Professional Documents
Culture Documents
DOI:
10.1016/j.chiabu.2019.104026
Publication date:
2019
Document Version:
Accepted author manuscript
Link to publication
Copyright
No part of this publication may be reproduced or transmitted in any form, without the prior written permission of the author(s) or other rights
holders to whom publication rights have been transferred, unless permitted by a license attached to the publication (a Creative Commons
license or other), or unless exceptions to copyright law apply.
Mickey T. Kongersleva,b,*, Bo Bacha, Gina Rossic, Anne M. Trauelsend, Nicolai Ladegaarde, Sille S.
a
Psychiatric Research Unit, Faelledvej 6, 4200 Slagelse, Region Zealand, Denmark.
b
Department of Psychology, University of Southern Denmark, Campusvej 55, 5230 Odense M,
Denmark.
c
Faculty of Psychology & Educational Sciences, Personality and Psychopathology Research Group,
Copenhagen, Denmark.
e
Department of Affective Disorders, Aarhus University Hospital – Psychiatry, Palle Juul-Jensens
*
Corresponding author at: Psychiatric Research Unit, Faelledvej 6, 4200 Slagelse, Region Zealand,
Denmark.
1
Funding: This research did not receive any specific grant from funding agencies in the public,
Summary declaration of interest statement: The authors declare they have no interests to declare.
2
Psychometric validation of the Childhood Trauma Questionnaire-Short Form (CTQ-SF) in a
ABSTRACT
BACKGROUND
The Childhood Trauma Questionnaire – Short Form (CTQ-SF) is a widely utilized self-report
instrument in the assessment and characterization of childhood trauma. Yet, research on the
instrument’s psychometric properties in clinical samples is sparse, and the Danish version of the
OBJECTIVES
To examine the structural validity, internal consistency reliability, and multi-method convergent
The study was based on data from four Danish clinical samples (N = 393): 1) Outpatients diagnosed
with personality disorders, 2) Patients commencing psychiatric treatment for non-affective first-
episode psychosis, 3) Patients diagnosed with first-episode or prolonged depression recruited from
general practitioners and an outpatient mood disorder clinic, and 4) detained delinquent boys.
METHODS
Confirmatory factor analysis was used to explore structural validity. Also, we calculated internal
parenting.
RESULTS
Confirmatory factor analyses indicated that the five-factor structure described in CTQ-SF manual
with three error correlated items best fitted the data, as compared to various other models.
3
Coefficients of congruence also supported factorial similarity across countries (i.e. US substance
abuser and a mixed Brazilian sample). Internal consistency reliability was acceptable and
CONCLUSION
These findings provide support for the reliability and validity of the Danish version of the CTQ-SF
in clinical samples.
KEYWORDS
4
Background
report that they have suffered from physical abuse, 36% from emotional abuse, 16% from physical
neglect, and 18% of girls and 8% of boys from sexual abuse in the past year (World Health
Organization, 2017). Thus, developmental trauma in the form of child maltreatment, comprising
various forms of abuse and neglect, is widespread worldwide (Gilbert et al., 2009; Moody,
Cannings-John, Hood, Kemp, & Robling, 2018; Stoltenborgh, Bakermans-Kranenburg, Alink, &
van Ijzendoorn, 2015). It poses a serious public health problem, given the fact that childhood
trauma is a robust and powerful pluripotent risk factor for both concurrent and future detrimental
developmental outcomes, including poor mental and physical health (Cicchetti, 2016; Karterud &
Kongerslev, 2019; Kessler et al., 2010; McCrory, Gerin, & Viding, 2017; Vachon, Krueger,
Rogosch, & Cicchetti, 2015; World Health Organization, 2017; Zeanah & Humphreys, 2018).
Furthermore, almost all common mental disorders across the lifespan have been shown to be
strongly associated with various forms of child maltreatment. Childhood adversity may thus be the
single greatest known environmental predictor of transdiagnostic psychiatric problems (Green et al.,
2010; Kessler et al., 2010; Zeanah & Humphreys, 2018). The strong link betwixt child maltreatment
and mental disorders is also underscored by research documenting the negative effects of childhood
trauma on presentation, severity, course, and treatment response in adolescent and adult patients
diagnosed with for example depression (Williams, Debattista, Duchemin, Schatzberg, & Nemeroff,
2016), bipolar disorder (Cakir, Tasdelen Durak, Ozyildirim, Ince, & Sar, 2015; Etain et al., 2013),
borderline personality disorder (Bo & Kongerslev, 2017; Bo, Sharp, Fonagy, & Kongerslev, 2017;
Levey, Apter, & Harrison, 2016; Zanarini, Frankenburg, Hennen, Reich, & Silk, 2006), and
psychosis (Schäfer & Fisher, 2011). Importantly, childhood trauma also increases the risk for
5
suicide and suicide attempts considerably (Angelakis, Gillespie, & Panagioti, 2019; Gerdner &
Allgulander, 2009; Zatti et al., 2017) – for example a recent systematic review reported a two- to
threefold increased risk of suicide attempts for all types of child maltreatment (Angelakis et al.,
2019). Consequently, assessment of childhood abuse and neglect is important in routine clinical
practice, in order to inform treatment planning and assess risk of suicide and prognosis.
