Title: The relationship between childhood adversities and dissociation in severe mental

illness: A meta-analytic review.

Authors: Sonya Rafiq a, Carolina Campodonico a, and Filippo Varese a,b

a School of Health Sciences, Division of Psychology and Mental Health, Faculty of Biology,

Medicine and Health, Manchester Academic Health Science Centre, The University of

Manchester. 2nd Floor, Zochonis Building, Brunswick Street, Manchester. M13 9PL, UK.

b Complex Trauma and Resilience Research Unit, Greater Manchester Mental Health NHS

Foundation Trust, Manchester (UK).

Corresponding author:

a Filippo Varese. School of Health Sciences, Division of Psychology and Mental Health,

Faculty of Biology, Medicine and Health, Manchester Academic Health Science Centre, The

University of Manchester. 2nd Floor, Zochonis Building, Brunswick Street, Manchester. M13

9PL, UK.

Email: filippo.varese@manchester.ac.uk

Telephone: 0161 306 0434

Running title: Childhood trauma and dissociation: Meta-analysis

This paper was published in Acta Psychiatrica Scandinavica

1

Abstract

Objective: Several studies have observed that dissociative experiences are frequently

reported by individuals with severe mental illness (SMI), especially amongst patients that

report a history of adverse/traumatic life experiences. This review examined the magnitude

and consistency of the relationship between childhood adversity (sexual abuse, physical

abuse, emotional abuse, neglect, bullying, natural disasters and mass violence) and

dissociation across three SMI diagnostic groups: schizophrenia, bipolar disorder and

personality disorders.

Method: A database search (EMBASE, PubMed, and PsycINFO) identified 30 eligible

empirical studies, comprising of 2199 clinical participants. Effect sizes representing the

relationship between exposure to childhood adversity and dissociation were examined and

integrated using a random-effects meta-analysis.

Results: The results indicated that exposure to childhood trauma was associated with

heightened dissociation across SMIs. Positive significant associations were also found

between specific childhood adversities and dissociation, with aggregated effect sizes in the

small-to-moderate range.

Conclusion: These findings support calls for the routine assessment of traumatic experiences

in clients with SMIs presenting with dissociative symptoms, and the provision of adequate

therapeutic support (e.g. trauma-focused therapies) to manage and resolve these difficulties.

Keywords: Trauma, dissociation, schizophrenia, bipolar disorder, personality disorder.

2

Summations

Small-to-moderate relationship between childhood trauma and dissociation across

SMIs.

Medium relationship between childhood trauma and dissociation in schizophrenia-

spectrum disorders.

All forms of childhood traumatic experiences (e.g. sexual abuse, physical abuse,

emotional abuse, emotional neglect, physical neglect) were found to be associated

with dissociation in people diagnosed with SMI.

Considerations

Only three studies were found eligible for the bipolar sample (i.e. low power),

therefore no strong conclusions can be drawn regarding the association between

childhood trauma and dissociative experiences in bipolar disorder.

The relationship between childhood trauma and dissociation in SMI requires further

corroboration using longitudinal designs to determine causality and directionality.

No firm conclusions can be drawn regarding which specific traumatic experience is

most strongly associated with dissociation, as the studies included in the analysis were

not independent (i.e. effects for sexual abuse and physical abuse estimated from the

same sample of participants).

Introduction

3

Dissociation has been defined by DSM-5 (1) as “a disruption of and/or discontinuity

in the normal integration of consciousness, memory, identity, emotion, perception, body

representation, motor control, and behaviour,” (1; p. 291). Dissociation can manifest as a

range of anomalous experiences, including dissociative amnesia, (e.g. inability to access

information about oneself/salient episodes of one’s life, which is qualitatively distinct from

mere forgetting) absorption, (e.g. losing awareness of one’s surroundings/passage of time

whilst being highly absorbed in certain activities) identity alteration, (e.g. experiencing one’s

self as comprising multiple distinct personas) derealisation, (e.g. feeling as if the world is

unreal) and depersonalisation (e.g. feeling disconnected/detached from one’s body/emotions;

(2, 3).

Dissociative experiences are common in the immediate aftermath of potentially

traumatic life experiences, and many researchers and clinicians regard dissociation as a

defence mechanism, that protects individuals against the overwhelming distress caused by

certain adverse life experiences (4). For example, at the time of a traumatic event,

dissociation allows for the individuals to “detach” from the source of distress, enabling them

to adapt and protect their internal and/or physical selves (4-6). However, the initial ‘adaptive’

response may become maladaptive over time, for example when used automatically or

indiscriminately in a variety of circumstances as a response to stress or reminders of trauma

(5), therefore, potentially impacting on the persons’ functioning and preventing the

integration of factors that may promote long-term recovery and well-being (e.g. the

correction of maladaptive trauma-related beliefs; 5, 7).

Growing empirical evidence has indicated that dissociative experiences are pervasive

within clients with SMIs; heightened dissociation is commonly reported by individuals who

received diagnoses of psychosis, (8, 9) bipolar disorder, (10) and personality disorders (11,

12). In line with evidence suggesting that dissociation is a common sequela of adverse and/or

4

traumatic life experiences, many empirical studies have demonstrated that individuals with

SMIs (13-16) exposed to childhood adversity (e.g. sexual abuse, physical abuse, emotional

abuse, neglect, bullying, etc.), experience greater dissociation compared to individuals

exposed to little or no trauma (17-21). However, the relationship between childhood trauma

and dissociation in individuals with SMIs remains contradictory, as some studies have found

no association between the two constructs (22, 23).

Several meta-analyses have attempted to investigate and clarify the relationship

between childhood trauma and dissociation across clinical and non-clinical individuals (4, 7).

However, previous evidence synthesis has largely neglected individuals with SMIs, either

because of the lack of sufficient empirical evidence at the time these syntheses were

conducted (7), or the use of study selection strategies that precluded the investigation of

dissociation in the context of specific mental health presentations (4). Also, previous reviews

have only considered a very limited range of traumatic exposures (i.e. physical abuse and

sexual abuse; 4, 7), consequently excluding the examination of other potentially traumatic

experiences that can plausibly impact on dissociation. Therefore, most of the empirical

studies that have investigated the relationship between childhood maltreatment and

dissociation have not been systematically reviewed.

Aims of the study

This meta-analysis examined and summarised evidence for an association between

childhood trauma and dissociation and considered the consistency of this relationship across

different SMIs (i.e. schizophrenia, bipolar and personality disorder). A secondary aim was to

investigate the relationship between specific types of childhood adversities and dissociation

and to determine the magnitude of these relationships across SMIs.

Method

Search procedure

5

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses

(PRISMA) guideline (24), alongside our PROSPERO review protocol (25) were followed for

the current meta-analysis (there was a deviation from the protocol in regards to quality

assessment, as it was not conducted because most quality assessment tools such as the

Effective Public Health Practise Project, are poorly suited to evaluate the population of

studies within this review). In the present review, we opted to employ the definition of SMI

(e.g. psychosis, personality disorder and bipolar disorder) employed in the UK as part of the

Improving Access to Psychological Therapies for SMI project (26). Additionally, the search

strategies and study selection criteria of previous relevant meta-analyses (27-31) were

examined to inform the selection of childhood adversities considered in the current evidence

synthesis, (i.e. physical abuse, sexual abuse, emotional abuse, physical neglect, emotional

neglect, natural disasters, mass violence and bullying) the eligible operational criteria used to

define dissociation, the diagnostic groups investigated (i.e. schizophrenia-spectrum disorders,

personality disorders and bipolar disorder), as well as the search terms used to identify

relevant papers considering the constructs of interest.

PubMed, EMBASE, and PsycINFO, were systematically searched between 1986 up

to and including October 2016 using the following search terms: Dissociation combined with

trauma related search terms and severe mental health related search terms (see supplementary

material for the complete list of search terms). Medical subject headings (MeSH) in PubMed

and subject headings in EMBASE and PsycINFO were used to further expand the literature

search.

To reduce file drawer effects the following steps were taken to identify all relevant

studies: 1) reference lists of eligible studies (i.e. backward search) and articles that cited

eligible studies (i.e. forward search) were scanned to locate studies not identified in the

6

database search; and 2) authors of all eligible studies were contacted regarding any relevant

unpublished studies (see supporting information; Table S1).

Inclusion and exclusion criteria

Studies published after 1986 were included as the first validated and widely used

measures of dissociation (e.g. DES) was published in 1986 (29). Also, studies utilising the

subsequent quantitative methodologies were eligible: 1) between-group comparisons

contrasting traumatised individuals (e.g. exposed to an eligible traumatic event, i.e. physical

abuse, sexual abuse, emotional abuse, physical neglect, emotional neglect, natural disasters,

mass violence and/or bullying) with non-traumatised comparable controls (e.g. patients with

identical diagnosis) on measures of dissociation; 2) between-group comparisons contrasting

individuals with high dissociation (e.g. scored in the pathological range of a dissociation

measure) with low dissociation comparable controls on measures of trauma; 3) correlational

studies examining the relationship between dissociation and trauma in an eligible clinical

sample or; 4) longitudinal studies investigating the temporal relationship between trauma

exposure and subsequent presence and/or severity of dissociative experiences. Studies written

in English, Spanish, Italian, Portuguese and Urdu were also eligible for inclusion (researchers

were bilingual). Moreover, studies had to employ: 1) validated self-report measures of

dissociation and; 2) validated self-report measures of exposure to potentially traumatic

experiences in childhood (i.e. prior to age 18). Furthermore, studies employing the following

diagnostic populations were included: 1) schizophrenia and related psychotic disorders; 2)

personality disorders or; 3) bipolar disorder based on ICD-9, ICD-10, DSM-III, DSM-III-R,

DSM-IV, DSM-IV-TR or DSM-5 diagnostic criteria.

Studies were excluded based on the subsequent criteria: 1) non-quantitative

methodology (e.g. qualitative studies, case studies, etc.); 2) measures of dissociation were

staff administered instead of self-report; 3) dissociative symptoms were either primarily drug

7

or alcohol induced or a medical (e.g. epilepsy) or organic cause was attributable; 4) non-

clinical sample; or 5) insufficient information was provided to calculate an effect size.

Eligibility was assessed independently by two researchers following a two-stage

procedure: title and abstract screening, and whole paper screening. In the first stage, S.R.

screened all the titles and abstracts, and C.C. screened 20% of all title and abstracts

(agreement ratings = 93.3%, adjusted kappa = .87). In the second stage, both S.R. and C.C.

reviewed all papers independently (agreement ratings = 96.0%, adjusted kappa = .92). Any

inter-coder discrepancy was resolved during consensus meetings and consultations with a

third researcher (F.V.).

Quality assessment

Eligible studies were quality assessed using the Effective Public Health Practise

Project tool (EPHPP; Thomas, 2003). The EPHPP has been found to be valid (Thomas et al.

