interventions for bereaved persons: a comprehensive quantitative
review. Psychol. Bull. 134, 648661.
Derogatis, L.R., 1977. SCL-90-R: administration, scoring and procedures manual for
the (revised) version. John Hopkins University School of Medicine, Baltimore.
Ehlers, A., 1999. Posttraumatische Belastungssto
rung (Posttraumatic stress disorder). Go
ttingen, Hogrefe.
Fortner, B.V., 2000. The Effectiveness of Grief Counseling and Therapy: A Quantitative Review. University of Memphis, Memphis, TN (Unpublished masters
thesis).
Franke G.H., SCL-90-R. Symptom-Checkliste von L. R. Derogatis Deutsche Version
(SCL-90-R. Symptom-Checklist by L. R. Derogatis German Version), 2002,
Gttingen; Beltz Test.
Horowitz, M.J., Siegel, B., Holen, A., Bonanno, G.A., Milbrath, C., Stinson, C.H., 1997.
Diagnostic criteria for complicated grief disorder. Am. J. Psychiatry 154,
904910.
Jacobs, S., 1999. Traumatic grief: Diagnosis, Treatment, and Prevention. Taylor &
Francis, Philadelphia.
Jacobson, N.S., Truax, P., 1991. Clinical signicance: a statistical approach to dening
meaningful change in psychotherapy research. J. Consult. Clin. Psychol. 59,
1219.
Kato, P.M., Mann, T., 1999. A synthesis of psychological interventions for the
bereaved. Clin. Psychol. Rev. 19, 275296.
Kersting, A., Brhler, E., Glaesmer, H., Wagner, B., 2011. Prevalence of complicated
grief in a representative population-based sample. J. Affect. Disord. 131,
339343.
Latham, A.F., Prigerson, H.G., 2004. Suicidality and bereavement: complicated grief
as psychiatric disorder presenting greatest risk for suicidality. Suicide Life
Threat. Behav. 34, 350362.
Maercker, A., Forstmeier, S., Enzler, A., Krsi, G., Ho
rler, E., Maier, C., Ehlert, U.,
2008. Adjustment disorders, PTSD and depressive disorders in old age: ndings
from a community survey. Compr. Psychiatry 49, 113120.
Maercker, A., Znoj, H.J., 2010. The younger sibling of PTSD: similarities and
differences between complicated grief and posttraumatic stress disorder. Eur.
J. Psychotraumatol 1, 5558.
Melges, F., DeMaso, D., 1980. Grief-resolution therapy. Reliving, revising, and
revisiting. Am. J. Psychother. 34, 5161.
Newson, R.S., Boelen, P.A., Hek, K, Hofman, A, Tiemeier, A., 2011. The prevalence and
characteristics of complicated grief in older adults. J. Affect. Disord. 132,
231238.
Ott, C.H., 2003. The impact of complicated grief on mental and physical health at
various points in the bereavement process. Death Stud. 27, 249272.
Pfoh, G., 2007. Interview zur Komplizierten Trauer-13 (IKT-13). [German version of
the PG-13]. Unpublished manuscript.
Prigerson, H.G., Bierhals, A.J., Kasl, S.V., et al., 1997. Traumatic grief as a risk factor
for mental and physical morbidity. Am. J. Psychiatry 154, 616623.
Prigerson, H.G., Frank, E., Kasl, S.V., Reynolds, C.F., Anderson, B., Zubenko, G.S.,
Houck, P.R., George, C.J., Kupfer, D.J., 1995. Complicated grief and bereavementrelated depression as distinct disorders: preliminary empirical validation
among elderly bereaved spouses. Am. J. Psychiatry 152, 2230.
Prigerson, H.G., Horowitz, M.J., Jacobs, S.C., Parkes, C.M., Aslan, M., Goodkin, K.,
Raphael, B., Marwit, S.J., Wortman, C., Neimeyer, R.A., Bonanno, G., Block, S.D.,
Kissane, D., Boelen, P., Maercker, A., Litz, B.T., Johnson, J.G., First, M.B.,
Maciejewski, P.K., 2009. Prolonged grief disorder: psychometric validation of
criteria proposed for DSM-V and ICD-11. PLoS Med. 6, e1000121, http://dx.doi.
org/10.1371/journal.pmed.1000121.
Prigerson, H.G., Maciejewski, P.K, n.d. Interview for Prolonged Grief (The PG-13).
Unpublished manuscript.
Prigerson, H.G., Vanderwerker, L.C., Maciejewski, P.K., 2008. A case for inclusion of
prolonged grief disorder in DSM-V. In: Stroebe, M.S., Hansson, R.O., Schut, H.,
Stroebe, W. (Eds.), Handbook of Bereavement Research and Practice: Advances
in Theory and Intervention. American Psychological Association, Washington,
DC, pp. 165186.
Rando, T.A., 1993. Treatment of Complicated Mourning. Research Press, Champaign, Ill.
Rosner, R., Hagl, M., 2007. Was hilft bei Trauer nach interpersonalen Verlusten?
Eine Literaturbersicht zu Behandlungsstudien bei Erwachsenen. [Effective
interventions after bereavement. A literature review of treatment studies on
adults]. Psychodyn. Psychother. 6, 4754.
Rosner, R., Kotouov, M., Pfoh, G. Diagnose der anhaltenden Trauersto
rung mit der
deutschen Version des PG-13 [Diagnosing prolonged grief disorder with the
German version of the PG-13]. In: Rosner, R., Pfoh, G, Rojas, R., Brandsttter, M.,
Rossi, R., Lumbeck, G., Kotouov, M., Hagl, M., Geissner E. (Eds.). Anhaltende
Trauersto
rung. Manuale zur Einzel- und Gruppentherapie [Prolonged grief
disorder. Manuals for individual and group treatment]. Hogrefe, Go
ttingen.
(in press).
Rosner, R., Kotouov, M., Pfoh, G., 2011a. Treatment of complicated grief. Eur. J.
Psychotraumatol. 2, 7995, http://dx.doi.org/10.3402/ejpt.v2i0.7995.
Rosner, R., Kruse, J., Hagl, M., June 2005. Quantitative and qualitative review of
interventions for the bereaved. In: Proceedings of the Paper presented at the
36th Annual Meeting of the Society of Psychotherapy Research. Montreal, CA.
Rosner, R., Lumbeck, G., Geissner, E., 2011b. Evaluation of a group therapy
intervention for comorbid complicated grief. Psychother. Res. 21, 210218.

