Evidence based treatment for adult patients
with child abuse related (Complex)PTSD:
a quantitative review
6
Ethy Dorrepaal, Kathleen Thomaes, Dick J Veltman, Adriaan W Hoogendoorn, Nel Draijer,
Anton JLM van Balkom. In preparation
78 |
treatment for (complex) ptsd: a quantitative review
Abstract
Introduction: Effective first line treatments for PTSD are well established, but
opinions about their generalizability to child abuse (CA) related Complex PTSD
are still devided. This study sought to determine the available evidence to guide
choice of treatment for this population.
Method: We performed a quantitative review of literature from 1989 to 2012 and
identified 24 RCT’s with mainly PTSD diagnosed as well as child abused study
populations. Six of these studies, with treatments specifically targeting CA related
(Complex) PTSD, were meta-analyzed, including variables such as inclusion/
exclusion criteria, effect size, dropout, recovery and improvement rates, pre- and
post-scores in both completers as well as intention-to-treat analysis.
Results: Results indicate that CA related PTSD patients profit with large effect
sizes and modest recovery and improvement rates from mainly cognitivebehavioral therapy. Treatments which include exposure showed better effect
sizes in completer analysis, but not in intention-to-treat analysis. Additionally, no
differential results were found in recovery and improvement rates. On all outcome
measures, results in the subgroup of Complex PTSD studies were less favorable as
compared to the ‘simple’ PTSD studies. In contrast to all CA related PTSD studies,
within the subgroup of Complex PTSD studies no superior effect size was found
for exposure, moreover, affect management showed better outcomes in terms of
dropout, recovery and improvement rates.
Conclusion: Our results suggest that a variety of mainly CBT treatments with
a mean duration of 12 to 24 weeks may be effective for CA related PTSD, but
these treatments do not suffice to obtain adequate end states, especially in the
Complex PTSD populations. Treatment for this population needs improvement to
reach better results with high compliance, and possibly should focus on relevant
comorbidity. Moreover, we propose that future research should aim to also include
Complex PTSD patients who are referred, unemployed, low educated, have
comorbid personality disorders, are on medication, and are previous treatment
non-responders, thereby improving generalizability of results.
treatment for (complex) ptsd: a quantitative review
| 79
Introduction
Effective treatments for posttraumatic stress disorder (PTSD) are well established:
first line treatments consist of several forms of cognitive behavior therapy (CBT),
such as prolonged exposure (PE), cognitive (processing) therapy (C(P)T) with and
without exposure, and Eye Movement Desensitization and Reprocessing (EMDR)
(Bradley et al., 2005; Cloitre, 2009). However, until now there is only sparse evidence
for effective treatments in child abuse (CA) related (Complex) PTSD. By interfering
with normal development, child abuse may result in PTSD complicated by problems
in affect regulation, memory and attention, self-perception, interpersonal relations,
somatization and systems of meaning (Herman, 1992). This syndrome is referred to
as ‘PTSD with associated features’ in DSM-IV-TR (APA, 2000) or ‘Complex PTSD’, and
is characterized by high comorbidity on both DSM-IV Axis I and II. Empirical studies
as well as neurobiological findings support the distinction of Complex PTSD from
‘simple’ PTSD (Ford, 1999; Lanius et al., 2010; Thomaes et al., 2010; 2011; Van der
Kolk et al., 2005; Zlotnick et al., 1996) and a prevalence of 1% of Complex PTSD is
observed in a student population (Ford et al., 2006).
To our knowledge, no reviews have been published that exclusively focused neither
on the efficacy of treatments of CA related Complex PTSD, nor on CA related PTSD.
Reviews on child abuse (Callahan et al., 2004; Kessler et al., 2003; Matsolff et
al., 2005; Peleikis et al., 2005; Taylor & Harvey, 2010) showed that a variety of
treatments are beneficial. Earlier reviews mainly included group treatments, while
more recently individual treatments showed favorable effect sizes. Structured
treatment characteristics such as availability of a manual, an instructional format
and providing homework increased effect in terms of PTSD symptoms, while
externalizing problems were unchanged after treatment (Taylor & Harvey, 2010).
However, these reviews included a limited number of Randomized Controlled Trials
(RCTs) with properly diagnosed PTSD, and these PTSD studies were not analyzed
separately to investigate differential treatment effects.
Reviews on PTSD (Bisson & Andrew, 2007; Benish et al., 2008; Bradley et al., 2005;
Cloitre, 2009; Powers et al., 2010; Schottenbauer et al., 2008; Seidler & Wagner,
2006) concluded that active treatments for PTSD are highly effective and superior
to waiting list controls. The largest body of evidence was assembled for cognitive
behavioral therapy (CBT), either exposure, cognitive therapy, in combination, or
EMDR. These reviews included only a small number of primary RCTs concerning
child abuse populations. Powers et al. (2010) found no significant difference in
effect sizes between studies with and without a child sexual abuse population,
based on 2 (partly) child abuse studies. After child abuse and chronic interpersonal
violence, Cloitre (2009) found a range of CBT treatments effective to achieve PTSD
symptom reduction. In addition, a few treatments were reported that focused
on other symptom domains, such as affect management and interpersonal skills,
which resulted in beneficial results on these domains as well.
80 |
treatment for (complex) ptsd: a quantitative review
Notably, these reviews on child abuse as well as on PTSD show considerable
differences in inclusion of studies (target population, study design), outcome
measures and methods of data analysis (effect size, recovery and/or improvement
rates, intention to treat analysis or completer analysis). Thus, conclusions on the
empirical evidence for effective treatments in child abuse related Complex PTSD
can hardly be drawn. So it is still unclear for clinicians, whether all Complex PTSD
patients are able to tolerate, and benefit from, immediate first-line treatments
equally well as ‘simple’ PTSD patients, and opinions are divided on this issue.
Complex PTSD as well as PTSD with Borderline Personality Disorder (BPD) have
been associated with poor treatment outcome (Cloitre & Koenen, 2001; Ford & Kidd,
1998) and a higher dropout rate after exposure (Cloitre et al., 2010; McDonagh et al.,
2005). Moreover, first line PTSD treatments may not address all relevant pathology
in the child abused population, such as poor affect regulation and interpersonal
problems.
Systematic inventories among experts recommend an initial focus on ‘stabilization’
in severely dysfunctional patients with an inability to tolerate strong affects before
exposure (Baars et al., 2011; Cloitre et al., 2011). Thus, regular exposure treatments
may not fit Complex PTSD patients in the first phase of their treatment. More
psycho-educational or stabilizing treatments targeting affect dysregulation,
irrational beliefs, and/ or lack of social and self soothing skills may prepare patients
to subsequent first line treatments such as exposure (e.g. Harned et al., 2010), or
directly reduce PTSD symptoms in other cases (e.g. Zlotnick et al., 1997).
Summarizing, after early severe child abuse PTSD may be complicated by additional
features referred to as Complex PTSD. Both child abuse reviews as well as PTSD
reviews conclude that effective treatments are available for child abuse or PTSD,
but research on the overlap between these populations is limited. It is unknown
which patients are more likely to benefit from one method or the other. Moreover,
the results of these reviews have questionable generalizability to the child abuse
related Complex PTSD population, since only a few studies on both child abuse
and PTSD were included and little attention was paid to indicators of complexity
like axis II comorbidity. Also, exclusion criteria such as suicidality and self-injurious
behavior may have resulted in the exclusion of Complex PTSD patients. Over time
more studies on both CA and PTSD have been published, but it is not clear yet how
these population characteristics are related to treatment effects/ compliance (of
different treatments: type, duration, manualized). Therefore our research question
was, what evidence is available to effectively treat the subgroup of CA related
Complex PTSD?
treatment for (complex) ptsd: a quantitative review
| 81
Method
Literature search
The literature search was restricted to the period 1989 to May 2012. We searched
MEDLINE using the following terms: child abuse OR childhood abuse OR child
sexual abuse OR PTSD OR posttraumatic AND treatment OR therapy AND
controlled trial OR clinical trial OR randomized OR review OR meta analysis.