2004) and reliable (Armijo-Olivio et al. 2012). The components included in the tool are: 1)

selection bias; 2) study design; 3) confounders; 4) blinding; 5) data collection methods and;

6) withdrawals and drop-outs (given that all studies were cross-sectional, this component was

found to be not applicable for the current analysis). Each component was rated as either

strong, moderate or weak. A global rating of strong (no weak rating), moderate (one weak

rating) or weak (two or more weak ratings) was given based on component ratings. All

studies were quality assessed by S.R and monitored by F.V; disagreements were discussed

and resolved during meetings

Effect size computation and statistical analyses

A series of random-effects meta-analyses were carried out using Comprehensive

Meta-Analysis, version 2. The random effects model was chosen as it allows for

heterogeneity and generalizability of results across studies (32). Pearson’s r was selected as

the main effect size metric for all analyses as most of the eligible studies reported

8

correlational effects. In the case of studies reporting correlational effects, correlation

coefficients and sample sizes on which correlations were estimated were extracted from the

primary studies. In the case of studies reporting between-group effects, the mean, standard

deviation and sample size were used to generate Cohen’s d, which were then converted to

effects of the r-family using appropriate computational methods (32). Similarly, when binary

data effects were reported, odds ratios (ORs) were estimated from appropriate descriptive

statistics (e.g. 2 x 2 tables), and then converted to r-family effects (32).

A number of studies included in the meta-analysis used different measures of trauma

and dissociation and reported both correlation and between-group analyses. To ensure

consistency and comparability across studies, and to avoid issues stemming from

dependency, a coding hierarchy was developed to guide the extraction of statistical

information. The following hierarchy was utilised: 1) when multiple separate effects for

childhood adversities (e.g. sexual and physical abuse) were reported within the same study

and no total trauma scores were provided, authors were contacted to provide information on

the summary score when appropriate, and when no information was provided, a composite

summary effect size was generated by merging the effects prior to the analysis (32); 2) when

measures of adversity exposure included multiple categories (i.e. none, low, moderate, or

severe trauma) we only extracted effects contrasting the no trauma exposure sample to the

most severe trauma sample; 3) when the dissociation scores were divided into different

severities (i.e. low, moderate, or high), the low dissociation sample was compared to the high

dissociation sample; 4) when the relationship between trauma and dissociation was

investigated at two different time points (i.e. baseline and post-assessments), only the

baseline results were selected; and 5) when between group and within group statistical

analyses were reported in the same study, within group information (i.e. correlation or

regression) was extracted; and 6) when multiple studies appeared to be drawn from the same

9

participant sample, the study with the greatest sample size was included to increase precision

(or if studies with overlapping samples had missing data, the study with the relevant data was

included in the meta-analysis).

A series of analyses were carried out to examine the main research questions of this

review. Firstly, analysis of the association between total childhood trauma (or composite

childhood trauma) and dissociation was performed for the whole sample. After the main

analysis, subgroup-analyses were conducted to investigate the differences between

schizophrenia, personality disorder and bipolar disorder (the statistical differences between

studies was also investigated). Secondly, subgroup analyses were conducted for different

types of childhood trauma (e.g. sexual abuse, physical abuse, emotional abuse, physical

neglect or emotional neglect, etc.) and dissociation when examined in primary studies. Again,

subgroup-analyses were conducted to investigate the differences between schizophrenia,

personality disorder and bipolar disorder. However, it was not possible to directly compare

the effects of different types of trauma (i.e. the effect of sexual abuse is stronger than that of

physical abuse) as in most cases these effects were estimated from the same sample of

participants (32).

Heterogeneity was examined in all analyses using the Q and I2 statistic, to determine

if heterogeneity was present, and to what degree the amount of statistical inconsistency

between studies existed. Following heterogeneity analysis, publication bias was assessed by

visual inspection of the funnel plots, and Egger’s test for funnel plot asymmetry (33) was

carried out to verify the influence of unpublished studies (i.e. no publication bias if studies

are located symmetrically around the mean effect size; 32). Duval and Tweedie’s, trim-and-

fill method (34) was applied to both the overall analysis and subgroup analyses when

appropriate to correct for the presence of publication bias. Furthermore, sensitivity analyses

10

(e.g. one study removed analysis, sub-group analyses) were conducted to determine the effect

of potentially influential studies on the meta-analytic results.

Results

Summary of studies

Figure 1 includes a flow chart summarising the search strategy for our review. The

overall number of studies retrieved from database searches totalled 6245. After title and

abstract screening, 332 full papers were screened, and 30 studies were included in the meta-

analysis. However, 31 effect sizes were extracted, as Braehler et al. (8) considered the

relationship between childhood adversity and dissociation in two separate diagnostic groups

of interest.

Demographic characteristics of eligible studies

The total number of participants included in the review totalled 2199. Most of the

studies had participants with a diagnosis of schizophrenia-spectrum disorders (n = 1192),

followed by personality disorder (n = 630), bipolar disorder (n = 303) and a mixed sample (n

= 74). The mean age was 35.28 (with information regarding age not reported for one study).

The ratio of male to females was 1156: 1108. Participants were recruited from a range of

countries including: USA (n = 630), Turkey (n = 373), Canada (n = 264), Germany (n = 263),

South Korea (n = 260), UK (n = 207), Spain (n = 116), Austria (n = 52), and Australia and

the UK (n = 34).

Measures of dissociation and childhood trauma

11

The details regarding measures of dissociation and childhood adversity are displayed

in Table 1. Seven different measures of dissociation were used to investigate the concept of

dissociation. The most commonly administered measures were the DES (k = 23) and

variations of the DES (DES-II, k = 2 and DES-T, k = 2). Furthermore, ten measures of

trauma were used to investigate childhood trauma. The most commonly administered

measures of childhood trauma were the CTQ (n = 16) and variations of the CTQ scale, i.e.

CTQ-SF (n = 6). Additionally, the specific types of adversity measured by the scales included

childhood sexual, physical and emotional abuse, and childhood emotional and physical

neglect. We found no eligible studies investigating the association between dissociative

experiences and exposure to mass violence, natural disasters or bullying in childhood.

Quality assessment

The global quality assessment ratings are presented in Table 1. The majority of

studies fell into the weak range (k = ), the remaining studies were rated as moderate (k = ). No

study achieved a strong global rating (component ratings displayed in Table S2). These

global quality ratings were affected by methodological limitations common in cross-sectional

literature (and quasi-experimental studies), specifically selection bias, study design and a

limited control of confounding variables

Design characteristics of eligible studies

The meta-analysis included between-group (n = 7) and correlational (n = 24) designs

to investigate the relationship between trauma and dissociation. No longitudinal design

studies were considered eligible for analysis, as the studies located did not meet the eligibility

criteria (i.e. did not investigate the relationship between childhood trauma and dissociation).

[Insert Figure 1 approximately here]

[Insert Table 1 approximately here]

Statistical analysis of the overall sample

12

The first set of analyses examined the relationship between total childhood trauma

and dissociation in the overall sample (i.e. irrespective of whether specific studies considered

samples of individuals with diagnoses of schizophrenia-spectrum disorders, personality

disorders or bipolar disorder). The results of the analysis are presented as a forest plot in

Figure S2 (see Figure 2). The analysis demonstrated a significant association between

exposure to potentially traumatic experiences in childhood and dissociation; k = 31, r = .33,

95% CI (.27, .39), p < .001. Based on Cohen’s (79) criteria (i.e. r = .10 indicates a small

effect; r = .30 indicates a medium effect and; r = .50 indicates a large effect), the summary

effect corresponded to a “medium-sized” association (when interpreting effect sizes, we urge

the reader to be cautious in associating larger effect sizes with more value, given the arbitrary

nature of criteria to evaluate the magnitude of summary effect sizes; 80).

Additionally, the Q and I2 analyses (Q = 97.50, p < .001, I2 = 69.23%) suggest a

significant amount of heterogeneity exists. Statistical heterogeneity was in the moderate

range according to widely used criteria to evaluate the amount of statistical inconsistency in

meta-analytic findings (i.e. I2: 25% = low, 50% moderate, 75% = high heterogeneity; 81).

However, the interpretation of I2 may be misleading given the arbitrary nature of the criteria.

[Insert Figure 2 approximately here]

Statistical analyses of clinical groups

The second set of analyses considered the association between childhood adversity

and dissociation within different diagnostic groups. Whereas the association between

childhood adversity and dissociation was found to be robust and significant in studies with

patients diagnosed with schizophrenia-spectrum (k = 20, r = .39, 95% CI [.31, .46], p < .001)

and personality disorders (k = 7, r = .24, 95% CI [.19, .29], p < .001), no significant

relationship was found in studies considering patients with bipolar disorder (k = 3, r = .15,

95% CI [-.04, .32], p = .114). Based on Cohen’s criteria (79), the relationship between

13

childhood adversity and dissociation appeared more robust within schizophrenia-spectrum

disorders studies (i.e. a medium summary effect size) than studies considering the other

diagnostic groups (i.e. small summary effect sizes). This was corroborated by a Q test

examining the difference between summary effects, which indicated that the magnitude of the

relationship significantly differed across clinical groups; Q = 10.03, df = 2, p = .007 (when

bipolar disorder is removed from analysis the significant difference between groups remains;

Q = 4.73, df =1, p = .030)

The Q test for schizophrenia-spectrum disorders was significant and the I2 statistic

indicated that heterogeneity was in the moderate range (Q = 50.06, p < .001, I2 = 64.04%).

Regarding personality disorders, no significant amount of heterogeneity was found (Q = 6.59,

p = .360, I2 = 9.01%). However, this result may be unreliable as the Q test may be biased in

meta-analyses that include a small number of studies (32).

Association between specific childhood adversities and dissociation across all SMIs

The second set of analyses investigated the relationship between specific childhood

adversities and dissociation within the combined SMI sample (see Table 2). The results

indicated that all types of childhood adversities for which it was possible to extract specific

effects, were positively associated with dissociation, with summary effect sizes within the

small-to-moderate range. Moderate statistical heterogeneity was observed in the analyses

considering sexual abuse, physical abuse, emotional abuse and aggregated neglect.

Associations between specific childhood adversities and dissociation within specific

diagnostic groups

The third set of analyses investigated the relationship between specific childhood

adversities and dissociation within different clinical groups. In the analyses considering the

14

schizophrenia-spectrum studies (see Table 2), all types of childhood adversity were

significantly associated with higher dissociation, with the notable exception of neglect

(results may be due to a few studies examined). Additionally, sexual abuse and aggregated

neglect demonstrated a medium amount of heterogeneity, while emotional abuse displayed

high levels of heterogeneity.