settings, due in part to the sensitivity of the subject which may make some patients or clinicians
reluctant to talk about it during a face-to-face interview (Read, Hammersley, & Rudegeair, 2018;
Zeanah & Humphreys, 2018). A brief self-report measure could therefore be useful to facilitate
systematic assessment of childhood maltreatment. Such a measure would also be time and cost
efficient making it possible to screen many patients, and it can be filled out in private, which makes
maltreatment (Dovran, Winje, Arefjord, & Haugland, 2012), the Childhood Trauma Questionnaire-
Short Form (CTQ-SF) (Bernstein & Fink, 1998; Bernstein et al., 2003) is internationally one of the
most widely used (Baker & Maiorino, 2010; Grassi-Oliveira et al., 2014). The CTQ-SF is a self-
report instrument suitable for use with adolescents and adults to screen for five types of
maltreatment: physical, emotional, and sexual abuse, and physical and emotional neglect, as well as
including a total scale score indicating the global level of childhood trauma. On the whole, the
CTQ-SF appears to meet the general requirements for a brief trauma screening instrument (Brewin,
2005), including satisfactory reliability and validity (Baker & Maiorino, 2010). Yet, results from
studies on the five-factor structure reported in the original manual (Bernstein & Fink, 1998) have
been mixed, with some studies supporting it whilst others only obtain partial support or suggest
alternative models. A number of studies, based on both clinical and non-clinical samples, and
6
including both adolescents and adults, predominantly from North America and Europe, have
obtained support for the five-factor structure reported in the CTQ-SF manual (Bernstein et al.,
2003; Dovran et al., 2013; Dudeck et al., 2015; Hernandez et al., 2013; Sacchi, Vieno, & Simonelli,
2018; Thombs, Bernstein, Lobbestael, & Arntz, 2009; Thombs, Lewis, Bernstein, Medrano, &
Hatch, 2007). Moreover, some of these studies also tested and demonstrated the structural
invariance or measurement equivalence of the original CTQ-SF five-factor structure across gender,
age and subsamples (Bernstein et al., 2003; Dovran et al., 2013; Thombs et al., 2007). Yet, other
studies, based on both clinical and non-clinical samples including adolescents and adults from for
example Sweden, Brazil, and Korea, have failed (or at least partially failed) to replicate the original
CTQ-SF five-factor structure (Gerdner & Allgulander, 2009; Grassi-Oliveira et al., 2014; Villano et
al., 2004). Particularly, the Physical Neglect scale has been found to be problematic in these studies,
and also appears more generally to have the poorest internal consistency of all the five scales in
most previous studies (Gerdner & Allgulander, 2009; Grassi-Oliveira et al., 2014). Particularly the
findings from studies in clinical samples suggest an alternative five-factor structure where items
number 2 and 26 loads onto the Emotional Neglect scale, rather than on the Physical Neglect scale
as would otherwise have been expected according to the CTQ-SF manual. This alternative CTQ-SF
five-factor was found to be the most appropriate when compared to the original factor solution in a
mixed Brazilian sample when the total CTQ-SF score was also included at the second order level
(Grassi-Oliveira et al., 2014). Another study, based on an Italian college sample, though finding that
the five-factor structure specified in the CTQ-SF manual provided best fit, also found that a four-
factor first order structure (wherein items from the Physical and Emotional neglect scales were
collapsed into one single Neglect scale) provided a good fit (Sacchi et al., 2018). Given these
indecisive research findings, more research examining the factorial structure in clinical samples is
7
The Danish version of the CTQ-SF (Bernstein & Fink, 2011) has already been used in
a number of clinical studies to explore childhood trauma in relation to psychopathology (e.g., Bach
& Fjeldsted, 2017; Trauelsen et al., 2015). However, whilst the original CTQ-SF manual contains
information on the factor structure and item loadings in adolescent and adult clinical samples, the
Danish manual only contains psychometric information based on a non-clinical Danish convenience
sample. Thus, information on the psychometric performance of the Danish CTQ-SF in clinical
samples is lacking.
psychometric properties, research on the CTQ-SF factor structure displays conflicting results.
Additionally, the instrument has never been psychometrically tested in Danish clinical sample.
Hence, the present study aimed to evaluate the psychometric properties of the Danish CTQ-SF in a
clinical sample. This involved: i) testing the instruments factor structure through comparing the fit
of the original second order five-factor model to that of competing one-, four- and five-factor
models reported in the literature as well as to a one factor model; ii) estimating the factorial
similarity for the best fitting factor model, based on factor loadings, with other international studies;
iii) assessing scale reliability; and iv) estimating convergent validity with an interview-based
Archival material from four clinical research studies in Denmark was used. Each of
these previous studies was conducted in accordance with the Declaration of Helsinki, approved by
the respective local ethics committees in Denmark, and reported to the Danish Data Protection
8
Agency. Written informed consent was obtained from all participants in the original studies. For the
present study we only had access to fully anonymized data. Apart from administration of the CTQ-
SF all participants were also tested with structured diagnostic interviews conducted by trained
clinicians. Adequate interrater-reliability for these ratings has been documented (see the original
study references below). The four clinical samples, from which our final sample was derived, can
1) Patients consecutively enrolled, as part of routine clinical assessment, from March 2012 to
(n = 142; 68% women; Mage = 29 years, SD = 8.4). Subjects fulfilling criteria for current
psychotic disorder, current manic episode, autism or organic disorders, or organic induced
disorders were excluded. The majority of the included patients (71%) met criteria for
Borderline Personality Disorder (Bach & Sellbom, 2016). Other types of specific personality
disorders as well as various mental state disorders also occurred frequently (Bach,
Simonsen, Christoffersen, & Kriston, 2017). Moreover, in this sample, 72 randomly selected
(Zanarini, 1992). These interviews were conducted and scored by a clinical psychologist
blinded to the participants self-reported responses on the CTQ-SF. We used these data to
test the convergent validity associations between the CTQ-SF clinical scales and interview-
behavior was also rated by another clinical psychologist to obtain estimates on interrater
Region Zealand in Denmark (n = 101; 26% women; Mage = 23 years, SD = 3.4; Trauelsen et
9
al., 2015). The patients were recruited consecutively over a two-year period from April 2011
to April 2013. Inclusion criteria for this study was meeting criteria for a diagnosis of non-
affective psychosis and being 18 to 35 years of age. Exclusion criteria were a previous
diagnosis of psychosis.
practitioners and an outpatient mood disorder clinic in the Central Denmark Region (n = 71;
78% women; Mage = 35 years, SD = 11.6; Ladegaard, Lysaker, Larsen, & Videbech, 2014).
