Psychometric validation of the Childhood Trauma Questionnaire-Short Form (CTQ-SF) in a Danish clinical sample

Child Abuse & Neglect 94 (2019) 104026

Contents lists available at ScienceDirect

Child Abuse & Neglect journal homepage: www.elsevier.com/locate/chiabuneg

Research article

Psychometric validation of the Childhood Trauma QuestionnaireShort Form (CTQ-SF) in a Danish clinical sample

T

⁎

Mickey T. Kongersleva,b, , Bo Bacha, Gina Rossic, Anne M. Trauelsend, Nicolai Ladegaarde, Sille S. Løkkegaardf, Sune Boa a

Psychiatric Research Unit, Faelledvej 6, 4200 Slagelse, Region Zealand, Denmark Department of Psychology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark c Faculty of Psychology & Educational Sciences, Personality and Psychopathology Research Group, Vrije Universiteit Brussel (VUB), Pleinlaan 2, 1050 Brussels, Belgium d Psychotherapeutic Ambulatory, Mental Health Center Amager, Digevej 10, Capital Region, 2300 Copenhagen, Denmark e Department of Affective Disorders, Aarhus University Hospital – Psychiatry, Palle Juul-Jensens Boulevard 175, 8200 Aarhus N, Denmark f Danish National Center for Psychotraumatology, Department of Psychology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark b

A R T IC LE I N F O

ABS TRA CT

Keywords: Adverse childhood experiences CTQ Danish Developmental trauma Reliability Validity

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 CTQ-SF has not been previously evaluated in clinical samples. Objectives: To examine the structural validity, internal consistency reliability, and multi-method convergent validity of the CTQ-SF in a heterogenous clinical sample from Denmark. Participants and setting: 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 consistency and multi-method convergent validity with interview-based ratings of adverse 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. 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 comparable to estimates previously published. Multi-method convergent validity associations further corroborated the validity of the CTQ-SF. Conclusion: These findings provide support for the reliability and validity of the Danish version of the CTQ-SF in clinical samples.

⁎

Corresponding author at: Psychiatric Research Unit, Faelledvej 6, 4200 Slagelse, Region Zealand, Denmark. E-mail addresses: [email protected] (M.T. Kongerslev), [email protected] (B. Bach), [email protected] (G. Rossi), [email protected] (A.M. Trauelsen), [email protected] (N. Ladegaard), [email protected] (S.S. Løkkegaard), [email protected] (S. Bo). https://doi.org/10.1016/j.chiabu.2019.104026 Received 25 February 2019; Received in revised form 23 May 2019; Accepted 26 May 2019 Available online 30 May 2019 0145-2134/ © 2019 Elsevier Ltd. All rights reserved.

Child Abuse & Neglect 94 (2019) 104026

M.T. Kongerslev, et al.

1. Background According to the World Health Organization (WHO) 23% of children worldwide 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 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. Unfortunately, however, childhood trauma oftentimes goes unrecognized in clinical 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 it easier for patients to disclose information on this highly sensitive topic. Among the available instruments for the retrospective assessment of childhood 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 including both adolescents and adults, predominantly from North America and Europe, have obtained support for the five-factor structure reported in the CTQSF manual (Bernstein et al., 2003; Dovran et al., 2013; Dudeck et al., 2015; Hernandez et al., 2013; Sacchi, Vieno, & Simonelli, 2018; Thombs, Lewis, Bernstein, Medrano, & Hatch, 2007; Thombs, Bernstein, Lobbestael, & Arntz, 2009). 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; GrassiOliveira 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 important and needed. 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. In summary, though internationally widely used and generally displaying adequate psychometric properties, research on the CTQSF 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 2

Child Abuse & Neglect 94 (2019) 104026

M.T. Kongerslev, et al.