In the analyses focusing on personality disorders, small but significant associations

were identified between all types of childhood trauma investigated and dissociation, with the

exception of emotional neglect (see Table 2). Furthermore, aggregated neglect demonstrated

a significant moderate amount of heterogeneity, but heterogeneity statistics for this set of

analyses should be interpreted with caution given the small number of studies that examined

associations between specific types of adversities and dissociation in individuals with

personality disorder. No statistical integration was carried out for bipolar disorder as there

were too few studies for the analysis to be meaningful (32).

[Insert Table 2 approximately here]

Publication bias analyses

The results of our publication bias analyses, indicated the possible presence of

publication bias in the analysis examining the relationship between total childhood trauma

and dissociation within the overall sample (Egger’s test p = .012). As a result, Duval’s and

Tweedie’s trim-and-fill method was applied to correct for this potential bias. When seven

hypothetical studies were included in the analysis, the summary effect size was reduced, but

remained substantial; (r = .28, 95% CI [.22, .34]). Further analyses found no evidence of

publication or other selection bias for the analyses reported (see Table S2).

Sensitivity analyses

One study removed analyses found no evidence of potential influential cases across

the analyses reported (except for the bipolar subgroup; the relationship between childhood

15

trauma and dissociation became significant after the removal of Yilmaz et al., 2016). As a

few studies reported only significant relationships (i.e. selective reporting) between certain

specific childhood adversities and dissociation (12, 82, 83), sensitivity analyses were

conducted by excluding these studies from all analyses. This did not alter the pattern of

findings reported earlier. Furthermore, two studies contained childhood adversity measures

that examined both eligible and non-eligible (e.g. witnessing domestic violence, loss, etc.)

childhood traumatic experiences (43, 84). When the studies were excluded, the significant

relationships between childhood adversity and dissociation in the overall sample and

schizophrenia-spectrum studies were consistent with those reported in our original analyses

(see supporting information; Table S3 for all further sensitivity analyses and statistics).

Finally, to ensure that the design of the eligible studies did not influence our meta-

analytic findings, an additional subgroup analysis contrasting correlational and between-

group effect sizes was carried out for meta-analyses conducted on the overall sample. No

significant difference (Q = .05, df = 1, p = .830) was found between the summary effects of

studies which used between-group (k = 7, r = .32, 95% CI (.23, .41), p < .001) and

correlational designs (k = 24, r = .33, 95% CI (.25, .41), p < .001).

Discussion

Summary and discussion of the findings

The primary aim of the present meta-analysis was to examine the relationship

between childhood adversity and dissociation across three SMI diagnostic groups (i.e.

schizophrenia-spectrum disorders, personality disorders and bipolar disorder). Our findings

indicated a robust small-to-medium relationship between childhood adversity and

dissociation, when the effects of all studies which considered SMI patients were aggregated

in a single analysis. The magnitude of this relationship is broadly consistent with that

16

observed by the previous meta-analyses of the link between traumatic life experiences and

dissociation in other clinical and non-clinical populations (4, 7). The magnitude of this

relationship was found to differ across diagnostic groups according to widely used criteria to

evaluate effect sizes (79). Whilst a moderate and significant relationship was found within

schizophrenia-spectrum disorders studies, the association between childhood adversity and

dissociative experiences was somewhat smaller in studies with individuals who received

personality disorder diagnoses. Furthermore, the results found no evidence of an association

between life adversities and dissociative experiences in individuals with bipolar disorder.

Whilst, this negative finding might be in part due to the very small number of eligible bipolar

studies, this finding parallel those of recent meta-analytic synthesis, indicating that

dissociative experiences are not a prominent complaint of clients with diagnoses of bipolar

and related disorders (85), despite the high levels of childhood adversity observed in the

clinical group (86).

In regards to the secondary aim of the present review (i.e. examine the relationship

between specific types of childhood adversity and dissociation across different SMIs), it is

notable that our search strategy did not identify any eligible studies considering the

association between dissociation and bullying, mass violence and conflict exposure, despite

the fact that in previous research these potentially traumatic experiences have been linked to

both SMIs and stress-and trauma-related symptoms, including dissociative experiences. The

findings of our meta-analytic integrations indicated that all other types of childhood

adversities considered in this evidence synthesis showed a significant relationship with

dissociation. Across all diagnostic samples, the association between emotional abuse and

dissociation was particularly robust. While conclusions cannot be drawn as to which

childhood adversity might be most associated with dissociation based on our findings (e.g.

due to non-independent nature of the sample of participants), primary studies that employed

17

multivariate analyses to test the relative contribution of different life adversities have often

found that emotional abuse is one of the most robust predictors of dissociative experiences (8,

36). This may be explained by the interpersonal nature of emotional abuse. Repetitive

childhood emotional abuse in the form of verbal insults, criticism, humiliation and isolation,

may be seen as more threatening compared to other types of abuse, as emotional abuse is

often perpetrated by somebody whom the victim is close to and is reliant upon for protection

and support (87). It has been proposed that the contradictory nature of an emotionally abusive

“caregiver” (being at the same time the source of abuse and comfort/protection/survival) may

be particularly conductive to dissociative responses that protect the survivors’ internal selves

whilst maintaining attachment to significant others in order to ensure survival (87, 88).

In many cases, our findings demonstrated considerable statistical heterogeneity. Thus,

the reported summary effect sizes should be interpreted with some caution, as guidance (e.g.

GRADE approach) suggests that considerable statistical heterogeneity decreases the quality

of evidence (insert REF) and indicates t the “true” effect may be somewhat smaller or larger

than indicated in the current meta-analysis (32). Given the noticeable methodological and

clinical heterogeneity of the included studies, the results of the heterogeneity analyses are not

surprising. The quality of included studies may partly explain the heterogeneity reported. The

majority of studies quality assessed using the EPHPP achieved weak global ratings (k =); this

may affect the meta-analytic evidence supporting an association between childhood trauma

and dissociation in SMI. Studies were consistently rated as weak because most of the eligible

studies were correlational, did not take confounding variables into account and were

vulnerable to selection bias. Additionally, components of the EPHPP, in particular selection

bias were rated as weak because the necessary information required to assess bias was not

present. Per EPHPP guidance, inferences about what the authors intended to do could not be

made, and instead judgements were required to be made using the information contained in

18

the study. However, while studies may not be entirely transparent regarding methodology

(e.g. selection bias, blinding, etc.) due to word count restrictions in published papers, it does

not necessarily imply that the method was not used in the study (Mhaskar, Djulbegovic,

Magazin, Soares, & Kumar, 2012). Therefore, the results of quality assessment may be

somewhat conservative in reflecting the true methodological quality of included studies.

A series of subgroup and sensitivity analyses were conducted in an attempt to explain

the observed statistical heterogeneity. The analyses showed that the inconsistency between

studies is unlikely to be due to study design, as the effect sizes did not substantially differ

across correlational and between-group designs. However, the relationship between

childhood adversity and dissociation substantially varied across diagnostic groups, which

may explain some of the statistical inconsistency observed in the overall analysis. As a wide

range of empirical studies was included in the meta-analysis, other unmeasured variables not

measured in the current study may account for the observed heterogeneity. Factors such as

the detailed features of the potentially traumatic events under scrutiny (e.g. timing, frequency,

severity, relationship to the abuser, etc.) affect the trauma and dissociation relationship. For

example, individuals abused by parents (vs others e.g. relatives, strangers or friends), or those

exposed to more than one type of abuse demonstrated higher dissociative scores (48, 89), and

as the number of perpetrators increase; the greater the dissociative symptoms experienced by

the victim (90). Additionally, socio-demographic characteristics of the respondents (e.g. age,

gender, ethnicity, etc.) may plausibly moderate the association between trauma exposure and

dissociative experiences. For instance, age was found to be a significant predictor of

dissociation (13), and previous research has found that the relationship between trauma and

dissociation varied across ethnicities (91) and gender (92). Given that only a small amount of

studies explored such factors, it was not possible to examine the contribution of those

moderators systematically using meta-analytic methods. While many possible sources of

19

heterogeneity were not controlled for, some of the primary studies accounted for a range of

factors shown to be related to self-reported dissociative experiences, such as substance abuse,

cognitive functioning and medical conditions such as epilepsy (92-94). As the corpus of

evidence pertaining to the relationship between trauma and dissociation in SMIs is

increasing, future systematic reviews may attempt to further explore the impact of these

effect moderators.

An important consideration relevant to the findings of the present review concerns the

conceptualisation and measurement of dissociation in the available literature. Whilst the

majority of studies included in this meta-analysis (and previous meta-analytic reviews of

dissociation studies; 4, 7, 85) used the DES to assess dissociative experiences, several

researchers have criticised the use of this measure on the grounds that dissociation should be

regarded as a multifaceted rather than a unitary construct (2, 95). Critics have argued that the

unitary model of dissociation is over-inclusive and implies that dissociative experiences such

as absorption and derealisation are qualitatively similar and only differ in “amount” of

dissociation (2). Therefore, even though the DES has good psychometric properties and is

easy to administer, its over-inclusive and unitary nature may limit the validity of research

findings gathered using this measure. On the other hand, the proposal that dissociation should

be conceptualised as a multifaceted construct, although theoretically plausible, is yet to

translate into the development of psychometrically sound measures assessing qualitative and

aetiologically distinct forms of dissociation (e.g. dissociative detachment and

compartmentalisation; 2, 95) that could replace the DES. Furthermore, the inclusion in the

DES of items assessing absorption has received some criticism, as this experience is not

necessarily pathological and is common in the general population. Therefore, it has been

argued that the use of the DES might inflate and bias any assessment of “true” dissociative

experiences (4). Despite this, numerous studies have indicated that absorption is robustly

20

associated with other dissociative experiences and other dimensions of psychopathology, (96,

97) and it has been therefore argued that its exclusion from dissociation tools used in both

clinical and research settings may be problematic (4).

A number of additional limitations should be taken into account when interpreting our

findings. All the studies included in the meta-analysis used retrospective measures of

childhood trauma. Some researchers argue that the use of retrospective measures of

childhood trauma inflate the relationship between trauma and dissociation. Also, it has been

claimed that patients may, in fact, exaggerate traumatic memories (98) or confuse fantasy

with factual memories (99). However, Dalenberg et al. (4) found that when using more

objective measures of childhood trauma (e.g. confirmation by therapist, child protective

service reports, etc.), the relationship between childhood sexual abuse and dissociation

remained moderate (r = .30) and similar to that observed in studies where self-report

measures (r = .32) were administered. Furthermore, research suggests that traumatic

experiences measured using self-report measures tend to be under-reported rather than over-

reported (100).

In the present meta-analysis, we did not employ an “off the shelf” quality assessment

tool to evaluate primary studies. Instead, we extracted specific methodological variables and

conducted a range of sensitivity analyses to quantitatively assess for potential biases (e.g.

selective reporting, study design, measures, etc.). Multiple sensitivity analyses were

conducted and the analyses demonstrated no difference in effect sizes after methodological

variables were accounted for .