These terms were searched as key words, title, abstract and Mesh terms. Findings
were cross-referenced with references from reviews. We included published
randomized original studies comparing interventions with other interventions
or control conditions in study populations combining PTSD and CA. Inclusion
criteria were (1) > 50% of participants who met DSM-III-R, DSM-IV, or DSM-IV-TR
criteria for posttraumatic stress disorder or PTSD main treatment target; (2) > 50%
of participants with CA or CA analyzed separately; (3) random assignment; (4) the
study participants were at least 18 years of age; (5) the study had to test a specific
psychotherapeutic treatment against a control condition and / or an alternative
treatment; (6) the study had to be reported in English.
Calculation of study outcome measures
The following measures from the original studies were drawn upon for the
calculation of PTSD effect sizes : (1) Clinician-Administered PTSD Scale (CAPS; Blake
et al., 1995); (2) PTSD Symptom Scale Self-Report (PTSD-SR; Foa et al., 1993); (3)
MPSS-SR (Falcetti et al., 1993); (4) Davidson Trauma Scale (Davidson et al., 1997).
Two calculations were performed for each original study: (1) an assessment of
effect sizes in the form of the standardized pre-post score difference of the groups
studied, and (2) a standardized score between conditions per study. For both
calculations Cohen’s d ((mean 1- mean 2)/ sd prepooled) was used as the measure of
the effect size (ES) using the following formula for the pooled standard deviation
of the pretreatment scores:
s=
(n1 − 1)s12+ (n2 − 1)s22
,
n1 + n2
resulting in an effect size also known as Glass’s delta. We considered pre-post
effectsizes (1) > 0.2 small; > 0.5 medium and > 0.8 large. For the direct comparison
of two forms of treatment (2), we calculated the post/post effect sizes (dpost/post),
and corrected these effect sizes for group differences at the beginning of the
study (dpre/pre) because such baseline differences may bias comparisons. Corrected
effect sizes dcorr were obtained by: dcorr=dpost/post – dpre/pre (Becker, 1988, Seidler et al.,
2006; Morris & Scott, 2007). Between condition effect sizes are considered medium
between 0.35 and 0.75.
82 |
treatment for (complex) ptsd: a quantitative review
Additionally, we present 1) percentage inclusion, conservatively defined as number
of patients screened by researchers, even if pre-screened by phone or clinicians, 2)
recovery rate, defined as percentage loss of PTSD diagnosis, and 3) improvement
rates using definitions for improvement as used by the authors (Bradley et al., 2005)
for both completers as well as intention-to-treat samples, with the aim to provide a
comprehensive overview. For all outcomes, we estimated non-reported intentionto-treat (ITT) or completers (CPL) scores based on published dropout percentages
per condition assuming that dropouts did not change.
Next, we computed a cumulative effect size of global PTSD symptoms across studies.
First, a joint effect size was calculated for the 3 original studies using more than
one PTSD measure. This joint effect size is equivalent to the arithmetical average of
the global scale scores. Subsequently, the average global effect size was calculated
from these primary-study effects, using fixed effect weights with software of
Metan (Harris, 2008). Subsequently, we computed confidence intervals for both
continuous (pre-post) as well as binary data (dropout, recovery and improvement
rates). For treatment versus control effect sizes no confidence limits are computed
yet. For confidence intervals the proportions of overlap (Cummings & Finch, 2005)
were computed to obtain an estimation if the pooled data of two groups of studies
differ significantly. Non overlapping intervals have a corresponding p value < 0.01
and intervals overlapping no more than 0.5 have a corresponding p value of < 0.05.
The p values are reported in the text only.
Results
Description of study selection
A total of 24 studies were identified that satisfied inclusion criteria, including both
a majority of patients with PTSD symptoms as well as a majority of patients with a
history of child abuse. The study populations differed with respect to 1) percentage
of patients meeting criteria for PTSD diagnosis or other PTSD indicators (e.g.,
level of PTSD symptoms), 2) percentage of patients meeting criteria for Complex
PTSD or other Complex PTSD indicators (e.g., percentage comorbid personality
disorders), 3) percentage of patients with child abuse history, 4) index trauma (e.g.,
child abuse or (adult) rape), and 5) target symptoms/ treatment goal (e.g., Complex
PTSD symptoms, PTSD symptoms or nightmares).
In Figure 1 we arranged these RCTs based these crireria, in order to visualize their
relevance in terms of our research question: what evidence based treatments are
available for CA related Complex PTSD? We identified five sections, varying in
target population (Figure 1), with the first (A) the most relevant to and specific for
our question and the others decreasing in relevance and specificity.
treatment for (complex) ptsd: a quantitative review
| 83
MAINLY PTSD + MAINLY CHILD ABUSE
E
Scheck 1998
Krakow 2001
Johnson 2011
C
Echebura 1997
Resick 2002
Kubany 2003
Kubany 2004
van der Kolk 2007
Resick 2008
Krupnick 2008
Zlotnick 2009
Mueser 2009
ALL PTSD +
MAINLY CHILD ABUSE
A
Zlotnick 1997
McDonnagh 2005
Cloitre 2010
Dorrepaal 2012
CA-RELATED
COMPLEX PTSD
B
Classen 2001
Cloitre 2002
Chard 2005
CHILD ABUSE
RELATED PTSD
D
Edmond 1999
Bradley 2003
Sikkema 2007
Paivio 2012
ALL CHILD ABUSE +
MAINLY PTSD
Figure 1
Section A: CA related Complex PTSD: all study participants Complex PTSD diagnosis or > 50% diagnosed with
personality disorders, all CA as index trauma, treatment target CA related Complex PTSD.
Section B: CA related PTSD treatment studies: all study participants diagnosed with PTSD, all CA as index
trauma, treatment target was CA related (Complex) PTSD
Section C: PTSD populations with > 50% CA (CA population analyzed separately), CA sometimes index trauma,
once separately analyzed, treatment target (Complex) PTSD symptoms
Section D: CA populations > 50% PTSD (or substantial PTSD symptomatology). CA index trauma, PTSD
symptoms treatment target.
Section E: > 50% PTSD diagnosis and > 50% CA, treatment target PTSD symptoms.
Section A: of these 24 RCTs, four studies (Zlotnick et al., 1997; McDonnagh et al.,
2005; Cloitre et al., 2010; Dorrepaal et al., 2012) in the middle circle of Figure 1 can
be characterized as CA related Complex PTSD treatment RCTs. All participants had
been diagnosed with Complex PTSD or at least 50% of them were diagnosed with
personality disorders. In these studies all patients suffered child abuse as the index
trauma and the treatment target was Complex PTSD symptomatology related to
child abuse.
Section B: three more studies (Classen et al., 2001; Cloitre et al., 2002; Chard, 2005)
in figure 1 can be characterized as CA related PTSD treatment studies, but not
Complex PTSD studies. All patients were diagnosed with PTSD according to DSMIV criteria, all had CA as index trauma, and the treatment target was PTSD related
to child abuse. All three studies included outcome measures covering Complex
PTSD symptoms (e.g. dissociation, interpersonal problems, affect modulation and
anger).