The data was collected from December 2010 to December 2012. The first-episode depressed
patients (n = 44) met DSM-IV criteria for major depressive disorder, and were all
psychotropic drug-naive. The patients with prolonged depression (n = 27) were in- and
outpatients required to meet full DSM-IV criteria for major depressive disorder for a period
of minimum two years, and was additionally required to have failed to respond to two or
more pharmacotherapy treatments with antidepressants of different classes. All the recruited
patients were also required to have depression as their primary diagnosis, and a depressive
symptom severity of moderate to severe when enrolled. Current substance use disorder,
neurological illness, head trauma and chronic somatic disease were exclusion criteria.
4) Juvenile delinquent boys sampled from three secure institutions and a prison ward in
Denmark (n = 80; Mage = 17 years, SD = 0.8) during August 2010 to October 2011.
Inclusion criteria were male gender, age from 15 to 18 years, remanded or sentenced, and
willing and able to give informed consent. Exclusion criteria were profound mental
assessments. The most common mental disorders in this sample were conduct disorder
(76%), personality disorders (65%), alcohol and substance abuse (58%), Attention-
Deficit/Hyperactivity Disorder (ADHD; 23%), anxiety disorders (18%) and mood disorders
10
(8%; Gillespie, Kongerslev, Sharp, Bo, & Abu.Akel, 2018; Kongerslev, Moran, Bo, &
We excluded one woman from sample 2 (i.e., patients diagnosed with first-episode psychosis)
because of completely missing data on all CTQ-SF items. The resulting combined Danish clinical
sample was diagnostically heterogeneous and comprised 393 respondents of which 45% (n = 177)
were women. Age ranged from 15 to 63 (M = 26 years, SD = 9.6). Of the 393 respondents included
in the present study, 391 had no missing data on the CTQ-SF. Of the two respondents with missing
data, one did not respond to item 5 and item 7 on the Emotional Neglect scale, and one did not
Materials
experiences when growing up. Each item is scored on a 5-point Likert scale (1 = never true, 2 =
rarely true, 3 = sometimes true, 4 = often true, 5 = very often true). Three items compose the
Minimization/Denial scale designed to detect socially desirable response style (false negatives). The
other 25 items are divided into five clinical subscales, with five items each: Emotional Abuse,
Physical Abuse, Sexual Abuse, Emotional Neglect, and Physical Neglect. Each of the five clinical
subscales’ scores can range from 5 to 25. Scores on the 25 items can also be summed to produce a
The CEQ-R (Zanarini, 1992) is a semi-structured interview that assesses 12 types of negative
parental/caretaker behaviors: emotional abuse, physical abuse, sexual abuse, emotional withdrawal,
11
physical neglect, inconsistent treatment, emotional denial, failure to protect, lack of real
relationship, verbal abuse, parentification of child, and malevolent parenting. In the present study,
each parental behavior was rated dichotomously as present (0) versus absent (1). Interrater
reliability for the CEQ-R, in the present study, was satisfactory: The median κ value for categorical
variables was 1 (range 0.31–1.00), whereas the median intraclass correlation coefficient value for
Statistical analyses
The original CTQ-SF manual (Bernstein & Fink, 1998) reports a factor analysis of the
25-item version of the CTQ-SF (excluding the three items from the Minimization/Denial Scale). To
examine if this original five-factor structure could be reproduced in our Danish clinical sample we
performed a confirmatory factor analysis in Mplus version 8.3 (Muthén & Muthén, 2017). Due to
non-normality of the data at item-level (skewness and kurtosis statistics exceeded the critical levels
of respectively 2 and 7; Ryu, 2011) we applied a robust maximum likelihood estimation using the
Satorra-Bentler chi-square. To evaluate model fit a selection of fit indices regarded as the most
informative, according to Kline (2005), were selected: the root mean square error of approximation
(RMSEA), the comparative fit index (CFI) and the standardized root mean square residual (SRMR).
The RMSEA is a parsimony-adjusted index with a built-in correction for model complexity. The
guideline is that RMSEA values ≥ .10 suggest unacceptable fit. Values ≤ .08 suggest approximate
or good model fit. A key advantage is that a confidence interval can be calculated for the RSMEA
value, which provides more information regarding model fit than a point estimate alone. The upper
bound of this confidence interval should be ≤ .10 for acceptable model fit (Chen, Curran, Bollen,
Kirby, & Paxton, 2008). The CFI assesses the relative improvement in fit compared with the
independence model (i.e., null model which assumes unrelated variables). A rule of thumb is that
12
values ≥ .90 indicate a reasonably good fit. The SRMR is a measure of the mean absolute residual
correlation, so values close to 0 are a better result. Ideally, the value of the SRMR should be < 0.08.