Alternatively, the omission of certain studies may have biased results. Despite

emailing authors of all eligible studies for any unpublished work they may have; no

unpublished study was located. However, in the current meta-analysis, the trim-and-fill

method was used to adjust for the potential influence of publication or other selection biases

21

in relevant analyses. When bias was accounted for the relationship between childhood

adversity and dissociation did not change substantially. Also, the literature search was

extended to languages other than English in order to minimise selection biases. Future meta-

analytic reviewers should strive to increase the comprehensiveness of the search and include

unpublished studies/reports to reduce the file drawer effect.

In the present meta-analysis, the initial screening phase (title and abstract) was

completed by one researcher. S.R screened all titles, abstracts and full papers, while C.C.

screened 20% of all titles and abstracts and all full papers. Research suggests the assessment

of eligibility at all stages should be completed by at least two researchers to minimise

bias/error (101, INSERT REF). To minimise error during the screening stage, S.R used

backward and forward screening on all 30 eligible studies to ensure the retrieval of all

possible studies.

Another potential limitation is the varied conceptualisation of which psychiatric

diagnoses should be regarded as “SMIs”. It is broadly recognised that the conceptualisation

of SMI lacks specificity and remains inconsistent (102). In the present review, we employed

the definition of SMI used in the UK as part of the Improving Access to Psychological

Therapies for SMI project (26), but we acknowledge that other diagnostic groups may have

qualified for inclusion if other SMI definitions were employed. For example, some

researchers have employed narrower definitions of SMIs (e.g. schizophrenia-spectrum

disorders and bipolar disorder only; 103, 104), whilst others used more inclusive ones (102,

105).

Future research

Regardless of the limitations mentioned above, the findings of this evidence synthesis

suggest that the dissociative experiences reported by individuals with SMIs, can be at least in

part represent psychological sequelae of exposure to adverse life experiences in childhood.

22

However, causality and directionality cannot be ascertained as only between-group and

correlational studies were included, as no eligible longitudinal studies were identified.

Further research is required to clarify the direction of the associations considered in the

present evidence synthesis. Other non-clinical longitudinal studies have demonstrated that

exposure to childhood trauma was associated with heightened levels of dissociation

compared to non-abused individuals (106); therefore, it is plausible that the same direction of

effects may be observed in the context of individuals who also received diagnoses of SMIs.

Also, many of the included studies in the review did not control for potential confounding

factors that may affect the relationship between childhood trauma and dissociation.

Therefore, to separate the effects of trauma from those possibly caused by other variables

(e.g. the sustained disturbances from the parent-child relationship) additional primary

research is required.

In relation to the secondary aim of the analysis, no firm conclusions can be drawn

regarding which traumatic experience is most strongly associated with dissociative

experiences. As the studies included in the analysis were not independent (32). Future studies

should explore further the relationship between specific types of childhood trauma and

dissociation. Using appropriate statistical methods and sufficiently large samples, to

disentangle the relative contribution of qualitatively different childhood adversities as well as,

the interplay between exposure to different adversities (e.g. the effect of

re-victimisation/polyvictimisation) and the impact of important trauma-related factors that

may further aggravate dissociative symptoms (e.g. the length and “severity” of potentially

traumatic experiences). Additionally, given the multifaceted nature of dissociation, future

research should consider the association between trauma/specific traumatic experiences and

specific features of dissociation. We initially planned to investigate those experiences (25),

however, too few studies were located (e.g. 43,44). Therefore, sufficient information could

23

not be extracted to undertake these analyses. Lastly, no studies were considered eligible for

natural disasters, mass violence, and bullying. Future studies should focus on these

adversities as well as, other common childhood adversities in the risk of developing

dissociation, as these traumatic experiences have been found to increase the risk of

developing dissociation in the general population (100, 107-109).

Clinical implications

These findings along with other meta-analytic results (4, 7), highlight the damaging

role of childhood trauma in the development of dissociative experiences. Therefore, it is

advisable for practitioners to routinely and sensitively enquire about abuse history in patients

who display dissociative symptoms (8, 11). Current research suggests that clinicians do not

routinely question individuals regarding trauma (110, 111). This may be due to clinicians’

beliefs about the biological aetiology of mental illness (112), the lack of awareness of the

adverse effects of social factors and life experiences on adult functioning (113), or concerns

that questioning service-users about potentially traumatic events may lead to further distress

and aggravation of symptoms (102). However, research indicates that most trauma survivors,

including those who received diagnoses of SMIs, do not experience any aggravation in their

mental health when asked about past traumatic experiences, and positively evaluate the

opportunity of talking about these experiences (114-116).

Additionally, clinicians should work collaboratively with service users to include and

understand the role played by trauma and dissociation in the development and maintenance of

their presenting difficulties, and incorporate appropriate interventions to ameliorate

dissociation and related difficulties in their treatment plans. Although a number of

psychological interventions for dissociative experiences are available (e.g. cognitive

behaviour therapy for depersonalisation; 117), their evaluation in clients with SMIs is in its

24

infancy (118). On the other end, a growing number of studies in recent years have evaluated

the safety and efficacy of trauma-focused interventions in people with SMIs, with promising

findings (119, 120). Research suggests that trauma-focused interventions can effectively

reduce the severity of dissociative symptoms in trauma survivors (4), and there is evidence

that trauma-focused therapies such as Prolonged Exposure and Eye Movement

Desensitisation and Reprocessing can lead to similar improvements in dissociative

experiences in those with SMIs (120). Thus, psychological treatments that address traumatic

experiences and incorporate dissociation into treatment plans, could potentially benefit

individuals with SMIs experiencing dissociation and other trauma-related symptoms.

Conclusion

In conclusion, our evidence synthesis of 30 clinical studies found evidence in support

of an association between childhood adversity and dissociation in individuals with SMI. To

ensure robust conclusions, publication and other selection biases have been taken into

account.

Acknowledgements

We thank Helga Masramon, Minyoung Sim, and Vedat Sar, for providing information

regarding their relevant published studies. We also thank the researchers who kindly provided

copies of their papers for the screening stage of the research synthesis: Alexander McGirr,

Camilla Callegari, Colin A. Ross, Harvey E. Dondershine, Park E. Dietz, and Joel Parris.

Declaration of interest

The authors declare that they have no conflict of interest.

25

References

1 American Psychiatric Association. Diagnostic and statistical manual of mental

disorders DSM-V. 5th ed. American Psychiatric Association, Arlington, 2000.

2 BROWN RJ. Different types of “dissociation” have different psychological

mechanisms. J Trauma Dissociation 2006;7:7-28. doi:10.1300/J229v07n04_02.

3 SPITZER C, BARNOW S, FREYBERGER HJ, GRABE HJ. Recent developments in

the theory of dissociation. WPA 2006;5:82-86.

4 DALENBERG CJ, BRAND BL, GLEAVES DH et al. Evaluation of the evidence

for the trauma and fantasy models of dissociation. Psychol Bull 2012;138:550-588.

doi:10.1037/a0027447.

5 CARLSON EA, YATES TM, SROUFE LA. Dissociation and the development of the

self. In DELL PF, O’NEIL JA, eds. Dissociation and the dissociative disorders,

Routledge, New York, 2009.

6 GENTILE JP, DILLON KS, GILLIG PM. Psychotherapy and pharmacotherapy for

patients with dissociative identity disorder. Innov Clin Neurosci 2013;10: 22-29.

7 VAN IJZENDOORN MH, SCHUENGEL C. The measurement of dissociation in

normal and clinical populations: meta-analytic validation of the dissociative

experiences scale. Clin Psychol Rev 1996;16:365-382. doi:10.1016/0272-

7358(96)00006-2.

8 BRAEHLER C, VALIQUETTE L, HOLOWKA D et al. Childhood trauma and

dissociation in first-episode psychosis, chronic schizophrenia and community

controls. Psychiatry Res 2013;210:36-42. doi:10.1016/j.psychres.2013.05.033.

9 GHOREISHI A, SHAJARI Z. Reviewing the dissociative symptoms in patients with

schizophrenia and their association with positive and negative symptoms. Iran

J Psychiatry Behav Sci 2014;8:13-18.

10 ERYILMAZ G, KESEBIR S, GUL IG, OZTEN E, KARAMUSTAFALIOGLU KO.

Dissociative experiences in bipolar disorder II: are they related to childhood trauma

and obsessive-compulsive symptoms? Arch Clin Psychiatry 2015;42:38-40.

doi:10.1590/0101-60830000000045.

26

11 SEMIZ UB, BASOGLU C, EBRINC S, CETIN M. Childhood trauma history and

dissociative experiences among Turkish men diagnosed with antisocial personality

disorder. Soc Psychiatry and Psychiatr Epidemiol 2007;42:865-873. doi:

10.1007/s00127-007-0248-2.

12 ZANARINI MC, RUSER TF, FRANKENBURG FR, HENNEN J, GUNDERSON

JG. Risk factors associated with the dissociative experiences of borderline patients. J

Nerv Ment Dis 2000;188:26-30. doi:10.1097/00005053-200001000-00005.

13 SAR V, TAYCAN O, BOLAT N et al. Childhood trauma and dissociation in

schizophrenia. Psychopathology 2012;43:33-40. doi:10.1159/000255961.

14 SCHALINSKI I, TEICHER MH. Type and timing of childhood maltreatment and

severity of shutdown dissociation in patients with schizophrenia-spectrum disorder.

PLoS One 2015;10. doi:10.1371/journal.pone.0127151.

15 HARIRI AG, GULEC MY, ORENGUL FFC, SUMBUL EA, ELBAY RY, GULEC

H. Dissociation in bipolar disorder: relationships between clinical variables and

childhood trauma. J Affect Disord 2015;184:104-110. doi:10.1016/j.jad.2015.05.023.

16 WATSON S, CHILTON R, FAIRCHILD H, WHEWELL P. Association between

childhood trauma and dissociation among patients with borderline personality

disorder. Aust N Z J Psychiatry 2006;40:478-481.

doi:10.1111/j.14401614.2006.01825.x.

17 CHAE S, SIM M, LIM M, NA J, KIM D. Multivariate analysis of the relationship

between childhood trauma and psychotic symptoms in patients with schizophrenia.

Psychiatry Investig 2015;12:397-401. doi:10.4306/pi.2015.12.3.397.

18 GOODMAN M, WEISS DS, MITROPOULOU V et al. The relationship between

pathological dissociation, self-injury and childhood trauma in patients with

personality disorders using taxometric analyses. J Trauma Dissociation 2003;4:65-88.

doi:10.1300/J229v04n02_05.