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 measure of problematic parenting behavior, namely the Revised Childhood Experiences Questionnaire (Zanarini, 1992). 2. Methods and materials 2.1. Data and procedures 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 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 briefly be summarized as follows: 1) Patients consecutively enrolled, as part of routine clinical assessment, from March 2012 to June 2014 at a psychiatric outpatient clinic specialized in treatment of personality disorders (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 participants were also administered an interview of parental behavior in childhood (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-based ratings of childhood trauma. Moreover, 19 randomly selected interviews on parenting behavior was also rated by another clinical psychologist to obtain estimates on interrater reliability for this instrument. 2) Patients commencing psychiatric treatment for non-affective first-episode psychosis in Region Zealand in Denmark (n = 101; 26% women; Mage = 23 years, SD = 3.4; Trauelsen et 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. 3) Patients diagnosed with first-episode or prolonged depression recruited from general 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 retardation, under the influence of alcohol or drugs or being psychotic on days of 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 (8%; Gillespie, Kongerslev, Sharp, Bo, & Abu Akel, 2018; Kongerslev, Moran, Bo, & Simonsen, 2012; Kongerslev, Bo, Forth, & Simonsen, 2015). 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 respond to item 16 on the Minimization/Denial scale. 2.2. Materials Childhood Trauma Questionnaire-Short Form (CTQ-SF). The CTQ-SF (Bernstein & Fink, 1998, 2011) is a 28-item retrospective selfreport questionnaire assessing traumatic 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 3

Child Abuse & Neglect 94 (2019) 104026

M.T. Kongerslev, et al.

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 total CTQ-SF score. Revised Childhood Experiences Questionnaire (CEQ-R), Parental Behavior Module. 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, 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 continuous variables was 0.88 (range 0.65–0.99).

2.3. 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 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, 2007), we used modification indices to decide which parameter could be set free. We selected the highest modification index (the value of this represents the estimated decrease in chi-square if a previously fixed parameter were to be estimated), 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 performing a chisquare difference test using the Satorra-Bentler scaled chi-square. 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 sample size increases. Models with the lowest AIC and BIC values are considered to show the best fit to the data. To gauge cross-country construct equivalence by factorial similarity, we calculated 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). Psychometric properties were further explored. To facilitate comparison with the 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.

4

Child Abuse & Neglect 94 (2019) 104026

M.T. Kongerslev, et al.

Table 1 Medians, means and standard deviations for the Danish Childhood Trauma Questionnaire-Short Form across samples. Sample

Personality disorder Psychosis Depression Delinquent boys Combined Men Women

EA

PA

SA

EN

PN

Total

MD

Md

M (SD)

Md

M (SD)

Md

M (SD)

Md

M (SD)

Md

M (SD)

Md

M (SD)

Md

M (SD)

14 9 7 7 9 8 13

13.87 (5.52) 10.96 (5.21) 9.31 (4.91) 7.33 (2.47) 10.98 (5.44) 9.39 (4.67) 12.91 (5.70)

6 5 5 6 5 5 5

7.78 6.41 5.90 6.48 6.83 6.88 6.76

5 5 5 5 5 5 5

7.75 6.77 5.70 5.08 6.59 5.63 7.76

16 12 12 8 12 10 14

15.34 (5.15) 12.15 (5.13) 12.50 (4.81) 8.58 (3.38) 12.64 (5.36) 11.39 (5.22) 14.15 (5.15)

8 7 6 7 7 7 7

9.16 8.22 7.17 7.36 8.20 8.12 8.29

51 41 38 34 41 37 47

53.89 44.36 40.59 34.81 45.19 41.32 49.88

0 0 0 0 0 0 0

0.16 0.44 0.51 0.85 0.43 0.57 0.26

(4.21) (2.52) (2.29) (1.63) (3.16) (3.49) (2.72)

(4.57) (4.25) (1.55) (0.67) (3.71) (2.15) (4.74)

(4.42) (3.21) (2.50) (2.49) (3.56) (3.57) (3.56)

(18.51) (15.29) (12.68) (7.89) (16.65) (15.29) (17.06)

(0.45) (0.83) (0.88) (1.16) (0.85) (0.96) (0.66)

Note. EA = Emotional Abuse; PA = Physical Abuse; SA = Sexual Abuse; EN = Emotional Neglect; PN = Physical Neglect; MD = Minimization/ Denial.