84 |
treatment for (complex) ptsd: a quantitative review
Section C: the participants of these nine additional studies (Echebura et al., 1997;
Resick et al., 2002; Kubany et al., 2003; Kubany et al., 2004; Van der Kolk et al., 2007;
Resick et al., 2008; Krupnick et al., 2008; Mueser et al., 2008; Zlotnick et al., 2009)
met criteria for a DSM-IV PTSD diagnosis, and the majority suffered child abuse, or
the CA population was analyzed separately (Resick et al., 2002). In some of these
studies, child abuse was the index trauma for part of the patients (Echebura et al.,
1997; Krupnick et al., 2008; Mueser et al., 2008; Resick et al., 2008), only in one study
these patients were separately analyzed (Van der Kolk et al., 2007). In other studies
e.g. rape or domestic violence were index trauma’s (Resick et al., 2002; Kubany et
al., 2003; 2004). Some studies aimed at improving Complex PTSD symptoms (e.g.
Resick et al., 2002; 2008), while another study excluded all comorbidity (Echebura
et al., 1997).
Section D: these five studies (Edmond et al., 1999; Bradley & Folingstad, 2003;
Sikkema et al., 2007; Paivio et al., 2010; Classen et al., 2011) focused on CA
populations. These were either in majority diagnosed with PTSD or substantial
PTSD symptomatology was established. All used PTSD symptoms as treatment
target, but PTSD patients were not analyzed separately. Some focused on special
populations (e.g. incarcerated women (Bradley & Folingstad, 2003), or HIV/ aids
patients (Sikkema et al., 2007)).
Section E: these three study populations (Scheck et al., 1998; Krakow et al., 2001;
Johnson et al., 2011) mostly had a PTSD diagnosis and a history of child abuse. One
(Scheck et al., 1998) shows many complex characteristics, while another (Krakow
et al., 2001) focused specifically on nightmares. The Johnson et al. (2011) study
included a majority of patients with a significant trauma history, but not certainly
more than 50% child abuse patients, while focusing on Intimate Partner Violence.
All three aimed at PTSD symptom reduction.
Since we were specifically interested in treatments focusing on CA related
Complex PTSD, we selected the studies of category A – closest to our aimed
population – and added the studies of category B – focusing on CA related to
child abuse – to increase the number of studies, since they all included outcome
measures covering Complex PTSD symptoms.
Of these 7 studies, numbered 1–7 in tables and text (see also references: review),
four conditions were classified: cognitive behavioral therapy (CBT), present centred
therapy (PCT), treatment as usual during waiting list (TAU) and waiting list only
(WL). CBT was further subdivided into ‘including exposure’ (EXP) (imaginary or
in vivo) and ‘including affect management’ (AM) (skills training to improve affect
regulation). Control conditions include TAU and WL.
treatment for (complex) ptsd: a quantitative review
| 85
Population characteristics of CA related (Complex) PTSD studies
Table 1 lists population characteristics of the seven included studies. The mean age
of study populations ranged between 34–40 years, most populations, except two3,6,
were predominantly Caucasian, most patients were well educated and employed.
Three studies2-4 reported advertisements as their method of recruitment and two
studies only included referred patients1,7.
Common exclusion criteria were the presence of organic brain disorder, psychotic
disorder, and substance abuse or dependence (table 1). Suicidality was an exclusion
criterion of five of seven studies2-6. One6 excluded suicidality only if it required
referral to a hospital. The two Complex PTSD diagnosed studies1,7 did not exclude
suicidality. Other comorbid conditions like eating disorders3, bipolair disorders3,5,6,
dissociative (identity) disorder1-3,5-7 and severe depression5, borderline3 or antisocial
PD7 were sometimes excluded. Two studies4,5 excluded patients with ongoing
abuse. Two2,5 studies used criteria involving clear memories of abuse under the age
of 16 years with anyone 5 years older.
Inclusion rates after screening were provided in 6 of 7 studies2-7 and ranged
between 30% and 81%, with a mean of 56% (table 1). Two studies5,6 categorized as
Complex PTSD populations show low inclusion rates. Combining the low inclusion
rate and exclusion criteria of these two (A) studies5,6, some caution about their
generalizability to the Complex PTSD population is warranted.
Six studies diagnosed PTSD with the CAPS1,3-7, and three studies reported blinded
measurements3,6,7.
As for comorbidity: three studies3,4,7 reported on comorbid major depressive
disorder (MDD) (40-55%) and two6,7 reported a mean of other Axis I diagnoses
of more than 2. Four studies3,4,5,6 reported on depression severity using the BDI,
with mean scores ranging from 18–25 (cut off being moderate 16; severe 24).
Dissociation was measured four times with the DES, with scores ranging from 15
to 25 1,4,5,7 (the cut off being 15–20). The four Complex PTSD (A) studies provided
some information about personality pathology: 100% Complex PTSD (including
personality pathology) in two studies1,7 and at least 50% personality disorders in
three studies5-7 (in study 5 we estimated this based on 15 of 29 patients meeting
criteria for Borderline or Avoidant PD in the CBT condition). Borderline personality
disorder comorbidity ranged from 11–53%5,6,7.
Only three studies reported on previous treatments1,3,7. Trauma history was
extensive throughout studies: many incidents were reported of several and severe
types of child abuse, mainly by fathers/ relatives, as well as a high adult abuse
prevalence.
Medication use only was reported once7, showing 70% use of psychiatric
medication, including 20% antipsychotic medication.
advertising,
letters/
presentations
local health
professionals
MCDONAGH5 74 Women 40 yrs,
NR
well educated,
2005
white, employed,
(A)
multiple trauma’s
child and adult
71 Women age 33
Ca- PTSD
81% white
14 yrs education
Self-referred
58 PTSD & CA
advertisements
46% white
52% college
75% work/ student
CLOITRE3
2002
(B)
CHARD4
2005
(B)
Schizophrenia,
dementia delirium
amnestic/ cogn
disorders; ritual abuse;
current Pt; alc /drug
dependence; current
suic (last month)
Advertisement
newspapers/
flyers/ radio/
community
agencies
55 38 yrs
-> 64% white 44%
52 college graduate
80% employed
CLASSEN2
2001
(B)
Somatic condition*, (hypo) mania,
schizo(fren), schizoaff,
psychosis, DID, org
psych, sev. depr, bipol
depr, psychot depr,
curr alc/ drug ab.,
withdr 3 mnth prior,
act. suic, hist. 2 TS or
gest, abusive partner.
81% CAPS
56% CAPS
Blind
SA < 16 yrs by 30% CAPS
> + 5 yrs. clear
CSA-PTSD
intrusion.
suic intent, substanse PTSD DX CSA
dep current trauma,
med disorders, unstable
medication last
3 mo, subst abuse
last 3 mo,
Organic or psychotic
PTSD related
mental disorder, subst to CA
dep, eating disorder,
dissociative disorder,
bipol I, BPD, suic attept
or hosp last 3 month
CSA discussed 58% TSC-40
> 6 month
before
> 1 memory
genit. SA 3-15
yrs by known
> + 5 yrs
Psychosis
Substance abuse
DID
Referred
CAPS
SIDES
C-PTSD based ?
on CSA. Ther
1 month, no
change
psyhotropic R/
exclusion criteria
recruitment
48 39 yrs
CSA< 17 yrs
White
High school
criteria
Assessment
Severity
CTQ
NR
SIDES 100%
NR
50 %
prev
hosp.
SCID
I & II
NR
ELS
STI
SAEQ
11 % BPD;
Min N= 15/29
as II in CBT
NR
SCID I & SCID II NR
Only 40% MDD
Reported
29%
SCID-I : 45%
Excl CMIS
ER past
BPD SAAIVS MDD
year
79% anx dis
SCID II
25% past subst.
16% past
eati. dis
NR
NR
77% CSA fam,
35% parent;
37% rape.
Age 6.9 yrs.
M 3.7 abusers
48% both
CSA & CPA,
39% only
CSA, 13%
CPA only
CAPS 70 High
BDI 18
comorbid
DES 15 CPA and
adult abuse,
Majority
penetration,
relative
CAPS 67 Onset 6.4 yrs
57%> 100
BDI 24
DES 19 times, 63%>
1 abuser, 84%
relative.