19 KONG SS, BAE JH. Impact of childhood trauma on post-traumatic stress

symptoms and dissociation: mediating effect of bullying. J Korean Acad Psychiatr

Ment Health Nurs 2013;22:193-204. doi:10.12934/jkpmhn.2013.22.3.193.

27

20 OH HY, KIM D, KIM Y. Reliability and validity of the Dissociative

Experiences scale among South Korean patients with schizophrenia. J Trauma

Dissociation 2015;16:577-591. doi:10.1080/15299732.2015.1037040.

21 RUIZ MA, POYTHRESS NG, LILIENFELD SO, DOUGLAS KS. Factor structure

and correlates of the Dissociative Experiences Scale in a large offender sample.

Assessment 2008;15:511-521. doi:10.1177/1073191108315548.

22 JOHNSTON C, DORAHY MJ, COURTNEY D, BAYLES T, O'KANE M.

Dysfunctional schema modes, childhood trauma and dissociation in borderline

personality disorder. J Behav Ther Exp Psychiatry 2009;40:248-255.

doi:10.1016/j.jbtep.2008.12.002.

23 SIMEON D, KNUTELSKA M, SMITH L, BAKER BR, HOLLANDER E. A

preliminary study of cortisol and norepinephrine reactivity to psychosocial stress in

borderline personality disorder with high and low dissociation. Psychiatry Res

2007;149:177-184. doi:10.1016/j.psychres.2005.11.014.

24 LIBERATI A, ALTMAN DG, TETZLAFF J et al. The PRISMA statement for

reporting systematic reviews and meta-analyses of studies that evaluate health care

interventions: explanation and elaboration. Ann Intern Med 2009;151:65-95.

doi:10.7326/0003-4819-151-4-200908180-00136.

25 RAFIQ, S, CAMPODONICO C, VARESE, F. The relationship between childhood

trauma and dissociation in people with severe mental health difficulties: a meta-

analytic review. PROSPERO: International prospective register of systematic reviews.

2016. CRD42016048705. Available from:

http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42016048705.

26 JOLLEY, S, GARETY P, PETERS E et al. Opportunities and challenges in

Improving Access to Psychological Therapies for people with Severe Mental Illness

(IAPT-SMI): evaluating the first operational year of the South London and Maudsley

(SLaM) demonstration site for psychosis. Behav Res Ther 2015;64:24-30.

doi:10.1016/j.brat.2014.11.006.

27 KRAMER DN, LANDOLT MA. Characteristics and efficacy of early psychological

interventions in children and adolescents after single trauma: a meta-analysis. Eur J

Psychotraumatol 2011;2:7858. doi:10.3402/ejpt.v2i0.7858.

28

28 LUSZCZYNSKA A, BENIGHT CC, CIESLAK R. Self-efficacy and health-related

outcomes of collective trauma a systematic review. European Psychol 2009;14:51-62.

doi:10.1027/1016-9040.14.1.51.

29 PILTON M, VARESE F, BERRY K, BUCCI S. The relationship between

dissociation and voices: a systematic literature review and meta-analysis. Clin

Psychol Rev 2015;40:138-155. doi:10.1016/j.cpr.2015.06.004.

30 STEEL Z, CHEY T, SILOVE D, MARNANE C, BRYANT RA, VAN OMMEREN

M. Association of torture and other potentially traumatic events with mental health

outcomes among populations exposed to mass conflict and displacement: a

systematic review and meta-analysis. JAMA 2009;302:537-549.

doi:10.1001/jama.2009.1132.

31 VARESE F, SMEETS F, DRUKKER M et al. Childhood adversities increase the risk

of psychosis: a meta-analysis of patient-control, prospective-and cross-sectional

cohort studies. Schizophr Bull 2012;38:661-671. doi:10.1093/schbul/sbs050. 30

32 BORENSTEIN M, HEDGES LV, HIGGINS J, ROTHSTEIN HR. Introduction to

meta- analysis. John Wiley & Sons, Ltd, Chichester, UK, 2009.

33 EGGER M, DAVEY-SMITH D, SCHNEIDER M, MINDER C. Bias in meta-analysis

detected by a simple, graphical test. BMJ 1997;315:629-634.

doi:10.1136/bmj.315.7109.62.

34 DUVAL S, TWEEDIE R. A nonparametric “trim and fill” method of accounting for

publication bias in meta-analysis. J Amer Statist Assoc 2000;95:89-98.

doi:10.2307/2669529.

35 ROSS CA, ANDERSON G, CLARK P. Childhood abuse and the positive symptoms

of schizophrenia. Psychiatr Serv 1994;45:489-491.

36 HOLOWKA DW, KING S, SAHEB D, PUKALL M, BRUNET A. Childhood abuse

and dissociative symptoms in adult schizophrenia. Schizophr Res 2003;60:87-90.

doi:10.1016/S0920-9964(02)00296-7.

37 ROSS CA, KEYES B. Dissociation and schizophrenia. J Trauma Dissociation

2004;5:69-83. doi:10.1300/J229v05n03_05.

29

38 LYSAKER PH, DAVIS LW, GATTON MJ, HERMAN SM. Associations of anxiety

related symptoms with reported history of childhood sexual abuse in schizophrenia

spectrum disorders. J Clin Psychiatry 2005;66:1279–84. doi:10.4088/JCP.v66n1010.

39 SCHAFER I, HARFST T, ADERHOLD V et al. Childhood trauma and dissociation

in female patients with schizophrenia spectrum disorders: an exploratory study. J

Nerv Ment Dis 2006;194:135-138. doi:10.1097/01.nmd.0000198199.57512.84.

40 DORAHY MJ, SHANNON C, SEAGAR L et al. Auditory hallucinations in

dissociative identity disorder and schizophrenia with and without a childhood trauma

history: similarities and differences. J Nerv Ment Dis 2009;3:892-898.

doi:10.1097/NMD.0b013e3181c299ea.

41 VOGEL M, SPITZER C, KUWERT P, MOLLER B, FREYBERGER HJ, GRABE

HJ. Association of childhood neglect with adult dissociation in schizophrenic

inpatients. Psychopathology 2009;42:124-130. doi:10.1159/000204763.

42 LADDIS A, DELL PF. Dissociation and psychosis in dissociative identity disorder

and schizophrenia. J Trauma Dissociation 2012;13:397-413.

doi:10.1080/15299732.2012.664967.

43 PERONA‐GARCELAN S, CARRASCOSO‐LOPEZ F, GARCÍA‐MONTES JM et al.

Dissociative experiences as mediators between childhood trauma and auditory

hallucinations. J Trauma Stress 2012;25:323-329. doi:10.1002/jts.21693.

44 SCHAFER I, FISHER HL, ADERHOLD V et al. Dissociative symptoms in patients

with schizophrenia: relationships with childhood trauma and psychotic symptoms.

Compr Psychiatry 2012;53:364-371. doi:10.1016/j.comppsych.2011.05.010.

45 VARESE F, BARKUS E, BENTALL RP. Dissociation mediates the relationship

between childhood trauma and hallucination-proneness. Psychol Med 2012;42:1025-

1036. doi:10.1017/S0033291711001826.

46 KIM D, BAE H, HAN C, OH HY, MACDONALD K. Psychometric properties of

the Childhood Trauma Questionnaire-short form (CTQ-SF) in Korean patients with

schizophrenia. Schizophr Res 2013;144:93-98. doi:10.1016/j.schres.2012.12.020.

30

47 LAFERRIERE-SIMARD MC, LECOMTE T, AHOUNDOVA L. Empirical testing of

criteria for dissociative schizophrenia. J Trauma Dissociation 2014;15:91-107.

doi:10.1080/15299732.2013.834860.43.

48 ALVAREZ MJ, MASRAMON H, PENA C et al. Cumulative effects of childhood

traumas: polytraumatization, dissociation, and schizophrenia. Community Ment

Health J 2015;51:54-62. doi:10.1007/s10597-014-9755-2.

49 SIMEON D, NELSON D, ELIAS R, GREENBERG J, HOLLANDER E.

Relationship of personality to dissociation and childhood trauma in borderline

personality disorder. CNS Spectr 2003;8:755-762.

50 LOFFLER-STASTKA H, SZERENCSICS M, BLUML V. Dissociation, trauma,

affect regulation and personality in patients with a borderline personality

organization. Bull Menninger Clin 2009;73:81-98. doi:10.1521/bumc.2009.73.2.81.

51 YILMAZ O, ATES MA, SEMIZ UB et al. Childhood traumas in patients with bipolar

disorder: association with alexithymia and dissociative experiences. Anatolian J

Psychiatry 2016;17:188-195. doi:10.5455/apd.188358.

52 VAN DER KOLK BA, PERRY JC, HERMAN JL. Childhood origins of self-

destructive behaviour. Am J Psychiatry 1991;148:1665-1671.

doi:10.1176/ajp.148.12.1665.

53 ZANARINI MC, GUNDERSON JG, MARINO MF, SCHWARTZ EO,

FRANKENBURG FR. Childhood experiences of borderline patients. Compr

Psychiatry 1989;30:18-25. doi:10.1016/0010-440X(89)90114-4.

54 BERNSTEIN DP, FINK L. Manual for the Childhood Trauma Questionnaire: a

retrospective self-report. Harcourt Brace & Company, New York, 1998.

55 BERNSTEIN DP, FINK L, HANDELSMAN L et al. Initial reliability and validity of

a new retrospective measure of child abuse and neglect. Am J

Psychiatry 1994;151:1132-1136.

56 SAR V, OZTURK E, IKIKARDES E. Validity and reliability of the Turkish version

of Childhood Trauma Questionnaire. Turkiye.Klinikleri J of Med Sci 2012;32:1054–

1063. doi:10.5336/medsci.2011-26947.

31

57 WINGENFELD K, SPITZER C, MENSEBACH C et al. The German version of the

Childhood Trauma Questionnaire (CTQ): preliminary psychometric properties.

Psychotherapie, Psychosomatik, Medizinische Psychologie 2010;60:442-450.

doi:10.1055/s-0030-1247564.

58 BERNSTEIN DP, STEIN JA, NEWCOMB MD et al. Development and validation of

a brief screening version of the Childhood Trauma Questionnaire. Child Abuse Negl

2003;27:169-190. doi:10.1016/S0145-2134(02)00541-0.

59 HERNANDEZ A, GALLARDO-PUJOL D, PEREDA N et al. Initial validation of the

Spanish Childhood Trauma Questionnaire-short form: factor structure, reliability and

association with parenting. J of Interpers Violence 2013;28:1498–1518.

doi:10.1177/0886260512468240.

60 KIM D, PARK SC, YANG H, OH DH. Reliability and validity of the Korean version

of the Childhood Trauma Questionnaire-short form for psychiatric outpatients.