In analogy to some previous studies that had to allow error estimates to covary to improve model fit
(e.g., Thombs et al., 2009; Thombs et al., 2007), we used modification indices to decide which
parameter could be set free. We selected the highest modification index (the value of this represents
combined with theoretical reasons: correlated error terms were only allowed in the model if they
also made substantive sense (i.e. an effect exists that relates the two variables, which was not
included in the specified CFA model). As soon as all fit indices indicated reasonably fit, no further
error terms were added. Finally, we checked if the model with correlated error terms indeed yielded
a significant improvement in fit when compared to the original model without error constraints by
Next, given the indecisive research findings on the factor structure of the CTQ-SF in
previous studies, fit statistics of the original five-factor model specified in the manual were
compared with fit statistics for alternative models: a one factor model (representing only a total
CTQ-SF score), an original second order five-factor model and competing models at first and
second order, that is an alternative five-factor model (Grassi-Oliveira et al., 2014) and a four-factor
model (Sacchi et al., 2018). To compare the fit of these non-nested models we used two criteria
(Claeskens & Hjort, 2008): Akaike information criterion (AIC) and a sample size adjusted Bayesian
information criterion (BIC). AIC is an asymptotically efficient criterion for model selection, which
means that it tends to select the model that minimizes prediction error as sample size increases. BIC
originates from the Bayesian tradition in statistics and is concerned with the statistical property of
consistency, which refers to the one “true model” being selected with increasing probability as
13
sample size increases. Models with the lowest AIC and BIC values are considered to show the best
congruency coefficients with similar factors for a sample of American adult substance abusers
reported in the original CTQ-SF manual (Bernstein & Fink, 1998) and with the first order original
five-factor loadings reported in a Brazilian study combining clinical and non-clinical, adult and
adolescent samples (Grassi-Oliveira et al., 2014). The factorial similarity were evaluated with the
commonly used indicator for congruency, Tucker Phi (Tucker, 1951). A Tucker Phi value in the
range of .85 to .94 corresponds to fair similarity, while a value higher than .95 suggest that the two
factors being compared can be considered equal (Lorenzo-Seva & ten Berge, 2006).
reported scale reliability in the manual (Bernstein & Fink, 1998), Cronbach’s coefficient alpha was
calculated to evaluate internal consistency. Because most of our data was non-parametrically
distributed, bivariate Spearman’s rho correlations were used to examine intercorrelations between
CTQ-SF scales and convergent associations of the CTQ-SF scales with the CEQ-R. The following
heuristic rules (Cohen, 1988) was used to interpret the effect size of the correlations: r = .10
indicates a small effect, r = .30 indicates a medium effect and r = .50 indicates a large effect. Given
the large number of tests, the Type I error rate was adjusted using a Bonferroni correction. The
conventional α = .05 was divided by the number of scales, yielding an adjusted alpha = .01.
Apart from the confirmatory factor analysis, conducted in Mplus, all other analyses
were performed using IBM SPSS Statics for MAC, version 25. Missing data was handled in Mplus
using full information maximum likelihood (FIML) and pairwise deletion in SPSS to maximize use
of information.
14
Results
Descriptive information
Table 1 provides means and standard deviations for the CTQ-SF scales across samples. In our
combined total sample kurtosis and skewness for the CTQ-SF total scale were 1.09 and 1.18,
respectively. For the Emotional Abuse, Physical Abuse, Sexual Abuse, Emotional Neglect, and
Physical Neglect subscales in the combined sample kurtosis and skewness values were –.47 and
0.74, 10.71 and 2.87, 8.50 and 2.91, –0.80 and 0.37, and 3.26 and 1.61, respectively.
We carried out confirmatory factor analysis to assess the structural validity of the
CTQ-SF in our Danish clinical sample. Because the original model did not reach adequate fit (see
Table 2), three pairs of error variances, that made substantive sense, were freed to covary: Item 9
and item 15, that refer to physical abuse, item 11 and item 12 that both include corporeal
punishment, and item 3 and item 11 that both refer to ’people in my family’. The goodness-of-fit
statistics of the original five-factor model, with three correlated error terms added, indicated good
model fit, chi-square = 673.437, df = 262, p < .001, RMSEA = .063, CFI = 0.909 and SRMR =
.071. This model, with correlated error terms, was also a significant improvement in fit compared to
the original model without error constraints, as demonstrated by a Satorra-Bentler scaled chi-square
difference test (scaling correction 0.3547, TRd 517.4292, ∆df = 3, p < .001). The standardized factor
loadings for this model, the original five-factor model with three error correlated items, are reported
in Table 3, and ranged from 0.34 (item 12 on the Physical Abuse scale) to 0.93 (item 20 on the
15
Next, the original model and competing models were compared in terms of fit
statistics. From Table 2 it is clear that based on AIC and BIC values the original model with three
correlated error terms outperformed all competing models in terms of fit to the data.
Factor congruency
Tucker Phi congruency coefficients corroborated factorial similarity of our best fitting
model (the original five-factor model with three pairs of error-correlated items) with a sample of
American adult substance abusers reported on in the original CTQ-SF manual (Bernstein & Fink,
1998) with congruency coefficients of 1.00 for Emotional Abuse, .97 for Physical Abuse, 1.00 for
Sexual Abuse, 1.00 for Emotional Neglect, and 1.00 for Physical Neglect, respectively.