CAPS 70
MPSSSR 70
BDI 24
only
Sexual
differenceExperiences
scores
Survey
DTS 70
DES 22
(Complex)
Prev. PTSD/
Trauma
rate PTSD
Axis II Trauma Co-morbidity treatm. BDI/DES severity
N* Population*
Inclusion
ZLOTNICK1
1997
(A)
Author
Year
(Figure 1)
Table 1. Population characteristics of randomized controlled trials with CA related PTSD (A+B), partly Complex PTSD (A).
86 |
treatment for (complex) ptsd: a quantitative review
71 40 yrs
educ 9.9 yrs,
17% employed
Referred
Longlasting psychosis,
DID, severe sunstanse
abuse interfering
with compliance,
antisocial PD
SCID I STI
SIDP-IV
SCID II NR
34% CAPS
Blind
CA- PTSD
CA related
79% CAPS
Complex PTSD
SIDES
blind
CA related
Sub dep, psychotic
PTSD
sympt cogn impair,
bipol, act suic requiring
hosp/emergency room.
No PTSD focused PT.
NR
93% CSA/ >
50% child
sexual &
physical ab.
> 50%
adult ab.
CAPS 64 90% CSA
BDI 21
80% CPA
Much adult
abuse
Mean nr
trauma: 6.5
75% axis II Dx, Mean DTS 85
52% BPD
2 prev DES 25
96% axis I Dx treatm.
55% MDD
70% medication: 47%
ssri/snri,
25% sedatives;
> 50%
antipsych
90% axis I;
Mean 2.4
50% axis II;
Mean 1.1
(BPD 24%)
BDI=Beck depression Interview; BPD= Borderline personality Disorder; CAPS= Clinician Administered PTSD Scale; CBT= Cognitive Behavioural therapy; CMIS= childhood Maltreatment
interview Scedule; CPA = child physical sbuse; CSA= child sexual abuse; CTQ= Childhood Trauma Questionnaire; DES= Dissociative Experiences Scale; DID= Dissociative Identity
Disorder; Dx= diagnosis; ELS= Evaluation of Lifetime Stressors Interview; ER= emergency room; MDD=major depressive Disorder; MPSS-SR= Modified Posttraumatic Stress Disorder
Symptom Scale; NR= not reported; SA= sexual abuse; SAAIVS= Sexual Assault and Additional Interpersonal Violence Scedule; SAEQ= Sexual Abuse Exposure Questionnaire; SCIDI=Structured Clinical Interview for DSM IV axis I; SCID-I/ II= structured ClinicaI Interview for DSM-IV; SIDES= Structured interview for Disorders of Extreme Stress; SIDP-IV= Structured
Interview for DSM IV Personality Disorders; STI= Structured Trauma Interview; TSC-40= Trauma Symptom checklist- 40; Ab.= abuse, antipsych= antipsychotic, educ= education. *=
Medication autonomic nervous system effects; pregnancy; cardio- vascular disease; antihypertensivum. 1,2,3,4,5,6,7 = study number referred to in text.
DORREPAAL7
2012
(A)
Table 1: Continued
104 88% > high school NR
CLOITRE6
33% unempl,
2010
33% married
(A)
35% white
treatment for (complex) ptsd: a quantitative review
| 87
88 |
treatment for (complex) ptsd: a quantitative review
Treatment characteristics
The seven RCTs included a total of 17 conditions: 11 treatment conditions and
six control conditions (4 WL only, 2 treatment as usual (TAU) during WL (table 2)).
Two studies2,5 included two active treatments, and one6 compared three active
treatment conditions.
The total number of patients was 478, with 297 receiving active treatment, 119 WL
only, and 56 TAU during WL. Six of seven studies1,3-7 provided data to calculate ES,
five completers1,3,4,5,7 and three5,6,7 intention-to-treat data. These studies included at
least one CBT condition, with overall 8 CBT conditions (Table 2), and one included
a present centered treatment (PCT) condition based on trauma genetic dynamics.
These nine treatment conditions were manualized, all reporting psycho-education
and homework as components of the treatment (Table 2). All CBT conditions
included cognitive therapy and/or restructuring. Five treatment conditions1,3,6 twice,7
explicitly aimed at improvement of affect regulation and dedicated a substantial
part of the treatment to affect management skills training. Five conditions2,3,5,6
twice
included some exposure. Four conditions3,6 twice,7 addressed interpersonal
functioning in explicit interpersonal skills training. Two conditions2 consisted of
group treatment only, in three conditions group treatment was combined with
individual care, once manualized4, twice with unstructured treatment as usual1,7.
Six active conditions consisted of individual treatment only3,5 twice, 6 (3x). Length of
treatment ranged from 12 to 24 weeks, including 14–27 sessions of 60–120 minutes
duration.
Dropout
The mean overall dropout rate was 22%, in active treatments 25% and in control
conditions 16% (Table 3 per study and table 4 aggregated data). CBT had higher
dropout rates 95%-CI= (21–32) as compared to PCT (-0.1, 24) (p< 0.05). Active
CBT conditions4-6 including some form of exposure without preceding affect
management showed a mean dropout rate of 32%, while the three conditions1,6-7
without exposure showed a mean dropout rate of 24% (ns). In the two studies with
direct comparisons between active treatment conditions5,6 exposure conditions
had a mean dropout rate of 40% as compared to a mean dropout of 18% in the noexposure active treatments conditions. Three studies1,4,5 reported characteristics of
dropout patients: e.g. higher PTSD1,4 and dissociation levels1, more severe trauma5,
anxiety5 and depression. One study5 also reported 100% dropout in the exposure
condition for BPD as compared to 0% in the present centered therapy, while
another7 found lower dropout rates for borderline personality disorder in the affect
management condition.
CBT: fear and
+
maladaptive cognitive
restructuring
PCT: traumagenic
dynamics
Affect dysregulation
and interpersonal
difficulties PTSD
7 weekly 120 minute CSA
followed by 7 90 min
individual sessions
16 weekly individual
sessions
20 weekly 120 minute CA
group sessions
added to TAU
MCDONAGH5
2005
CLOITRE6
2010
DORREPAAL7
2012
+
PE +
STAIR +
+
STAIR+
+
+
PE +
STAIR +
CBT +
PCT -
CPT-SA +
+
WA
PE
TF+?
-
+
PE –
STAIR +
-
PE +
STAIR –
CBT – (only
PE in CBT
breathing
PCT retraining)
PCT: problem
solving
_
STAIR +
PF+ some basic PF+/-?
assumptions?
+/2 sessions
Affect
Exposure
management techniques
+
PE +
STAIR +
CBT –
PCT: probl
solving skills
STAIR
interpersonal
learning?
-
Social skills
+
PE +
STAIR +
+
+
+
?
+
Home work
C(S)A= child (sexual) abuse; CBT= cognitive behavioral therapy; CR= cognitive restructuring; CT= cognitive therapy; PE= prolonged exposure; PF= present focused; STAIR=
skills raining in affect and interpersonal regulation; TF= trauma focused; WA= written accounts; 1,2,3,4,5,6,7 = study number referred to in text.
CSA
PTSD and CPTSD
PTSD/ fear and
attachment,
17 weekly 90 min
group sessions
combined with
10weekly 60 min
individual sessions
CHARD4
2005
CSA
Affect regulation
PTSD
8 weekly 60 min & 8 CA
twice a weekly 90 min
individual sessions
CLOITRE3
2002
+
TF: work through,
integrate.