3. Results 3.1. 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. 3.2. Confirmatory factor analysis 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, Δdf = 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 Sexual Abuse scale). 3.3. Competing factor models 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. 3.4. 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, Table 2 Model fit for the confirmatory factor analysis models of the Danish Childhood-Trauma Questionnaire-Short Form in a clinical sample. Model

χ2 (df)

RMSEA [90% CI]

CFI

SRMR

AIC

Sample size adjusted BIC

Original Original with error terms Original second order 1-factor model Alternative 5-factor model Second order alternative 5-factor model 4-factor model Second order 4-factor model

815.923 (265) 673.437 (262) 859.277 (270) 2379.739 (279) 825.405 (265) 864.520 (270) 899.685 (269) 901.130 (271)

.073 .063 .075 .139 .074 .075 .077 .077

0.878 0.909 0.870 0.537 0.876 0.869 0.861 0.861

0.091 0.071 0.096 0.207 0.092 0.096 0.098 0.100

22438.799 22261.224 22496.847 24685.341 22453.252 22507.826 22554.541 22556.733

22506.439 22331.251 22560.508 24741.840 22520.892 22571.487 22618.998 22619.597

[.067, [.057, [.069, [.134, [.068, [.069, [.072, [.072,

.079] .069] .080] .144] .079] .081] .083] .083]

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. 5

Child Abuse & Neglect 94 (2019) 104026

M.T. Kongerslev, et al.

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.’’ 8. I thought that my parents wished I had never been born. 14. People in my family said hurtful or insulting things to me. 18. I felt that someone in my family hated me. 25. I believe I was emotionally abused.

0.62 0.78 0.83 0.84 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 hospital. 11. People in my family hit me so hard that it left me with bruises or marks. 12. I was punished with a belt, a board, a cord, or some other hard object. 15. I believe that I was physically abused. 17. I got hit or beaten so badly that it was noticed by someone like a teacher, neighbor, or doctor.

0.37 0.52 0.34 0.80 0.43

SEXUAL ABUSE 20. Someone tried to touch me in a sexual way, or tried to make me touch them. 21. Someone threatened to hurt me or tell lies about me unless I did something sexual with them. 23. Someone tried to make me do sexual things or watch sexual things. 24. Someone molested me. 27. I believe that I was sexually abused.

0.93 0.68 0.88 0.82 0.90

EMOTIONAL NEGLECT 5. There was someone in my family who helped me feel that I was important or special (R). 7. I felt loved (R). 13. People in my family looked out for each other (R). 19. People in my family felt close to each other (R). 28. My family was a source of strength and support (R).

0.62 0.86 0.88 0.86 0.92

PHYSICAL NEGLECT 1. I didn’t have enough to eat. 2. I knew that there was someone to take care of me and protect me (R). 4. My parents were too drunk or high to take care of the family. 6. I had to wear dirty clothes. 26. There was someone to take me to the doctor if I needed it (R).

0.48 0.73 0.57 0.54 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.

1.00 for Emotional Neglect, and 1.00 for Physical Neglect, respectively. Furthermore, factorial similarity was indicated with a fivefactor 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 Emotional Neglect, and .99 for Physical Neglect, respectively.

3.5. 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 Table 4 Intercorrelations among the Danish Childhood Trauma Questionnaire-Short Form scales in the combined sample.

Total EA PA SA EN PN

EA

PA

SA

EN

PN

MD

.90

.52 .44

.57 .50 .36

.89 .74 .30 .39

.73 .53 .38 .29 .62

–.49 –.44 –.16 –.18 –.48 –.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 = Emotional Neglect; PN = Physical Neglect; MD = Minimization/Denial. 6

Child Abuse & Neglect 94 (2019) 104026

M.T. Kongerslev, et al.

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 median US based alpha coefficients reported in the original manual. Sample

EA

PA

SA

EN

PN

Total

Personality disorder Psychosis Depression Delinquent boys Combined Men Women Median alpha values based on seven US samples†

.87 .88 .87 .79 .89 .89 .88 .89

.86 .75 .82 .50 .82 .88 .70 .82

.91 .94 .75 .97 .92 .87 .93 .92

.90 .92 .88 .85 .92 .92 .90 .89

.81 .67 .52 .47 .73 .73 .73 .66

.94 .92 .91 .85 .94 .94 .93 –

Note. EA = Emotional Abuse; PA = Physical Abuse; SA = Sexual Abuse; EN = Emotional Neglect; PN = Physical Neglect. † Median Alpha values for the five CTQ-SF subscales reported in the original English manual (Bernstein & Fink, 1998).