Furthermore, factorial similarity was indicated with a five-factor solution derived from a Brazilian
sample combining clinical and non-clinical samples (Grassi-Oliveira et al., 2014) with congruency
coefficients of 1.00 for Emotional Abuse, .98 for Physical Abuse, 1.00 for Sexual Abuse, .99 for
Intercorrelations
Table 4 displays intercorrelations between all the CTQ-SF scales. Correlations among
the five subscales ranged from .29 to .74 (ps < .001) indicating generally medium to large effects
(Cohen, 1988). These relatively high intercorrelations amongst the five subscales indicate modest
discriminant validity and suggest that it is feasible to extract a factor representing non-specific or
global childhood trauma, thereby supporting the use of the CTQ-SF total scale. Accordingly, we
proceeded to compute clinical subscale-total correlations (Table 4). The clinical subscale-total
correlations ranged from .52 to .90 (ps < .001.), revealing large effect sizes by conventional
standards. As would be expected the Minimization/Denial scale was negatively correlated with all
16
other CTQ-SF scales (rs ranged from –.16 to –.49, ps ranged from <. 01 to < 001., indicating small
Internal consistency
Cronbach’s coefficient alpha values for the CTQ-SF total and subscales based on the
combined sample and for the respective subsamples are reported in Table 5. Alpha values for the
CTQ-SF total scale were high, ranging from .85 to .94 across all samples. For the five CTQ-SF
clinical subscales in the combined sample and stratified by gender the alpha values ranged from .70
to .93, and were remarkably similar to the median alpha values reported in the original CTQ-SF
Convergent validity coefficients are reported in Table 6. As can be seen, on the whole
all clinical CTQ-SF scales were substantially correlated with the CEQ-R items indicating good
convergent validity. Of the total of 72 significant correlations (ps < .01) between the CTQ-SF scales
and CEQ-R items, 62 reached a medium to large effect size (rs ranged from .31 to .81). Four of the
five CTQ-SF clinical subscales (i.e. Emotional Abuse, Physical Abuse, Sexual Abuse, and Physical
Abuse) showed large positive correlations with their corresponding item on CEQ-R (rs ranged from
.52 to .81). The CTQ-SF Emotional Neglect scale was the only subscale without a distinct counter-
part scale on the CEQ-R. Still, the Emotional Neglect scale showed substantial correlations with
most of the CEQ-R items, notably Emotional withdrawal (r = .63), Lack of real relationship (r =
Discussion
17
The present study is the first to formally evaluate the psychometric performance of the
Danish version of the CTQ-SF in a heterogenous clinical sample. Our study provides support for
the Danish translation of the instrument in terms of structural validity, scale reliability, and multi-
method convergent validity. The original CTQ-SF five-factor model with three correlated error
terms provided the best fit to the data when compared with competing factor models. Additionally,
we could demonstrate factorial congruency of our best fitting model with factor models based on a
Concerning reliability, the obtained coefficient alpha’s were very similar to those
reported in the original CTQ-SF manual, both in terms of magnitude and pattern (Bernstein & Fink,
1998), as well as to those reported in various other studies across countries and groups (e.g.,
Bernstein & Fink, 2011; Gerdner & Allgulander, 2009; Jiang et al., 2018; Thombs et al., 2009).
Consistent with previous research (Bernstein & Fink, 2011; Gerdner & Allgulander, 2009), the
subscales of Physical Abuse and Physical Neglect showed the lowest alpha coefficients, except in
the subsample of patients diagnosed with depression and in the subgroup of men in the combined
sample where the Sexual Abuse subscale was the lowest. Relatedly, the Physical Neglect scale
displayed the lowest internal consistency estimate of all the CTQ-SF clinical scales across all
We also obtained support for the convergent validity of the CTQ-SF clinical scales
with interview-ratings of adverse parental behaviors. This is a strong test of convergent validity,
possibility for shared method variance to inflate the estimates (Campbell & Fiske, 1959).
The applied implications of the present study are that clinicians in Denmark now have
a validated instrument to briefly screen for a wide range of childhood trauma. Effective intervention
hinges on detection. The CTQ-SF can easily be administered as part of routine assessment, to
18
facilitate recognition of various types of childhood trauma in diverse clinical groups. Such
recognition may be beneficial for patients and clinicians when making treatment plans, including
informing assessment of suicide risk, treatment needs, and prognosis. Moreover, assessment of
childhood trauma may also be valuable for formulating individualized case-formulations (Karterud
& Kongerslev, 2019) or make use of adjunct trauma informed care (Hopper, Bassuk, & Olivet,
Some limitations should be considered when interpreting the results of this study. First
and foremost, our findings may not be generalizable to clinical groups different from those we have
investigated. This pertains especially to the convergent validity data which was only performed in a
small subsample of adult outpatients diagnosed with personality disorders. Moreover, our sample
size was too small to allow for factor analytic comparison between men and women and different
age groups. This must be addressed in future research, preferably together with potential differences
based on ethnicity/cultural background, for which we did not have data in the present study. This
way the factor invariance can be further examined across important subgroups. We provided some
preliminary evidence, in this study, for the cross-culturally factorial invariance of the five-factor
structure by calculating coefficients of congruency with a sample from US and Brazil. Still, future
studies could aim at collecting data from different countries to perform multi-group confirmatory
analyses, so that degree of measurement invariance can be directly tested by different measurement
models such as a congeneric model (same pattern of factor loadings), a tau-equivalent model (equal
factor loadings) and a parallel model (equal factor loadings and same amount of error). Also,
though a few prior studies, as mentioned in the introduction of this paper, have found evidence
indicating measurement invariance across age for the CTQ-SF there is indeed a need for more
future studies to test this – the factorial or measurement invariance of the CTQ-SF with respect to
age. Inspired by developmental research, such future studies could preferably use a longitudinal
19
study design to further explore the factorial invariance of the CTQ-SF (Widaman, Ferrer, & Conger,
2010).