PF: maladaptive
patterns past
CSA
ZLOTNICK
1997
1
+
Psycho
education CT/ CR
24 weekly 90 minute
group sessions
Target
Symptoms
CLASSEN2
2001
Index
trauma
Affect dysregulation
Format
15 weekly 120 minute CSA
group sessions
added to TAU
1st Author
year
Table 2. Treatment characteristics of randomized controlled trials with CA related (Complex) PTSD (A+B).
treatment for (complex) ptsd: a quantitative review
| 89
1
36 (30)
35 (28)
31 (22)
27 (24)
14+7
34
23 (16)
23 (17)
MCDONAGH5
2005
29 (17)
22 (20)
23 (20)
CHARD4
2005
CLOITRE3
2002
CLASSEN2
2001
ZLOTNICK
1997
Study
N ITT
(CPL)
18.4
15.1
16.9
67.1
67.5
70.0
CAPS
CBT
PCT
WL only
41%
9%
13%
22.85
24.74
57.57
57.52
16.3
16.6
69
69
26.39
23.67
16.6
18.6
69
73
-
22.00
25.83
SD
65.46
68.30
17%
20%
29%
11%
-
66.88
74.69
M
CAPS
CPT-SA
MA during WL
MPSS
CPT-SA
MA during WL
MFFS-SR
STAIR PE
WL only
CAPS
STAIR PE
WL only
Un
clear
29%
25%
DTS
AM
TAU during WL
TSC-40
TFT+PFT
WL only
Drop
out
Measure
Condition
Prescore
38.5
44.9
62.5
7.54
57.70
9.00
62.96
31
62
29
58
-
45.76
73.06
M
SD
27.7
22.1
17.0
9.51
27.47
11.04
30.68
25.2
22.7
27.6
28.6
-
34.12
29.86
CPL
1.75
1.39
0.44
2.14
–0.01
15.9
57.7
53.1
47.2
65.5
2.29
0.22
18.4
64.0
9)
2.36
0.43
14)
8.1
3.8
0.91
0.07
CPL
42.0
62.7
28.8
22.4
18.6
-
SD
1.03
1.26
0.40
1.78
–0.01
9)
1.91
0.17
1.68
0.39
9)
1.64
0.77
14)
17.0
14.1
9)
0.63
0.05
ITT
Pre-post
(Cohen’s d)
2.31
0.87
ITT
40.6
59.7
9)
-
-
9)
52.2
73.5
M
Postscore
1.79
9)
1.73
1.29
9)
0.87
-
-
9)
0.58
ITT
93%
26%
77%
25%
-
87%
41%
CPL
28%
32%
17%
9)
78%
21%
9)
55%
22%
-
-
9)
61%
30%
ITT
Recovery
rate
0.63 47%
1.29
0.86 35%
0.93
0.3710) –0.2310) 20%
2.15
2.07
1.93
1.45
-
0.84
CPL
Tx vs other
(Cohen’s dcorr)
Change/ improvement
-
79%
4%
46%
4%
-
-
CPL
-
9)
66%
3%
9)
33%
4%
-
-
ITT
Improvement
rates
Table 3. Inclusion and drop-out rates, means and SDs pre- and post treatment & effect sizes pre-post and therapy versus other, recovery rates and improvement
rates, on PTSD diagnosis/severity (completers and intention-to-treat) per study.
90 |
treatment for (complex) ptsd: a quantitative review
33 (28)
38 (28)
33 (20)
91.4
80.5
DTS
CBT group
TAU during WL
21.8
23.1
12.87
12.65
11.14
36.7
39.9
38.2
18%
12%
18.29
21.15
15.86
63.08
64.34
64.50
15%
26%
39%
CAPS
STAIR/EXP
STAIR/Sup
Sup/EXP
PTSD SR
STAIR/EXP
STAIR/Sup
Sup/EXP
29.4
30.3
69.6
66.5
14.0
14.5
19.0
9.95
5.43
6.52
66.7
65.5
32.70
32.32
39.72
8)
27.28
20.88
23.61
27.4
29.8
11.46
12.79
9.83
19.37
23.04
18.34
1.12
0.68
2.21
2.85
2.84
8)
1.94
2.35
2.22
0.99
0.63
1.87
2.10
1.59
0.44
0.35
–0.6411) –0.2211)
0.2312) 0.5112)
–0.4113) 0.2913)
-
8)
1.65 –0.2811) 0.3011) 72%
0.1412) 0.3912) 64%
1.73
1.34 –0.4113) –0.0913) 55%
-
61%
47%
33%
55%
24%
8)
32%
33%
10%
45%
21%
27%
24%
6%
AM= affect management; CAPS= Clinician Administered PTSD Scale; CBT= cognitive behavior therapy?; CPL= Completer; DTS= Davidson trauma Scale; ITT= Intention-to-treat;
M= mean; MA= minimal attention; MFFS-SR= Modified Posttraumatic Stress Disorder Symptom Scale; MPSS= Modified PTSD Symptom Scale; PTSD-SR= PTSD Self rating scale;
PE= prolonged exposure; PFT= present focused therapy; SD= standard deviation; STAIR= skills training in affect and interpersonal regulation; TAU = treatment as usual; TSC40= trauma Symptom Check list; TFT= trauma focused therapy; WL= waiting list.
1,2,3,4,5,6,7
= study number referred to in text; 8) M CPL is estimated from M ITT, N ITT and N CPL, assuming non-completers did not change; 9) M ITT is estimated from M CPL, N ITT
and N CPL, assuming non-completers did not change; 10) CBT vs. PCT, 11) STAIR/EXP vs. Sup/EXP, 12) STAIR/Sup vs. Sup/EXP, 13) STAIR/EXP vs. STAIR/Sup, 14) change scores as reported
in article, unknown if standardized
DORREPAAL7
2012
38 (31)
33 (29)
CLOITRE6
2010
Table 3. Continued
treatment for (complex) ptsd: a quantitative review
| 91
92 |
treatment for (complex) ptsd: a quantitative review
Within the Complex PTSD studies (A) the comparison between different types of CBT
showed lower drop-out out in affect management only (24%; 95%-CI= (15,33)) and
affect management combined with exposure (15%; 95%-CI= (2,29)) as compared to
exposure only (40%; 95%-CI= (27,53)) (p < 0.05) (Table 5).
In sum, present centred therapy showed lowest drop-out rates. Within CBT
treatments, drop-out rates for exposure were high as compared to affect
management, significantly in Complex PTSD studies.
Effect sizes
The effect sizes in table 3 and 4 indicate that active treatments resulted in
substantial improvement from pre to post treatment in this patient population
ranging from 0.6 to 2.8 (Table 3) with a mean of 1.7 for completers and 1.3 for
intention-to-treat (Table 4). The effect sizes of control conditions ranged from no
effect to medium effect sizes with a small mean effect size of 0.4 in completers,
and 0.3 in intention-to-treat, (in WL only3-6 as well as TAU during WL1,7 conditions).
The effect sizes of active treatments versus control (WL only plus TAU during WL)
comparisons ranged from 0.4 to 2.2 (table 3), with large mean effects of 1.2 in
completers and 0.9 in intention-to-treat (table 4).
In Table 4 we also aggregated data for CA related PTSD (A & B) per type of active
condition, showing large effect sizes pre post for all types of active treatment
conditions. The 95% confidence interval (CI)(1.58, 2.57) for treatments including
exposure only (EXP no AM) in contrast with affect management only treatments
(AM no EXP) 95%-CI= (0.96,1.76) indicated more favorable results for exposure in
completers (p< 0.05). This was also the case for combined affect management and
exposure 95%-CI= (1.52,2.67). In intention-to-treat no significant superiority of
one type of treatment over the other was observed. Generally the same pattern
between treatment types was observed in treatment versus control effect sizes,
except affect management only showing a medium effect size in intention-to-treat.