Psychiatry Investig 2011;8:305-311. doi:10.4306/pi.2011.8.4.305.

61 ROSS CA. Dissociative identity disorder. Diagnosis, clinical features, and treatment

of multiple personality. 2nd ed. John Wiley & Sons, New York, 1997.

62 GRAY MJ, LITZ BT, HSU JL, LOMBARDO TW. Psychometric properties of the

Life Events Checklist. Assessment 2004;11:330-341.

doi:10.1177/1073191104269954.

63 SANDERS B, BECKER-LAUNSEN E. The measurement of psychological

maltreatment: early data on the Child Abuse and Trauma Scale. Child Abuse Negl

1995;19:315–323. doi:10.1016/S0145-2134(94)00131-6.

64 CUSACK KJ, FRUEH BC, BRADY KT. Trauma history screening in a community

mental health center. Psychiatric Serv 2004;55:157-162.

doi:10.1176/appi.ps.55.2.157.

65 DAVIDSON J, SMITH R. Traumatic experiences in psychiatric outpatients. J

Trauma Stress 1990;3:459-475. doi:10.1002/jts.2490030314. 66 HERMAN JL, VAN DER KOLK B. Traumatic antecedent of borderline personality

disorder. In VAN DER KOLK B, ed. Psychological trauma, American Psychiatric

Press, Washington, 1987.

32

67 NIJENHUIS ERS, VAN DER HART O, VANDERLINDEN J. The traumatic

experiences checklist. In NIJENHUIS ERS, ed. Somatoform dissociation:

Phenomena, measurement, and theoretical issues, Van Gorcum, Assen, 1999.

68 BERNSTEIN EM, PUTNAM FW. Development, reliability, and validity of a

dissociation scale. J Nerv Ment Dis 1986;174:727–735. doi:10.1097/00005053-

198612000-00004.

69 FREYBERGER HJ, SPITZER C, STIEGLITZ RD, KUHN G, MAGDEBURG N,

BERNSTEIN- CARLSON E. Fragebogen zu dissoziativen Symptomen (FDS).

Deutsche Adaption,Reliabilität und Validität der deutschen Dissociative Experiences

Scale (DES). Psychotherapie, Psychosomatik, Medizinische Psychologie

1998;48:223–229.

70 SPITZER C, FREYBERGER HJ, STIEGLITZ RD et al. Adaptation and psychometric

properties of the German version of the dissociative experience scale. J Trauma Stress

1998;11:799–809. doi:10.1023/A:1024457819547.

71 PARK J, CHOE BM, KIM M et al. Standardization of Dissociative Experiences

Scale-Korean version (I). Korean J Psychopathol 1995;4:105–125.

doi:10.1080/15299732.2015.1037040.

72 YARGIC LI, TUTKUN H, SAR V. Reliability and validity of the Turkish version of

the Dissociative Experiences Scale. Dissociation: Progress Dissociative Disord

1995;16:577-591. doi:10.1080/15299732.2015.1037040.

73 CARLSON EB, PUTNAM FW. An update on the Dissociative Experiences Scale.

Dissociation: Progress Dissociative Disord 1993;6:16–27.

74 ICARAN E, COLOM R, ORENGO-GARCIA F. Validation study of the dissociative

experiences scale in Spanish population sample. Actas Luso Esp Neurol Psiquiatr

Cienc Afines 1996;24:7–10.

75 WALLER N, PUTNAM FW, CARLSON EB. Types of dissociation and dissociative

types: a taxometric analysis of dissociative experiences. Psychol Methods

1996;1:300-321. doi:10.1037/1082-989X.1.3.300.

76 DELL PF. The Multidimensional Inventory of Dissociation (MID): a

comprehensive measure of pathological dissociation. J Trauma Dissociation

2006;7:77–106. doi:10.1300/J229v07n02_06.

33

77 SCHALINSKI I, SCHAUER M, ELBERT T. The Shutdown Dissociation scale (Shut-

D). Eur J Psychotraumatol 2015;6:25652. doi:10.3402/ejpt.v6.25652.

78 BRIERE J. Trauma Symptom Inventory professional manual. Psychological

Assessment Resources, Odessa, 1995.

79 COHEN J. A power primer. Psychol Bull 1992;112:155-159. doi:10.1037/0033-

2909.112.1.155. 33.

80 GLASS GV, MCGAW B, SMITH ML. Meta-Analysis in Social Research. Sage,

Beverly Hills, 1981.

81 HIGGINS JP, THOMPSON SG, DEEKS JJ, ALTMAN DG. Measuring inconsistency

in meta-analyses. BMJ 2003;327:557-560. doi:10.1136/bmj.327.7414.557.

82 LOFFLER-STASTKA H, SZERENCSICS M, BLUML V. Dissociation, trauma,

affect regulation and personality in patients with a borderline personality

organization. Bull Menninger Clin 2009;73:81-98. doi:10.1521/bumc.2009.73.2.81.

83 SCHAFER I, HARFST T, ADERHOLD V et al. Childhood trauma and dissociation

in female patients with schizophrenia spectrum disorders: an exploratory study. J

Nerv Ment Dis 2006;194:135-138. doi:10.1097/01.nmd.0000198199.57512.84.

84 VAN DER KOLK BA, PERRY JC, HERMAN JL. Childhood origins of self-

destructive behaviour. Am J Psychiatry 1991;148:1665-1671.

doi:10.1176/ajp.148.12.1665.

85 LYSSENKO L, SCHMAHL C, BOCKHACKER L, VONDERLIN R, BOHUS M,

KLEINDIENST N. Dissociation in psychiatric disorders: a meta-analysis of studies

using she Dissociative Experiences Scale. Am J Psychiatry 2017;175:37-46.

doi:10.1176/appi.ajp.2017.17010025.

86 PALMIER-CLAUS J, BUCCI S, BERRY K, VARESE F. The relationship between

childhood adversity and bipolar disorder: a systematic review and meta-analysis. Br J

Psychiatry 2016;209:454-459. doi:10.1192/bjp.bp.115.179655.

87 FREYD JJ. Betrayal trauma: traumatic amnesia as an adaptive response to childhood

abuse. Ethics Behav 1994;4:307-329. doi:10.1207/s15327019eb0404_1.

34

88 CARLSON EA. A prospective longitudinal study of attachment

disorganization/disorientation. Child Dev 1998;69:1107-1128. doi: 10.1111/j.1467-

8624.1998.tb06163.x.

89 Greenfield SF, Strakowski SM, Tohen M, Batson SC, Kolbrener ML. Childhood

abuse in first-episode psychosis. Br J Psychiatry 1994;164:831-834.

doi:10.1192/bjp.164.6.831.

90 VAN DEN BOSCH LM, VERHEUL R, LANGELAND W, VAN DEN BRINK W.

Trauma, dissociation, and posttraumatic stress disorder in female borderline patients

with and without substance abuse problems. Aust N Z J Psychiatry 2003;37:549-555.

doi:10.1046/j.1440-1614.2003.01199.x.

91 ANGLIN DM, POLANCO-ROMAN L, LUI F. Ethnic variation in whether

dissociation mediates the relation between traumatic life events and attenuated

positive psychotic symptoms. J Trauma Dissociation 2015;16:68-85.

doi:10.1080/15299732.2014.953283.

92 SEEDAT S, STEIN MB, FORDE DR. Prevalence of dissociative experiences in a

community sample: relationship to gender, ethnicity, and substance use. J Nerv Ment

Dis 2003;191:115-120. doi:10.1097/01.NMD.0000050940.16782.6B.

93 ALPER K, DEVINSKY O, PERRINE K et al. Dissociation in epilepsy and

conversion non epileptic seizures. Epilepsia 1997;38:991-997.

doi:10.1111/j.15281157.tb01481.x.

94 PERZOW SE, PETRENKO CL, GARRIDO EF, COMBS MD, CULHANE SE,

TAUSSIG HN. Dissociative symptoms and academic functioning in maltreated

children: a preliminary study. J Trauma Dissociation 2013;14:302-311.

doi:10.1080/15299732.2012.736928.

95 HOLMES EA, BROWN RJ, MANSELL W et al. Are there two qualitatively distinct

forms of dissociation? A review and some clinical implications. Clin Psychol Rev

2005;25:1-23. doi:10.1016/j.cpr.2004.08.006.

96 ALLEN JG, COYNE L, CONSOLE DA. Dissociation contributes to anxiety and

psychoticism on the Brief Symptom Inventory. J Nerv Ment Dis 1996;184:639-641.

doi:10.1097/00005053-199610000-00010.

35

97 LEAVITT F. MMPI profile characteristics of women with varying levels of normal

dissociation. J Clin Psychol 2001;57:1469-1477. doi:10.1002/jclp.1110.

98 MERCKELBACH H, MURIS P, RASSIN E. Fantasy proneness and cognitive

failures as correlates of dissociative experiences. Pers Individ Dif 1999;26:961-967.

doi:10.1016/s0191-8869(98)00193-7.

99 GIESBRECHT T, MERCKELBACH H, GERAERTS E. The dissociative experiences

taxon is related to fantasy proneness. J Nerv Ment Dis 2007;195:769-772.

doi:10.1097/NMD.0b013e318142ce55.

100 WILLIAMS LM. Recall of childhood trauma: prospective study of women's

memories of child sexual abuse. J Consult Clin Psychol 1994;62:1167-1176.

doi:10.1037/0022-006X.62.6.1167.

101 HIGGINS JP, GREEN S, eds. Cochrane handbook for systematic reviews of

interventions (Vol. 4). John Wiley & Sons, Chichester, 2011.

102 MAURITZ MW, GOOSSENS PJ, DRAIJER N, VAN ACHTERBERG T. Prevalence

of interpersonal trauma exposure and trauma-related disorders in severe mental

illness. Eur J Psychotraumatol 2013;4. doi:10.3402/ejpt.v4i0.19985.

103 HARDOON S, HAYES JF, BLACKBURN R et al. Recording of severe mental

illness in United Kingdom primary care, 2000–2010. PLoS 2013;8.

doi:10.1371/journal.pone.0082365.

104 OSBORN D, LEVY G, NAZARETH I, KING M. Suicide and severe mental

illnesses. Cohort study within the UK general practice research database. Schizophr

Res 2008;99:134-138. doi:10.1016/j.schres.2007.11.025.

105 RUGGERI M, LEESE M, THORNICROFT G, BISOFFI G, TANSELLA M.

Definition and prevalence of severe and persistent mental illness. Br J Psychiatry

2000;177:149-155. doi:1192/bjp.177.2.149.

106 TRICKETT PK, NOLL JG, REIFFMAN A, PUTNAM FW. Variants of intra familial

sexual abuse experience: implications for short-and long-term development. Dev

Psychopathol 2001;13:1001–1019.