Summarized, the findings from the present study provide evidence in support of the
reliability and validity of the Danish version of the CTQ-SF in clinical samples, and generally
suggest adequate psychometric properties comparable to those previously reported in the American
References
Angelakis, I., Gillespie, E. L., & Panagioti, M. (2019). Childhood maltreatment and adult
Bach, B., & Fjeldsted, R. (2017). The role of DSM-5 borderline personality symptomatology and
traits in the link between childhood trauma and suicidal risk in psychiatric patients.
017-0063-7
Bach, B., & Sellbom, M. (2016). Continuity between DSM-5 Categorical Criteria and Traits
Criteria for Borderline Personality Disorder. Canadian Journal of Psychiatry, 61(8), 489-
494. doi:10.1177/0706743716640756
Bach, B., Simonsen, E., Christoffersen, P., & Kriston, L. (2017). The Young Schema Questionnaire
doi:10.1027/1015-5759/a000272
Baker, A. J. L., & Maiorino, E. (2010). Assessment of emotional abuse and neglect with the CTQ:
Issues and estimates. Children and Youth Services Review, 32, 740-748.
20
Bernstein, D. P., & Fink, L. (1998). Childhood Trauma Questionnaire: A Restrospective Self-
Bernstein, D. P., & Fink, L. (2011). Childhood Trauma Questionnaire: A Retrospective Self-Report.
Bernstein, D. P., Stein, J. A., Newcomb, M. D., Walker, E., Pogge, D., Ahluvalia, T., . . . Zule, W.
(2003). Development and validation of a brief screening version of the Childhood Trauma
2134(02)00541-0
attachment, and psychopathology among clinically referred adolescents with and without
Dysregulation, 4, 4. doi:10.1186/s40479-017-0055-7
Bo, S., Sharp, C., Fonagy, P., & Kongerslev, M. (2017). Hypermentalizing, attachment, and
epistemic trust in adolescent BPD: Clinical illustrations. Personality Disorders, 8(2), 172-
182. doi:10.1037/per0000161
Brewin, C. R. (2005). Systematic review of screening instruments for adults at risk of PTSD.
Cakir, S., Tasdelen Durak, R., Ozyildirim, I., Ince, E., & Sar, V. (2015). Childhood trauma and
treatment outcome in bipolar disorder. Journal of Trauma & Dissociation, 17(4), 397-409.
doi:10.1080/15299732.2015.1132489
Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validity by the multitrait-
21
Chen, F., Curran, P. J., Bollen, K. A., Kirby, J., & Paxton, P. (2008). An empirical evaluation of the
use of fixed cutoff points in RMSEA test statistic in structural equation models. Sociological
Claeskens, G., & Hjort, N.L. (2008). Model Selction and Model Averaging. Cambridge: Cambridge
University Press.
Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences. Hillsdale, N.J.: L.
Erlbaum Associates.
Dovran, A., Winje, D., Arefjord, K., & Haugland, B. S. M. (2012). Traumatic events and
year systematic literature review. Journal of Child & Adolescent Trauma, 5(1), 16-32. doi:
10.1080/19361521.2012.644654
Dovran, A., Winje, D., Overland, S. N., Breivik, K., Arefjord, K., Dalsbo, A. S., . . . Waage, L.
doi:10.1111/sjop.12052
Dudeck, M., Vasic, N., Otte, S., Streb, J., Wingenfeld, K., Grabe, H. J., . . . Spitzer, C. (2015).
Factorial validity of the Short Form of the Childhood Trauma Questionnaire (CTQ—SF) in
22
Etain, B., Aas, M., Andreassen, O. A., Lorentzen, S., Dieset, I., Gard, S., . . . Henry, C. (2013).
Gerdner, A., & Allgulander, C. (2009). Psychometric properties of the Swedish version of the
Gilbert, R., Widom, C. S., Browne, K., Fergusson, D., Webb, E., & Janson, S. (2009). Burden and
doi:10.1016/S0140-6736(08)61706-7
Gillespie, S. M., Kongerslev, M. T., Sharp, C., Bo, S., & Abu.Akel, A. M. (2018). Does affective
theory of mind contribute to proactive aggression in boys with conduct problems and
doi:https://doi.org/10.1007/s1057
Grassi-Oliveira, R., Cogo-Moreira, H., Salum, G. A., Brietzke, E., Viola, T. W., Manfro, G. G., . . .
different age groups: findings from confirmatory factor analysis. PLoS One, 9(1), e87118.
doi:10.1371/journal.pone.0087118
Green, J. G., McLaughlin, K. A., Berglund, P. A., Gruber, M. J., Sampson, N. A., Zaslavsky, A. M.,
& Kessler, R. C. (2010). Childhood adversities and adult psychiatric disorders in the
national comorbidity survey replication I: associations with first onset of DSM-IV disorders.
Hernandez, A., Gallardo-Pujol, D., Pereda, N., Arntz, A., Bernstein, D. P., Gaviria, A. M., . . .
23
Questionnaire-Short-Form: factor structure, reliability and association with parenting.
Hopper, E. K., Bassuk, E. L., & Olivet, J. (2009). Shelter from the storm: Trauma-Informed Care in
homelessness services settings. The Open Health Services and Policy Journal, 2, 131-151.
doi:10.2174/1874924001003010080
Jiang, W. J., Zhong, B. L., Liu, L. Z., Zhou, Y. J., Hu, X. H., & Li, Y. (2018). Reliability and
validity of the Chinese version of the Childhood Trauma Questionnaire-Short Form for
doi:10.1371/journal.pone.0208779
doi:10.3389/fpsyg.2019.00518
Kessler, R. C., McLaughlin, K. A., Green, J. G., Gruber, M. J., Sampson, N. A., Zaslavsky, A. M., .
doi:10.1192/bjp.bp.110.080499
Kline, R. B. (2005). Principles and Practice of Structural Equation Modeling. New York:
Guildford Press.