R. Rosner et al. / Journal of Affective Disorders 167 (2014) 5663

Schaal, S., Elbert, T., Neuner, F., 2009. Prolonged grief disorder and depression in
widows due to the Rwandian genocide. Omega 59, 209219.
Shear, K., Frank, E., Houck, P.R., Reynolds, C.F., 2005. Treatment of complicated grief.
A randomized controlled trial. J. Am. Med. Assoc. 293, 26012608.
Shear, M.K., Simon, N., Wall, M., Zisook, S., Neimeyer, R., Duan, N., Reynolds, C.,
Lebowitz, B., Sung, S., Ghesquiere, A., Gorscak, B., Clayton, P., Ito, M., Nakajima,
S., Konishi, T., Melhem, N., Meert, K., Schiff, M., O'Connor, M.F., First, M., Sareen,
J., Bolton, J., Skritskaya, N., Mancini, A.D., Keshaviah, A., 2011. Complicated grief
and related bereavement issues for DSM-5. Depress. Anxiety 28, 103117.
Shear, M.K., Zuckoff, B.S., Frank, E., 2001. The syndrome of traumatic grief. CNS
Spectrsc. 6, 339346.
Simon, N.M., Shear, K.M., Thompson, E.H., Zalta, A.K., Perlman, C., Reynolds, C.F.,
Frank, E., Melhem, N.M., Silowash, R., 2007. The prevalence and correlates of

63

psychiatric comorbidity in individuals with complicated grief. Compr. Psychiatry 48, 395399.
Stroebe, M., Schut, H., Stroebe, W., 2007. Health outcomes of bereavement. Lancet
370, 19601973.
Wagner, B., Knaevelsrud, C., Maercker, A., 2006. Internet-based cognitive-behavioral therapy for complicated grief: a randomized controlled trial. Death Stud.
30, 429453.
Wittchen, H.U., Pster, H., 1997. Diagnostisches Expertensystem fr Psychische
Sto
rungen. DIA-X Interviews [Diagnostic Expert System for Mental Disorders].
Swets & Zeitlinger, Frankfurt.
Wittouck, C., van Autreve, S., de Jaegere, E., Portzky, G., van Heeringen, K., 2011. The
prevention and treatment of complicated grief: a meta-analysis. Clin. Psychol.
Rev. 31, 6978.