Additionally we observed small to medium effect sizes (ranging from 0.09 to 0.51)
in direct comparisons between active treatments within studies5,6: both in pre-post
as well as treatment versus other treatment effect sizes, relatively favorable results
for exposure only were found in completers analysis, whereas in intention-to-treat
relatively unfavorable results for exposure only were found (Table 3).
Comparing pre post effect sizes in CA related Complex PTSD studies1,5-7 (A) 95%CI= (1.28, 1.86) versus CA related non-Complex PTSD studies3,4 (B) 95%-CI= (1.68,
2.86) revealed less treatment gain for the Complex population (table 5) (p< 0.05) in
completers analysis, as confirmed in intention-to-treat analysis (p< 0.05).
26%
8
CBT
9%b
16%
15%
18%
1
5
3
2
PCT5
1,7
a
78
68
72
68
66
74
64
68
68
69
61
64
45
29
44
23
33
34
CPL
70
62
65
47
37
51
36
42
42
ITT
0.4
0.4
0.4
c
1.4
2.1e
1.4d,e
2.1d
1.7
1.7c
CPL
0.4
0.3
0.3
f
1.3
1.7
1.1
1.4
1.3
1.3f
ITT
Effect size (d)
Pre versus post
Treatment versus control
0
0
0
1
1
2
2
5
6
N
0.9
1.7
0.6
1.7
1.3
1.2
CPL
0.9
1.1
0.5
1.2
0.9
0.9
ITT
Effect size (d) post-postcorr
1
3
4
1
2
2
3
7
8
N
41%
24%
27%g
35%h
74%
73%
70%
72%
h
68%g
CPL
30%
20%
22%i
32%
58%
52%
48%
52%
50%i
ITT
Recovery rate 8
1
2
3
0
2
2
2
6
6
24%
4%
k
k
11%j
38%
44%
52%
45%
45%j
CPL
21%m
3%m
9%l
30%
34%
38%
34%
34%l
ITT
Improvement rate 9
N
Post
AM= affect management; CA= child abuse; CBT= cognitive behavioral therapy; CPTSD= Complex posttraumatic stress disorder; CPL= Completer; Exp= exposure; ITT= Intentionto-treat; PCT= present centered therapy; TAU= treatment as usual; WL= waiting list
* Without Cloitre 2010 due to incomparability of score ranges.
1,2,3,4,5,6,7
= study number supplying condition: 1= Zlotnick 1997; 3= Cloitre 2002; 4= Chard 2005; 6= Cloitre 2010; 7= Dorrepaal 2012.
8
Recovery rate = percentage decrease of PTSD diagnoses as reported by author
9
Improvement rate= percentage of patients that improved as defined by author
Pre-post effectsizes, recovery and improvement rates were tested: active vs control; CBT vs. PCT; the 3 subtypes of CBT amongst each other; and the two types of control
conditions:
a,b,d,e ,k,m
Differences between groups of studies with the same superschrift are significant at p-level < 0.05 as a result of modest overlap of confidence intervals.
c,f,g,h,i,,j,l
Difference between groups of studies with the same superschrift are significant at p-level < 0.01 as a result of non-overlap of confidence intervals.
– TAU during WL
3,4,5,
– WL only
Control conditions
1,3,4,5,7
2
24%
22%
3
– incl AM no exp 1,6,7
– incl AM & exp 3,6
32%
3
– incl exp no AM 4,5,6
b
25%a
9
1,3,4,5,6,7
DO
N
Treatment type
All active conditions
CA-PTSD (A+B)
PTSD score
Pre* Post*
Change in PTSD
Table 4. Aggregated drop-out, pre and post scores and effect sizes for PTSD symptom changes, recovery and improvement rate across CA-related PTSD studies by
type of study population and treatment.
treatment for (complex) ptsd: a quantitative review
| 93
9%a
16%
1
3
2
– TAU during WL1,7
18%
13%
78
70
75
68
63
74
66
70
65
69
69
62
67
45
27
44
31
37
19
38
CPL
70
65
68
47
33
51
45
46
29
46
ITT
0.4
0.4
0.4
e
1.4
1.9
1.4
2.0
1.6
2.3d
1.6
d,e
CPL
0.4
0.4
0.4
g
1.3
1.6
1.1
1.2
1.2
1.8f
1.2
f,g
ITT
Effect size (d)
Pre versus post
Treatment versus control
1
0
2
1
3
2
4
N
0.9
0.6
1.3
0.9
1.9
0.9
CPL
0.9
0.5
0.6
0.5
1.4
0.6
ITT
Effect size (d) post-postcorr
1
1
2
1
1
2
2
5
2
6
N
41%
20%
30%i
35%j
71%
73%
51%
65%j
87%h
60%
h,i
CPL
30%
17%
24%l
32%
61%n
52%m
31%m,n
46%
67%k
44%
1
0
1
0
1
2
1
4
2
4
N
35%
24%
24%
32%
44%p
10%p
35%
65%o
o
CPL
21%
21%
27%s
34%r
6%r,s
26%
51%q
26%q
ITT
Improvement rate9
Post
k,l
ITT
Recovery rate8
AM= affect management; CA= child abuse; CBT= cognitive behavioral therapy; CPTSD= Complex posttraumatic stress disorder; CPL= Completer; Exp= exposure; ITT=
Intention-to-treat; PCT= present centered therapy; TAU= treatment as usual; WL= waiting list
* Without Cloitre 2010 due to incomparability of score ranges
1,2,3,4,5,6,7
= study number supplying active condition: 1= Zlotnick 1997; 3= Cloitre 2002; 4= Chard 2005; 6= Cloitre 2010; 7= Dorrepaal 2012
8
Recovery rate = percentage decrease of PTSD diagnoses as reported by author
9
Improvement rate = percentage of patients that improved as defined by author
Pre-post effectsizes, recovery and improvement rates were tested: CA related Complex PTSD (A) vs CA related PTSD (B); active vs control; CBT vs. PCT and the 3 subtypes of CBT
amongst each other, and the two types of control conditions:
a,b,c,d,f,j,l,m,s
Differences between groups of studies with the same superschrift are significant at p-level < 0.05 as a result of modest overlap of confidence intervals.
e,g,h,I,k,n,o,p,q,r
Difference between groups of studies with the same superschrift are significant at p-level < 0.01 as a result of non-overlap of confidence intervals.
1
– WL only5
Control conditions
1,5,7
PCT5
15%c
24%b
3
1
b,c
28%a
2 40%
6
– incl AM no exp1,6,7
5,6,
26%
22%
7
2
DO
N
– incl AM & exp6
– incl exp no AM
CBT
Within CA-CPTSD (A):
CA-PTSD (B)3,4
1,5 (2x),6 (3x),7
CA-CPTSD (A)
Treatment type
Post*
PTSD score
Pre*
Change in PTSD
Table 5. Aggregated drop-out, pre- and postscores and effect sizes for PTSD symptom changes, recovery and improvement rate across CA-related Complex PTSD
studies by type of study population and treatment.
94 |
treatment for (complex) ptsd: a quantitative review
treatment for (complex) ptsd: a quantitative review
| 95
We additionally compared treatment gains per type of treatment within the
Complex PTSD studies only (section A) (Table 5) and now observed no evidence for
differential pre post effect sizes between type of active treatments in completers or
intention-to-treat analysis (large p values). The relatively modest results in the more
complex populations was also evident within the (A) category. The two studies1,7
that included Complex PTSD diagnosed patients showed lower effect sizes (table
3) compared to the studies5,6 that included a PTSD diagnosed population with a
minimal 50% personality disorder comorbidity as indicator of complexity, possibly
also including partly less complex patients.
This suggests that in in completers analysis only exposure treatment is superior
over affect management. Specifically for Complex PTSD study populations
however, treatment results are less favorable, and no differential effects are found.