Kongerslev, M., Moran, P., Bo, S., & Simonsen, E. (2012). Screening for personality disorder in
Kongerslev, M. T., Bo, S., Forth, A. E., & Simonsen, E. (2015). Assessment of the affective
24
Unemotional Traits among incarcerated boys: A study of reliability, criterion validity, and
Ladegaard, N., Lysaker, P. H., Larsen, E. R., & Videbech, P. (2014). A comparison of capacities for
social cognition and metacognition in first episode and prolonged depression. Psychiatry
Levey, E. J., Apter, G., & Harrison, A. M. (2016). The global problem of child maltreatment:
Lorenzo-Seva, U., & ten Berge, J. M. F. (2006). Tucker's congruence coefficient as a meaningful
McCrory, E. J., Gerin, M. I., & Viding, E. (2017). Annual Research Review: Childhood
maltreatment, latent vulnerability and the shift to preventative psychiatry - the contribution
of functional brain imaging. Journal of Child Psychology and Psychiatry, 58(4), 338-357.
doi:10.1111/jcpp.12713
Moody, G., Cannings-John, R., Hood, K., Kemp, A., & Robling, M. (2018). Establishing the
Muthén, L. K., & Muthén, B. O. (2017). Mplus User’s Guide (8th ed.). Los Angeles: Muthén &
Muthén.
Read, J., Hammersley, P., & Rudegeair, T. (2018). Why, when and how to ask about childhood
25
Ryu, E. (2011). Effects of skewness and kurtosis on normal-theory based maximum likelihood test
1066-1074. doi:10.3758/s13428-011-0115-7
Sacchi, C., Vieno, A., & Simonelli, A. (2018). Italian validation of the Childhood Trauma
doi:10.1037/tra0000333
Schäfer, I., & Fisher, H. L. (2011). Childhood trauma and psychosis - what is the evidence?
Stoltenborgh, M., Bakermans-Kranenburg, M. J., Alink, L. R. A., & van Ijzendoorn, M. H. (2015).
The prevalence of child maltreatment across the globe: Review of a series of meta-analyses.
Thombs, B. D., Bernstein, D. P., Lobbestael, J., & Arntz, A. (2009). A validation study of the
Thombs, B. D., Lewis, C., Bernstein, D. P., Medrano, M. A., & Hatch, J. P. (2007). An evaluation
Trauelsen, A. M., Bendall, S., Jansen, J. E., Nielsen, H. G. L., Pedersen, M. B., Trier, C. H., . . .
10.1016/j.schres.2015.03.014
26
Tucker, L. R. (1951). A Method for the Synthesis of Factor Analytic Studies. Washington, DC:
Vachon, D. D., Krueger, R. F., Rogosch, F. A., & Cicchetti, D. (2015). Assessment of the harmful
Villano, C. L., Cleland, C., Rosenblum, A., Fong, C., Nuttbrock, L., Marthol, M., & Wallace, J.
(2004). Psychometric utility of the childhood trauma questionnaire with female street-based
Widaman, K. F., Ferrer, E., & Conger, R. D. (2010). Factorial Invariance within Longitudinal
Structural Equation Models: Measuring the Same Construct across Time. Child
Williams, L. M., Debattista, C., Duchemin, A. M., Schatzberg, A. F., & Nemeroff, C. B. (2016).
Childhood trauma predicts antidepressant response in adults with major depression: data
from the randomized international study to predict optimized treatment for depression.
World Health Organization (2017). Child Maltreatment: Infographics. Geneva: World Health
Organization.
Zanarini, M. C., Frankenburg, F. R., Hennen, J., Reich, D. B., & Silk, K. R. (2006). Prediction of
the 10-year course of borderline personality disorder. American Journal of Psychiatry, 163,
27
Zatti, C., Rosa, V., Barros, A., Valdivia, L., Calegaro, V. C., Freitas, L. H., . . . Schuch, F. B.
(2017). Childhood trauma and suicide attempt: A meta-analysis of longitudinal studies from
Zeanah, C. H., & Humphreys, K. L. (2018). Child abuse and neglect. Journal of the American
doi:10.1016/j.jaac.2018.06.007
28
Table 1
Medians, means and standard deviations for the Danish Childhood Trauma Questionnaire-Short Form across samples
EA PA SA EN PN Total MD
Sample Md M Md M Md M Md M Md M Md M Md M
(SD) (SD) (SD) (SD) (SD) (SD) (SD)
Personality 14 13.87 6 7.78 5 7.75 16 15.34 8 9.16 51 53.89 0 0.16
disorder (5.52) (4.21) (4.57) (5.15) (4.42) (18.51) (0.45)
Psychosis 9 10.96 5 6.41 5 6.77 12 12.15 7 8.22 41 44.36 0 0.44
(5.21) (2.52) (4.25) (5.13) (3.21) (15.29) (0.83)
Depression 7 9.31 5 5.90 5 5.70 12 12.50 6 7.17 38 40.59 0 0.51
(4.91) (2.29) (1.55) (4.81) (2.50) (12.68) (0.88)
Delinquent 7 7.33 6 6.48 5 5.08 8 8.58 7 7.36 34 34.81 0 0.85
boys (2.47) (1.63) (0.67) (3.38) (2.49) (7.89) (1.16)
Combined 9 10.98 5 6.83 5 6.59 12 12.64 7 8.20 41 45.19 0 0.43
(5.44) (3.16) (3.71) (5.36) (3.56) (16.65) (0.85)
Men 8 9.39 5 6.88 5 5.63 10 11.39 7 8.12 37 41.32 0 0.57
(4.67) (3.49) (2.15) (5.22) (3.57) (15.29) (0.96)
Women 13 12.91 5 6.76 5 7.76 14 14.15 7 8.29 47 49.88 0 0.26
(5.70) (2.72) (4.74) (5.15) (3.56) (17.06) (0.66)
Note. EA = Emotional Abuse; PA = Physical Abuse; SA = Sexual Abuse; EN = Emotional Neglect; PN = Physical Neglect; MD =
Minimization/Denial.