107 LAOR N, WOLMER L, KORA M, YUCEL D, SPIRMAN S, YAZGAN Y.

Posttraumatic, dissociative and grief symptoms in Turkish children exposed to

36

the 1999 earthquakes. J Nerv Ment Dis 2002;190:824-832.

doi:10.1097/01.NMD.0000041959.54021.A7.

108 NEUNER F, PFEIFFER A, SCHAUER-KAISER E, ODENWALD M, ELBERT T,

ERTL V. Haunted by ghosts: prevalence, predictors and outcomes of spirit possession

experiences among former child soldiers and war-affected civilians in Northern

Uganda. Soc Sci Med 2012;75:548-554. doi:10.1016/j.socscimed.2012.03.028.

109 PENNING SL, BHAGWANJEE A, GOVENDER K. Bullying boys: the traumatic

effects of bullying in male adolescent learners. J Child Adolesc Ment

Health 2010;22:131-143. doi:10.2989/17280583.2010.528580.

110 CUNNINGHAM T, SHANNON C, CROTHERS I et al. Enquiring about traumatic

experiences in psychosis: a comparison of case notes and self-report questionnaires.

Psychosis 2016;8:301-310. doi:10.1080/17522439.2015.1115542.

111 ROSSITER A, BYRNE F, WOTA AP et al. Childhood trauma levels in individuals

attending adult mental health services: an evaluation of clinical records and structured

measurement of childhood trauma. Child Abuse Negl 2015;44:36-45.

doi:10.1016/j.chiabu.2015.01.001.

112 READ J, FRASER A. Abuse histories of psychiatric inpatients: to ask or not to ask?

Psychiatr Serv 1998;49:355-359. doi:10.1176/ps.49.3.355.

113 READ J, DILLON J, LAMPSHIRE D. How much evidence is required for a

paradigm shift in mental health? Acta Psychiatr Scand 2014;134:287-294.

doi:10.1111/acps.12216.

114 CUNNINGHAM T, SHANNON C, CROTHERS I et al. The perspectives of people

with psychosis about participating in trauma-related research. Psychosis 2016;9:82-

85. doi:10.1080/17522439.2016.1145729.

115 NEWMAN E, KALOUPEK DG. The risks and benefits of participating in trauma-

focused research studies. J Trauma Stress 2004;17:383-394.

doi:10.1023/B:JOTS.0000048951.02568.3a.

116 GRIFFIN MG, RESICK PA, WALDROP AE, MECHANIC MB. Participation in

trauma research: is there evidence of harm? J Trauma Stress 2003;16:221-227.

doi:10.1023/A:1023735821900.

37

117 HUNTER ECM, BAKER D, PHILLIPS ML, SIERRA M, DAVID AS. Cognitive-

behaviour therapy for depersonalisation disorder: an open study. Behav Res Ther

2005;43:1121-1130. doi:10.1016/j.brat.2004.08.003.

118 FARRELLY S, PETERS E, AZIS M, DAVID A, HUNTER EC. A brief CBT

intervention for depersonalisation/derealisation in psychosis: study protocol for a

feasibility randomised controlled trial. Pilot Feasibility Stud 2016;2.

doi:10.1186/s40814-016-0086-7.

119 DE BONT PA, VAN DEN BERG DP, VAN DER VLEUGEL BM et al. A multi-site

single blind clinical study to compare the effects of prolonged exposure, eye

movement desensitization and reprocessing and waiting list on patients with a current

diagnosis of psychosis and co-morbid post-traumatic stress disorder: study protocol

for the randomized controlled trial treating trauma in psychosis. Trials 2013;14:151.

doi:10.1186/1745-6215-14-151.

120 VAN DEN BERG DP, DE BONT PA, VAN DER VLEUGEL BM et al. Prolonged

exposure vs eye movement desensitization and reprocessing vs waiting list for

posttraumatic stress disorder in patients with a psychotic disorder: a randomized

clinical trial. Jama Psychiatry 2015;72:259-267.

doi:10.1001/jamapsychiatry.2014.2637.

38

Excluded (n = 302)

Non-eligible clinical sample: 90

Non-clinical sample: 33

Non-valid/or no dissociation measure: 20

Non-valid/or no childhood trauma measure: 38

Combined adulthood and childhood trauma: 13

No analysis between trauma and dissociation: 82 a

Non-quantitative methodology: 8 b

Studies in non-eligible language: 3

Case studies: 3

Conference/dissertation abstracts: 5

Un-retrieved studies: 1

Overlapping samples: 2 c

Duplications: 4

PubMed (n = 1553)

PsycINFO (n = 2119)

EMBASE (n = 2573)

Total number of retrieved studies (n = 6245)

Stage 1: Studies following duplicate removal (n = 4859)

Stage 2: Studies following title and abstract screening

(n = 332)

Stage 3: Studies following full paper screening (n = 30)

Forward and backward screening: No studies meeting eligibility criteria d

Figure 1. Flowchart of search stratergy and studies to be included in the systematic reviewa Researchers only explored the difference in childhood trauma or dissociation scores between groups (i.e. psychosis group reported higher levels of dissociation compared to the bipolar group), they did not analyse the relationship between childhood trauma and dissociation.

b Non-quantitative methodology included reviews, qualitative and non-empirical studies.

c The same participants were recruited in more than one study.

d The forward and backward screening of all 30 eligible studies led to the retrieval of 12 studies. None of the 12 studies located met the eligibility criteria for the following reasons: non-eligible clinical sample (9 studies); no analysis between childhood trauma and dissociation (2 studies) and; non-clinical sample (1 study).

39

Table 1. Summary of eligible study characteristics

Author, date Country Design Diagnosis Sample size

n (F:M) Mean age

Measures

Dissociation TraumaQuality rating

PsychosisRoss et al. 1994 (35) Canada Between Schizophrenia 83 25 F: 56 M a 41.9 DES DDIS Moderate

Holowka et al. 2003 (36) Canada Correlation Schizophrenia 26 7 F: 19 M 30.8 DES CTQWeak

Ross and Keyes, 2004 (37) USA Between Schizophrenia 60 23 F: 37 M 40.1 DES DDISModerate

Lysaker et al. 2005 (38) USA Between Schizophrenia, schizoaffective disorder 40 40 M 47.6 d TSI

TAA (brief revised version)

Moderate

Schafer et al. 2006 (39) Germany Correlation Schizophrenia-spectrum disorders 15 15 F 34.6 b

DES (German version)

CTQWeak

Dorahy et al. 2009 (40) UK and Australia Between Schizophrenia 34 7 F: 27 M 41.6 c DES-T CTQ

Moderate

Vogel et al. 2009 (41) Germany Between Schizophrenia-spectrum disorders 80 27 F: 53 M 22.2 d

DES (German version)

CTQModerate

Sar et al. 2010 (13) Turkey Correlation Schizophrenia 70 38 F: 32 M 38.3DES (Turkish version)

CTQWeak

Laddis and Dell, 2012 (42) USA Correlation Schizophrenia 40 14 F: 26 M 42.1 MID TEQ Weak

Perona-Garcelan et al. 2012 (43) Spain Correlation Schizophrenia-spectrum disorders 71 17 F: 54 M 39.1

DES-II (Spanish version)

TQWeak

40

Schafer et al. 2012 (44) Germany Correlation Schizophrenia-spectrum disorders 93 48 F: 97 M a 34.0 b

DES (German version)

CTQWeak

Varese et al. 2012 (45) UK CorrelationSchizophrenia, schizoaffective disorder, delusional disorder

45 10 F: 35 M 44.6 d DES CATWeak

Braehler et al. 2013 (8) Canada Correlation First episode psychosis 62 16 F: 46 M 23.2 DES CTQWeak

Braehler et al. 2013 (8) Canada Correlation Schizophrenia, schizoaffective disorder 43 11 F: 32 M 31.5 DES CTQ

Weak

Kim et al. 2013 (46) South Korea Correlation Schizophrenia 100 59 F: 41 M 37.8 DES-T

CTQ-SF (Korean version)

Weak

Laferriere-Simard et al. 2014 (47) Canada Correlation

Schizophrenia, Schizophreniform disorder, schizoaffective disorder, brief psychotic disorder

50 16 F: 34 M 32.0 DES CTQ

Weak

Alvarez et al. 2015 (48) Spain Correlation Schizophrenia, schizoaffective disorder 45 20 F: 25 M 41.1

DES-II (Spanish version)

CTQ-SF (Spanish version)

Weak

Chae et al. 2015 (17) South Korea Correlation Schizophrenia 92 47 F: 51 M a 43.0 b

DES (Korean version)

CTQ-SF (Korean version)

Weak

Oh et al. 2015 (20) South Korea Correlation Schizophrenia 68 42 F: 26 M 34.5

DES (Korean version)

CTQ-SF (Korean version)

Weak

Schalinkski and Teicher, 2015 (14) Germany Correlation Schizophrenia-spectrum

disorders 75 26 F: 49 M 31.0 Shut-D LECWeak

Personality disorders

Zanarini et al. 2000 (12) USA Between Borderline personality disorder 290 233 F: 57 M 26.9 DES CEQ-R Moderate

41

Goodman et al. 2003 (18) USA Correlation Personality disorders 93 33 F: 62 M a 38.5 b DES CTQ Weak

Simeon et al. 2003 (49) USA Correlation Borderline personality disorder 20 8 F: 12 M 37.9 DES CTQ-SF Weak

Watson et al. 2006 (16) UK Correlation Borderline personality disorder 132 105 F: 34 M

a 32.6 b DES CTQ Weak

Simeon et al. 2007 (23) USA Correlation Borderline personality disorder 13 6 F: 7 M 39.2 DES CTQ-SF Weak

Loffler-Stastka et al. 2009 (50) Austria Correlation Personality disorders 52 25 F: 27 M 38.0DES (German version)

CTQ (German version)

Weak

Johnston et al. 2009 (22) UK Correlation Borderline personality disorder 30 27 F: 3 M 40.0 WDS CTQ Weak

Bipolar disorder

Eryilmaz et al. 2015 (10) Turkey Correlation Bipolar disorder-II 33 19 F: 14 M 32.4DES (Turkish version)

CTQ (Turkish version)

Weak

Hariri et al. 2015 (15) Turkey Between Bipolar disorder 200 122 F: 78 M 38.5 dDES (Turkish version)

CTQ (Turkish version)

Moderate

Yilmaz et al. 2016 (51) Turkey Correlation Bipolar disorder 70 23 F: 47 M 34.7DES (Turkish version)

CTQ (Turkish version)

Weak

Mixed sample

van der Kolk et al. 1991 (52) USA Correlation

Bipolar disorder and borderline, antisocial, schizotypal personality disorder

74 39 F: 35 M Not reported DES TAQ

Weak

a Participants did not complete all measures of either childhood trauma or dissociation. Therefore, the gender distributions displayed in the table were based on the whole

sample (i.e. including the participants whose data was later excluded as they did not complete the measures).