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 other CTQ-SF scales (rs ranged from –.16 to –.49, ps ranged from < . 01 to < 001., indicating small to moderate effect sizes).

3.6. 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 manual for seven American samples (Bernstein & Fink, 1998).

3.7. Multi-method convergent validity 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 = .62), and Malevolent parenting (r = .63).

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. CEQ-R items Emotional abuse Physical abuse Sexual abuse Emotional withdrawal Physical neglect Inconsistent treatment Emotional denial Failure to protect Lack of real relationship Verbal abuse Parentification of patient Malevolent parenting

EA

PA ***

.52 .38** .43*** .52*** .42*** .56*** .56*** .54*** .60*** .61*** .42*** .66***

SA ***

EN **

.44 .67*** .15ns .38** .37** .26* .25* .46*** .29* .39** .12ns .41***

.39 .18ns .81*** .35** .25* .29* .35** .47*** .37** .33** .26* .56***

PN ***

.51 .41*** .26* .63*** .38** .51*** .52*** .55*** .62*** .50*** .42*** .63***

CTQ Total ***

.61 .40** .31** .66*** .65*** .47*** .58*** .62*** .52*** .58*** .58*** .64***

.61*** .48*** .49*** .66*** .49*** .56*** .60*** .66*** .66*** .60*** .44*** .73***

Note. All values are Spearman’s rho correlations. N = 72. CEQ-R = Childhood Experiences Questionnaire, Parental Behavior Module; EA = Emotional Abuse; PA = Physical Abuse; SA = Sexual Abuse; EN = Emotional Neglect; PN = Physical Neglect. ns non-significant. * p < .05. ** p < .01. *** p < 001. 7

Child Abuse & Neglect 94 (2019) 104026

M.T. Kongerslev, et al.

4. Discussion 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 US and Brazilian sample. 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 samples, except among women in our combined sample. 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, considering the monoconstruct-heteromethod design employed, effectively minimizing the 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 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, 2009) when indicated together with other forms of specialized treatment. 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 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 manual and in previous international studies. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sector. Declaration of interest The authors declare they have no interests to declare. References Angelakis, I., Gillespie, E. L., & Panagioti, M. (2019). Childhood maltreatment and adult suicidality: A comprehensive systematic review with meta-analysis. Psychological Medicine. https://doi.org/10.1017/S0033291718003823 Advance online publication. 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. Borderline Personality Disorder and Emotional Dysregulation, 4, 12. https://doi.org/10.1186/s40479-017-0063-7. Bach, B., & Sellbom, M. (2016). Continuity between DSM-5 categorical criteria and traits criteria for borderline personality disorder. The Canadian Journal of Psychiatry, 61(8), 489–494. https://doi.org/10.1177/0706743716640756. Bach, B., Simonsen, E., Christoffersen, P., & Kriston, L. (2017). The Young Schema Questionnaire 3 short form (YSQ-S3). European Journal of Psychological Assessment, 33(2), 134–143. https://doi.org/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. Bernstein, D. P., & Fink, L. (1998). Childhood trauma questionnaire: A restrospective self-report manualSan Antonio: Pearson.

8

Child Abuse & Neglect 94 (2019) 104026

M.T. Kongerslev, et al.