Recovery rates
Five studies1,3-6 reported outcome in terms of recovery rate. In eight active conditions
a mean recovery rate of 50% (95%-CI= (44, 56)) was found, comparing favorably
with a recovery rate of 22% (95%-CI= (14, 30)) in the control conditions (table 4)
in intention-to-treat analysis (p< 0.01), also found in completers analysis (p< 0.01).
Table 4 shows a higher recovery rate for CBT (95%-CI= (65, 79)) as compared to
PCT (95%-CI= (12, 58)) (p< 0.01) in completers, but not in intention-to-treat. No
difference was observed in the direct comparison5 (table 3). In contrast to effect
size comparisons, no clear evidence for differential recovery rates between types
of CBT was found.
The Complex PTSD studies (A) showed a lower recovery rate (95%-CI= (36, 51)) as
compared to the PTSD studies (B) (95%-CI= (56, 79)), (p< 0.01) in intention-to-treat
analysis, which also was seen in completers analysis, in line with differential effect
sizes (Table 5).
Within Complex PTSD studies, we now found better recovery rates for affect
management only (95%-CI= (39, 65) versus exposure only (95%-CI= (20, 44) in
intention-to-treat analysis (p< 0.05) (table 5). The same was found for combined
exposure and affect management versus exposure only as well (p< 0.01). The same
pattern was found in completers analysis.
Taken together, modest recovery rates are observed overall, favoring CBT over PCT
in completers analysis. Again in the Complex PTSD studies recovery rates show
unfavorable results as compared to the PTSD studies. Recovery rates for affect
management (only or in combination with exposure) exceeded those for exposure
only, in Complex PTSD studies.
Improvement rates
Improvement rates for active treatments had a mean of 45% in completers and
34% in intention-to-treat (Table 4). Although this contrasts favorably with 11% and
96 |
treatment for (complex) ptsd: a quantitative review
9% respectively in control conditions (p< 0.01), a majority of patients failed to reach
this criterion. No differential improvement rates between CBT and PCT or types of
CBT were found.
Again, Complex PTSD patients (A) showed unfavorable results in terms of
improvement rate of 35% (95%-CI= (25, 44)) versus 65% in ‘simple’ PTSD patients (B)
(95%-CI= (52,78)) in completers, and in intention-to-treat analysis as well (p’s< 0.01)
(table 5).
Between type of treatment within the Complex PTSD (A) studies, exposure only
(95%-CI= (-4, 16)) showed lower improvement rates (table 5) as compared to affect
management only (95%-CI= (23, 45)) (p< 0.01) as well as compared to combined
exposure and affect management (95%-CI= (11, 43)) (p< 0.05) in intention-to-treat
(p< 0.05), as was also observed in completers (ns). Lower improvement rates for
exposure only were also found in direct comparisons5,6 in table 3.
In sum, overall most patients fail to reach criteria for improvement. In the Complex
PTSD studies improvement rates are even lower, especially in exposure as
compared to affect management, in line with recovery rates.
Pre and post treatment symptom level
The mean pretreatment score on the CAPS was around 70 (range 64–86) and
indicated severe PTSD symptom levels. Post treatment the mean score for
completers of active treatment conditions was 34 (range 9–67) (tables 3 and 4).
The PTSD (B) studies active conditions reached a favorable mean post score of 19,
whereas the Complex PTSD (A) studies had a mean post score of 38 for completers
(Table 5), which is about the cut of score for the PTSD diagnosis, while in intentionto-treat levels these scores were even higher.
This means that the predominantly CBT treatments studied are effective in
reducing PTSD in the CA related PTSD population, but do not meet the needs of
the Complex PTSD population sufficiently.
Complex PTSD outcome measures
Data on Complex PTSD domains could not be aggregated due to limited
measurements and variability in measures used. Outcomes that are measured
in more than one study are dissociation, interpersonal problems, depressive
symptoms and anger. Results were mixed, some improved in the same direction
as PTSD scores, whereas others did not. Thus, although some complex features are
measured, lack of uniformity of measures and consequent measurement hinders
aggregation of data.
treatment for (complex) ptsd: a quantitative review
| 97
Discussion
Our research question was: What evidence is available to effectively treat the
subgroup of child abuse related Complex PTSD patients and to guide our choice of
optimal treatment? Little guidance was found in the empirical literature, because
to our knowledge no review on CA related (Complex) PTSD has been published to
date. Current reviews concerning child abuse or PTSD provide little evidence how
to treat these complex patients best. This was the main reason for the systematic
review described here.
We identified twenty-four RCTs addressing a combination of PTSD and child
abuse. They were highly variable with regard to study population, index trauma,
treatment target, and outcome measures.
Five groups of studies (A-E), varying in target population, were identified. We
selected the studies assigned to the sections A and B. In these studies all patients
were diagnosed with PTSD and all suffered from child abuse as the index trauma.
Moreover, in all patients the treatment target was (Complex) PTSD related to
CA. A subgroup of these studies (section A) included a presumed Complex PTSD
population, which was analyzed separately. Studies assigned to the section C,
focused on PTSD populations which partly suffered from a history of child abuse.
Child abuse was, however, not the index trauma in the majority of the PTSD
patients in category C. In section D, the study populations were characterized by a
CA history and treatment focus. Although a part of these patients was diagnosed
with PTSD, this subgroup was not analyzed separately. Section E pertained to
studies which focused on populations partly suffering from child abuse and partly
diagnosed with PTSD.
The study populations of the selected studies in the categories A and B
predominantly consisted of Caucasian patients, who were mostly well educated,
employed, and severely traumatized. Suicidal patients were excluded frequently.
The included patients were recruited mostly by advertisements. Information about
previous treatments and medication use lacked in most studies.
Results of the meta-analysis showed that these patients improved substantially
from different types of treatments targeting CA-related (Complex) PTSD,
as indicated by large effect sizes. Patients in active treatment conditions,
predominantly cognitive behavioral treatments, showed superior outcomes in
effect sizes, recovery and improvement rates compared to control conditions.
However, post treatment scores were substantial and only around half of patients
no longer met criteria for a PTSD diagnosis. In addition, only a minority showed
clinically relevant improvement. CBT and PCT were generally equally effective,
only differing in dropout rate favoring PCT, and recovery rate favoring CBT in
completers analysis. CBT treatments ranged from affect management to exposure,
98 |
treatment for (complex) ptsd: a quantitative review
always including psycho-education and cognitive therapy. Between types of CBT,
exposure showed a superior effect sizes to affect management in the completers
analysis. In the intention-to-treat analysis, however the superiority of exposure over
affect management disappeared, suggesting that patients either improve well or
dropout. The superior effect of exposure over affect management in completers
analysis was not supported by recovery and improvement rates.
When we compared CBT treatment outcome after 12 to 24 weeks for the most
Complex PTSD populations (assigned to category A) with PTSD populations
(assigned to category B), we found that the Complex PTSD population benefitted
less. Moreover, the superior completer effect sizes for exposure treatments were
not found in Complex PTSD studies, where no differential treatment effect sizes
were observed. In contrast, for Complex PTSD patients, in the completers as well as
the intention-to-treat analysis, more favorable dropout, recovery and improvement
rates for affect management was observed compared with exposure treatment.
This suggests that despite the large effect sizes which are in line with earlier PTSD
reviews, the post scores, improvement and recovery rates indicate a clear need for
better treatments. This is even more the case for the Complex PTSD population in
which the results are less favorable as compared to the PTSD population. This is in
line with the finding in the meta-analysis of Bradley et al. (2005) showing that lower
effect sizes are related to more exclusion criteria.
Additionally, a substantial proportion of patients drops out of treatment, without
relieve of their disturbing symptoms, while the remaining patients achieve large
effects in completer analysis. Our findings indicate the relevance of comparing
results between (sub)populations, which may be illustrated by a difference in dropout between exposure and affect management in the Complex PTSD population,
while they did not differ significantly in the PTSD populations. Probably, patients
who are able to tolerate exposure treatment, will follow treatment successfully.