42

b Participants did not complete all measures of either childhood trauma or dissociation. Therefore, the mean age displayed in the table was based on the whole sample (i.e.

including the participants whose data was later excluded as they did not complete the measures).

c The mean age was not reported for the schizophrenia group, instead, the mean age for the whole sample (i.e. schizophrenia and dissociative identity disorder) was reported.

However, as they were no significant differences in age between the groups, the whole sample mean age was reported in the table.

d In the original studies the mean age was reported for separate groups (e.g. male and female, abused group vs non-abused group). Therefore, to calculate the mean age of the

sample of interest, the mean age for separate groups were combined using a formula.

Abbreviations: DDIS = Dissociative Disorders Interview Schedule.

Childhood trauma measures: Childhood Experiences Questionnaire-Revised (CEQ-R; 53), Childhood Trauma Questionnaire (CTQ; 54, 55 [English version]; 56 [Turkish

version]; 57 [German version]), Childhood Trauma Questionnaire-short form (CTQ-SF; 58; [English version]; 59 [Spanish version]; 60 [Korean version]), Dissociative

Disorders Interview Scale (DDIS; 61), Life Events Checklist (LEC; 62), The Child Abuse and Trauma Scale (CATS; 63), Trauma Assessment for Adults (TAA- brief revised

version; 64), Trauma Questionnaire (65), Traumatic Antecedent Questionnaire (TAQ; 66), Traumatic Experiences Questionnaire (TEQ; 67).

Dissociation measures: Dissociative Experience Scale (DES; 68 [English version]; 69, 70 [German version]; 71 [Korean version]; 72 [Turkish version]), Dissociative

Experience Scale-II (DES-II; 73 [English version]; 74 [Spanish version]), Dissociative Experience Scale-Taxon (DES-T; 75), Multidimensional Inventory Scale (MID; 76),

Shutdown Dissociation Scale (Shut-D; 77), Trauma Symptom Inventory Scale (TSI; 78)

43

44

Figure 2: Forest plot of the relationship between total childhood trauma and dissociation across SMIs.

45

Table 2: The relationship between specific childhood trauma and dissociation in the overall sample and subgroups

Childhood trauma k r (95% CI), p value Q Statistic I2 (%)

Overall sample

Sexual abuse 20 r = .26 (.17, .34), p < .001 Q = 58.41, p < .001 67.47

Physical abuse 19 r = .27 (.20, .34), p < .001 Q = 32.34, p = .021 44.34

Emotional abuse 18 r = .34 (.23, .44), p < .001 Q = 65.93, p < .001 74.21

Physical neglect 17 r = .23 (.17, .29), p < .001 Q = 21.33, p = .166 24.98

Emotional neglect 16 r = .10 (.03, .18), p = .009 Q = 22.86, p = .087 34.38

Neglect 3 r = .30 (.22, .38), p < .001 Q = 1.93, p = .382 0.00

Aggregated neglect a 36 r = .19 (.14, .24), p < .001 Q = 63.73, p = .002 44.08

Schizophrenia spectrum disorder

Sexual abuse 13 r = .30 (.18, .41), p < .001 Q = 37.60, p < .001 68.09

Physical abuse 13 r = .32 (.24, .40), p < .001 Q = 19.18, p = .084 37.45

Emotional abuse 12 r = .41 (.27, .54), p < .001 Q = 47.11, p < .001 76.65

Physical neglect 11 r = .30 (.23, .37), p < .001 Q = 9.74, p = .467 0.00

Emotional neglect 11 r = .11 (.02, .20), p = .014 Q = 12.57, p = .249 20.46

Neglect 1 r = .23 (-.07, .49), p = .129 Q = 0.00, p = 1.000 0.00

Aggregated neglect a 23 r = .22 (.15, .28), p < .001 Q = 34.66, p = .042 36.53

Personality disorder

Sexual abuse 5 r = .24 (.13, .35), p < .001 Q = 6.69, p = .153 40.20

Physical abuse 4 r = .23 (.13, .32), p < .001 Q = 0.42, p = .936 0.00

Emotional abuse 4 r = .25 (.14, .36), p < .001 Q = 2.65, p = .449 0.00

Physical neglect 4 r = .14 (.05, .24), p = .004 Q = 1.86, p = .603 0.00

Emotional neglect 3 r = .06 (-.21, .31), p = .679 Q = 7.63, p = .022 73.78

Neglect 1 r = .29 (.20, .37), p < .001 Q = 0.00, p = 1.000 0.00

Aggregated neglect a 8 r = .14 (.03, .25), p < .001 Q = 19.59, p = .007 64.26

Note. Cl = confidence intervals; a Physical neglect, emotional neglect and neglect were combined to generate the aggregated neglect effect size.

46

SUPPORTING INFORMATION

Sonya Rafiq a, Carolina Campodonico a, and Filippo Varese a,b

The relationship between childhood adversities and dissociation in severe mental illness: A meta-analytic review.

Full search stratergy: All search terms used in the database search

Table S1: Author contacts during the meta-analytic review

Table S2: Quality assessment

Table S2: Publication analyses

Table S3: Sensitivity analyses for the overall sample and subgroups

47

Search terms

Dissociation (multiple personalit*, OR dissociat*, OR absorption, OR depersonalisation, OR derealisation, OR depersonalization, OR derealization) AND

trauma (child abuse, OR maltreat*, OR victim*, OR trauma*, OR advers*, OR emotional abuse, OR physical abuse, OR psychological abuse, OR sexual

abuse, OR bullied, OR bully*, OR neglect*, OR persecution, OR mass violence, OR torture, OR combat*, OR terror*, OR war*, OR conflict*, OR refugee*)

AND severe mental health (SMI OR severe mental illness*, OR serious mental illness*, OR chronic mental illness*, OR enduring mental illness*, OR severe

and enduring mental illness*, OR persistent mental illness*, OR severe mental disorder*, OR serious mental disorder*, OR chronic mental disorder*, OR

persistent mental disorder*, OR schizophrenia, OR schizophrenic*, OR schizoaffective, OR psychoti*, OR psychosis, OR paranoi*, OR delusion*, OR

voice*, OR, hallucinat*, OR personality disorder*, OR affective illness*, OR affective disorder*, OR affective dysfunction*, OR bipolar*, OR bi polar*, OR

bi-polar*, OR mood disorder*, OR mood dysfunction*, OR manic, OR mania, mood swing*, OR hypomani*, OR mixed episode*, OR rapid cycling, OR

rapid cycle, OR mood disturbance*, OR Rcbd).

The following words were MeSH terms: schizophrenia, psychotic disorders, bipolar disorder, child abuse, physical abuse, sexual abuse, bullying, violence,

torture, absorption, depersonalization, and depersonalization. Subject headings in EMBASE and PsycINFO included: multiple personality, dissociation,

absorption, depersonalization, child abuse, victim, emotional abuse, physical abuse, sexual abuse, bullying, neglect, persecution, torture, terrorism, war,

conflict, refugee, schizophrenia, psychotic disorder, psychosis, paranoia, delusion, voice, hallucination, personality disorder, bipolar disorder, mood disorder,

mania, mood, hypomania, and rapid cycling bipolar disorder.

48

Table S1. Author contacts

Contact regarding Author responses Data retrieved

Missing data from eligible articles:

8 authors contacted:

5 responses

3 non-responders

2 data sources retrieved and included

1 author clarified data

Papers that were un-retrieved due to unavailable journal subscription

9 authors contacted:

8 responses

1 un-retrievable contact

8 papers retrieved

Unpublished data from authors of papers included in the analysis

30 authors contacted:

5 responses

25 non-responders

No data source retrieved

49

Table S2: Quality assessment – INSERT TABLE AND CHANGE OTHER TABLE NUMBERS

Table S2: Publication analyses for the overall sample and subgroups

Diagnostic group Egger’s test Trim-and-fill method Missing studies

SMI p = .012 r = .28, 95% CI (.22, .34) 7

Schizophrenia spectrum disorders p = .059 r = .35, 95% CI (.27, .42) 4

Personality disorders p = .964 r = .24, 95% CI (.20, .29) 2Note: CI = confidence intervals

50

Table S3: Sensitivity analyses conducted on the overall sample and subgroups

Sensitivity analyses r (95% CI), p value a Q Statistic, I2 (%) Change in

effect size b

1. Selective reporting: Overall analysis k = 28, r = .33, (.26, .39), p < .001 Q = 92.78, p < .001, I2 = 70.90 No

Selective reporting: Schizophrenia k = 19, r = .38, (.30, .45), p < .001 Q = 46.93, p < .001, I2 = 61.65 No

Selective reporting: Personality disorder k = 5, r = .19, (.12, .26), p < .001 Q = 2.46, p = .652, I2 = 00.00 No

2. Childhood trauma measure: Overall analysis k = 29, r = .32, (.26, .38), p < .001 Q = 94.71, p < .001, I2 = 70.44 No

Childhood trauma measure: Schizophrenia k = 19, r = .39, (.30, .46), p < .001 Q = 49.83, p < .001, I2 = 63.88 No

3. Reported total trauma: Overall analysis k = 26, r = .36, (.30, .42), p < .001 c Q = 48.86, p = .003, I2 = 48.83 No

Reported total trauma: Schizophrenia k = 18, r = .40, (.32, .47), p < .001 c Q = 35.00, p = .006, I2 = 51.42 No

Reported total trauma: Personality disorder k = 5, r = .21, (.07, .34), p = .003 c Q = 2.85, p = .584, I2 = 00.00 No

Reported total trauma: Bipolar disorder k = 2, r = .24, (.11, .35), p < .001 c Q = 0.06, p = .800, I2 = 00.00 Yes

4. CTQ measures: Overall analysis k = 22, r = .33, (.25, .41), p < .001 Q = 79.56, p < .001, I2 = 73.61 No

Non-CTQ measures: Overall analysis k = 9, r = .33, (.25, .42), p < .001 Q = 14.35, p = .073, I2 = 44.27 No

5. DES measures: Overall analysis k = 27, r = .34, (.28, .40), p < .001 Q = 91.48, p < .001, I2 = 71.58 No

Non-DES measures: Overall analysis k = 4, r = .22, (.02, .41), p < .033 Q = 5.65, p = .130, I2 = 46.85 YesNote. CI = confidence intervals. a Findings after the sensitivity analyses. b Whether the effects sizes from the analyses investigating the relationship between childhood

trauma and dissociation, altered as a result of the sensitivity analyses (e.g. medium effect size to small effect size). c When total trauma scores generated from composite

summary effect sizes generated by CMA2 (five studies) were removed from the overall and subgroup analysis, only total trauma scores reported in the studies were

51

included in the analysis.

52