29
Table 2
Model fit for the confirmatory factor analysis models of the Danish Childhood-Trauma Questionnaire-Short Form in a clinical sample
Sample size
Model 2 (df) RMSEA [90% CI] CFI SRMR AIC adjusted BIC
Original 815.923 (265) .073 [.067, .079] 0.878 0.091 22438.799 22506.439
Original with error terms 673.437 (262) .063 [.057, .069] 0.909 0.071 22261.224 22331.251
Original second order 859.277 (270) .075 [.069, .080] 0.870 0.096 22496.847 22560.508
1-factor model 2379.739 (279) .139 [.134, .144] 0.537 0.207 24685.341 24741.840
Alternative 5-factor model 825.405 (265) .074 [.068, .079] 0.876 0.092 22453.252 22520.892
Second order alternative 5-factor model 864.520 (270) .075 [.069, .081] 0.869 0.096 22507.826 22571.487
4-factor model 899.685 (269) .077 [.072, .083] 0.861 0.098 22554.541 22618.998
Second order 4-factor model 901.130 (271) .077 [.072, .083] 0.861 0.100 22556.733 22619.597
Note. S-B = RMSEA = root mean square error of approximation; CI = confidence interval; CFI = comparative fit index; SRMR =
standardized root mean square residual; AIC = Akaike information criterion; BIC = Bayesian information criterion.
30
Table 3
Standardized factor loadings for the Danish Childhood Trauma Questionnaire-Short Form based
on confirmatory factor analysis in a clinical sample.
Factor/item Loadings
EMOTIONAL ABUSE
3. People in my family called me things like ‘‘stupid,’’ ‘‘lazy,’’ or ‘‘ugly.’’ 0.62
8. I thought that my parents wished I had never been born. 0.78
14. People in my family said hurtful or insulting things to me. 0.83
18. I felt that someone in my family hated me. 0.84
25. I believe I was emotionally abused. 0.71
PHYSICAL ABUSE
9. I got hit so hard by someone in my family that I had to see a doctor or go to the 0.37
hospital.
11. People in my family hit me so hard that it left me with bruises or marks. 0.52
12. I was punished with a belt, a board, a cord, or some other hard object. 0.34
15. I believe that I was physically abused. 0.80
17. I got hit or beaten so badly that it was noticed by someone like a teacher,
neighbor, or doctor. 0.43
SEXUAL ABUSE
20. Someone tried to touch me in a sexual way, or tried to make me touch them. 0.93
21. Someone threatened to hurt me or tell lies about me unless I did something 0.68
sexual with them.
23. Someone tried to make me do sexual things or watch sexual things. 0.88
24. Someone molested me. 0.82
27. I believe that I was sexually abused. 0.90
EMOTIONAL NEGLECT
5. There was someone in my family who helped me feel that I was important or 0.62
special (R).
7. I felt loved (R). 0.86
13. People in my family looked out for each other (R). 0.88
19. People in my family felt close to each other (R). 0.86
28. My family was a source of strength and support (R). 0.92
PHYSICAL NEGLECT
1. I didn’t have enough to eat. 0.48
2. I knew that there was someone to take care of me and protect me (R). 0.73
4. My parents were too drunk or high to take care of the family. 0.57
6. I had to wear dirty clothes. 0.54
26. There was someone to take me to the doctor if I needed it (R). 0.61
Note. All standardized factor loadings were statistically significant at p < .001. (R) denotes a
reverse-scored item. Fit indices were as follows: root mean square error of approximation = 0.063,
comparative fit index = 0.909, and standardized root mean square residual = 0.071.
31
Table 4
Intercorrelations among the Danish Childhood Trauma Questionnaire-Short Form scales in the
combined sample.
EA PA SA EN PN MD
EN .62 –.48
PN –.24
Note. All values are Spearman’s rho correlations. All correlations are significant at p < .001, except
for the correlation between the PA and MD scale which is significant at p < .01. N varies from 391
to 393 due to missing data. EA = Emotional Abuse; PA = Physical Abuse; SA = Sexual Abuse; EN
32
Table 5
Internal consistency (Cronbach’s coefficient alpha) for the Danish Childhood Trauma
Questionnaire-Short Form (CTQ-SF) scales across samples and gender, and in comparison, to the
Sample EA PA SA EN PN Total
Median alpha values based on seven US samples† .89 .82 .92 .89 .66 –
PN = Physical Neglect. †Median Alpha values for the five CTQ-SF subscales reported in the
33
Table 6
Associations between the Danish Childhood Trauma Questionnaire-Short Form and interview-
rated caretaker behavior based on the CEQ-R in a sample of outpatients diagnosed with
personality disorders.
Note. All values are Spearman’s rho correlations. N = 72. CEQ-R = Childhood Experiences
Sexual Abuse; EN = Emotional Neglect; PN = Physical Neglect. *p < .05; **p < .01; ***p < 001;
ns
non-significant.
34