The differential dropout rates between PCT (9%) and CBT (26%) as well as between
exposure and other treatments stress the importance of additional intentionto-treat analysis to obtain more balanced results. These findings concur with
findings that more complex patients (with comorbid personality disorder) show
higher dropout during exposure therapy (McDonnagh et al., 2005), while within
a Complex PTSD population the least complex patients (without comorbid
borderline personality disorder) dropout more frequently from affect management
(Dorrepaal et al., 2012a). In the literature it has been noted that dropout risk may be
highest in the first exposure sessions (McDonnagh et al, 2005) and may decrease
after improvement of negative mood regulation or better working alliance (Cloitre
et al., 2002).
treatment for (complex) ptsd: a quantitative review
| 99
It is unknown if treatment outcome can be improved by varying treatment type and
duration, since most treatments studied to date are 12-24 week CBT treatments. An
integrated approach focusing not only on PTSD outcome but also on (complex)
associated features is therefore considered relevant. For instance a treatment
schedule starting with affect management was shown to be both endurable as
well as effective (Cloitre et al., 2002; Cloitre et al., 2010), although not tested yet in
a fully diagnosed Complex PTSD population, nor separately analysed for the PD
subgroup.
Strenghts and limitations
In contrast to most PTSD meta-analyses, which are only based on the results of
completer analyses, we additionally presented aggregated intention-to-treat
data, and when necessary estimated them based on dropout rates and completer
results. This is of relevance since dropout rates differed between treatments.
Additionally, we calculated multimodal outcome measures, presenting not only
effect sizes but also recovery and improvement rates, inclusion rates and post
scores, resulting in a more comprehensive overview. Moreover, we attempted to
focus on well-circumscribed populations, trying to avoid conclusions that would
be over-generalizing and unspecific. However, even within the Complex PTSD
study populations some characteristics like inclusion rates and exclusion criteria
indicated different levels of complexity/ severity.
Due to our stringent inclusion criteria the number of identified studies was modest,
limiting the likelihood to identify differential treatment or population effects.
Moreover, we did not analyze follow-up data, which might provide important
information, for instance in the Cloitre et al. (2010) study most differential results
did not show before follow up. We were only able to analyze treatment outcome
in terms of PTSD symptom severity, which is likely to introduce a bias with regard
to patients’ perceived needs and treatment effects. A meta-analysis on CA studies
(Taylor & Harvey, 2010) for instance found CBT treatments superior in terms of PTSD
outcome, but not on externalizing problems. Lastly, we defined Complex PTSD
study populations arbitrarily as Complex PTSD diagnosed or PTSD with at least
50% PD comorbidity, so the possibility that patients in a study population meeting
the latter criteria did not all have Complex PTSD cannot be ruled out. Given the
low inclusion rate, the presence of exclusion criteria like suicidality and in these
two presumed Complex PTSD studies (McDonnagh et al., 2005; Cloitre et al., 2010),
and the fact that most patients were self-referred, Caucasian, well educated and
employed (except 7) indeed warrant some caution about the generalizability of
the results, especially for exposure, to the most Complex PTSD population.
100 |
treatment for (complex) ptsd: a quantitative review
Future research
Trials in CA related Complex PTSD populations are needed with well assessed
and deliberately searched and targeted Complex PTSD populations and minimal
exclusion criteria. The main question is if more favorable effects and low dropout rates for this population are possible using established approaches for
regularly included study populations, such as cognitive (processing) therapy
versus exposure versus eye movement and desentisation reprocessing (EMDR).
More personality disorder/ Complex PTSD oriented treatment programs may be
good candidates for comparisons. To investigate possible treatment x population
interactions, a study comparing treatments having the same number of sessions
within a Complex PTSD diagnosed population is warranted. This could also add to
the results of Cloitre et al. (2010) who compared 8 sessions of affect management
with 8 sessions of exposure with 16 sessions both of these modalities successively.
It remains unclear whether extending the 8 session stabilizing affect management
condition to 16 sessions would result in similar results as compared to the 16
session combined treatment or 16 sessions exposure only.
For generalizability of results it is important to broadly include referred populations
with minimal exclusion criteria like suicidality behavior or substance abuse, which
are characteristic for the Complex population. In addition, personality disordered
non-white, less educated, unemployed, medicated and previously treated
patients should be included. This important gap of knowledge identified in this
review confirms earlier publications (Bradley et al. 2005; Cloitre, 2009; 2011). Axis II
assessment and separate analysis on patients with and without Axis II are needed
to address questions about generalization of effects to more Complex populations.
Moreover differential dropout and analysis on both completers as well as intentionto-treat are warranted to acquire a balanced picture.
Concerning measurement of treatment outcome, till now a wide range of
measures was used to address complex features limiting comparability. Therefore,
more uniform measures for Complex PTSD symptoms are needed. It may turn
out that profiling based on severity of symptom clusters like intrusions, affect
dysregulation, interpersonal functioning or self-esteem will be important in
indicating for treatment components, as opposed to all DSM–IV comorbidities.
Concluding, the results of this review suggest that a variety of treatments may be
effective for CA related PTSD, but that they may not suffice to obtain adequate end
states in the more Complex PTSD populations. Therefore, it is important firstly to
be able to differentiate properly between ‘simple’ and ‘complex’ PTSD populations
and secondly to compare the effect of different combinations and sequences of
a variety of treatment modalities in well-established Complex PTSD populations.
treatment for (complex) ptsd: a quantitative review
| 101
Reference list meta analysis
1.
Zlotnick, C., Shea, T. M., Rosen, K., Simpson, E., Mulrenin, K., Begin, A., Pearlstein, T. (1997).
An affect management group for woman with posttraumatic stress disorder and histories
of childhood sexual abuse. J Trauma Stress, 10, 425–36.
2.
Classen, C. Koopman, C., Nevillmanning , K., & Spiegel, D. (2001). A Preliminary Report
Comparing Trauma-Focused and Present-Focused Group Therapy Against a Wait-Listed
Condition Among Childhood Sexual Abuse Survivors with PTSD. Journal of Aggression,
Maltreatment & Trauma, 4, 2, 265–288.
3.
Cloitre, M., Koenen, K. C., Cohen, L. R., & Han, H. (2002). Skills training in affective and
interpersonal regulation followed by exposure: a phase-based treatment for PTSD related
to childhood abuse. Journal of Consulting & Clinical Psychology, 70, 1067–1074.
4.
Chard, K. M. (2005). An evaluation of cognitive processing therapy for the treatment of
posttraumatic stress disorder related to childhood sexual abuse. J Consult Clin Psychol, 73,
965–971.
5.
McDonagh, A., Friedman, M., McHugo, G., Ford, J., Sengupta, A., Mueser, K. et al. (2005).
Randomized trial of cognitive-behavioral therapy for chronic posttraumatic stress disorder
in adult female survivors of childhood sexual abuse. Journal of Consulting & Clinical
Psychology, 73, 515–524.
6.
Cloitre, M., Stovall-McClough, K. C., Nooner, K., Zorbas, P., Cherry, S., Jackson, C. L. et al.
(2010). Treatment for PTSD related to childhood abuse: a randomized controlled trial.
American Journal of Psychiatry, 167, 915–924.
7.
Dorrepaal, E., Thomaes, K., Smit, J. H., van Balkom, A. J., Veltman, D. J., Hoogendoorn, A.
W. et al. (2012b). Stabilizing Group Treatment for Complex Posttraumatic Stress Disorder
Related to Child Abuse Based on Psycho-education and Cognitive Behavioural Therapy: A
Multisite Randomized Controlled Trial. Psychother. Psychosom., 81, 217–225.