Bernstein, D. P., & Fink, L. (2011). Childhood trauma questionnaire: A retrospective self-report. Vejledning (Dansk version)Stockholm: Pearson. 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 Questionnaire. Child Abuse & Neglect, 27(2), 169–190. https://doi.org/10.1016/S0145-2134(02)00541-0. Bo, S., & Kongerslev, M. (2017). Self-reported patterns of impairments in mentalization, attachment, and psychopathology among clinically referred adolescents with and without borderline personality pathology. Borderline Personality Disorder and Emotion Dysregulation, 4, 4. https://doi.org/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. https://doi.org/10.1037/per0000161. Brewin, C. R. (2005). Systematic review of screening instruments for adults at risk of PTSD. Journal of Traumatic Stress, 18(1), 53–62. https://doi.org/10.1002/jts. 20007. 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. https://doi.org/10.1080/15299732.2015.1132489. Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validity by the multitrait-multimethod matrix. Psychological Bulletin, 56(2), 81–105. https://doi. org/10.1037/h0046016. 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 Methods & Research, 36(4), 462–494. https://doi.org/10.1177/0049124108314720. Cicchetti, D. (2016). Socioemotional, personality, and biological development: Illustrations from a multilevel developmental psychopathology perspective on child maltreatment. Annual Review of Psychology, 67, 187–211. 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 posttraumatic reactions among children and adolescents in out-of-home placement: A 25-year systematic literature review. Journal of Child & Adolescent Trauma, 5(1), 16–32. https://doi.org/10.1080/19361521.2012.644654. Dovran, A., Winje, D., Overland, S. N., Breivik, K., Arefjord, K., Dalsbo, A. S., ... Waage, L. (2013). Psychometric properties of the Norwegian version of the Childhood Trauma Questionnaire in high-risk groups. Scandinavian Journal of Psychology, 54(4), 286–291. https://doi.org/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 German psychiatric patients, Inmates, and University Students. Psychological Reports, 116(3), 685–703. https://doi.org/10.2466/16.03.PR0. 116k27w5. Etain, B., Aas, M., Andreassen, O. A., Lorentzen, S., Dieset, I., Gard, S., ... Henry, C. (2013). Childhood trauma is associated with severe clinical chracteristics of bipolar disorder. The Journal of Clinical Psychiatry, 74(10), 991–998. https://doi.org/10.4088/JPC.13m08353. Gerdner, A., & Allgulander, C. (2009). Psychometric properties of the swedish version of the childhood trauma questionnaire-short form (CTQ-SF). Nordic Journal of Psychiatry, 63(2), 160–170. https://doi.org/10.1080/08039480802514366. Gilbert, R., Widom, C. S., Browne, K., Fergusson, D., Webb, E., & Janson, S. (2009). Burden and consequences of child maltreatment in high-income countries. Lancet, 373(9657), 68–81. https://doi.org/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 psychopathic tendencies? Child Psychiatry and Human Development, 49(6), 906–916. https://doi.org/10.1007/s1057. Grassi-Oliveira, R., Cogo-Moreira, H., Salum, G. A., Brietzke, E., Viola, T. W., Manfro, G. G., ... Arteche, A. X. (2014). Childhood Trauma Questionnaire (CTQ) in Brazilian samples of different age groups: findings from confirmatory factor analysis. PloS One, 9(1), e87118. https://doi.org/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. Archives of General Psychiatry, 67(2), 113–123. https:// doi.org/10.1001/archgenpsychiatry.2009.186. Hernandez, A., Gallardo-Pujol, D., Pereda, N., Arntz, A., Bernstein, D. P., Gaviria, A. M., ... Gutierrez-Zotes, J. A. (2013). Initial validation of the Spanish Childhood Trauma Questionnaire-Short-Form: Factor structure, reliability and association with parenting. Journal of Interpersonal Violence, 28(7), 1498–1518. https://doi. org/10.1177/0886260512468240. 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. https://doi.org/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 inpatients with schizophrenia. PloS One, 13(12), e0208779. https://doi.org/10.1371/journal.pone.0208779. Karterud, S. W., & Kongerslev, M. T. (2019). A Temperament-Attachment-Mentalization-Based (TAM) theory of personality and its Ddsorders. Frontiers in Psychology, 10(518), https://doi.org/10.3389/fpsyg.2019.00518. Kessler, R. C., McLaughlin, K. A., Green, J. G., Gruber, M. J., Sampson, N. A., Zaslavsky, A. M., ... Williams, D. R. (2010). Childhood adversities and adult psychopathology in the WHO World Mental Health Surveys. The British Journal of Psychiatry, 197(5), 378–385. https://doi.org/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 incarcerated adolescent boys: Preliminary validation of an adolescent version of the Standardised Assessment of Personality – Abbreviated Scale (SAPAS-AV). BMC Psychiatry, 12, 94. https://doi.org/10.1186/1471-244X-10-10. Kongerslev, M. T., Bo, S., Forth, A. E., & Simonsen, E. (2015). Assessment of the affective dimensions of psychopathy with the Danish version of the Inventory of Callous-Unemotional Traits among incarcerated boys: A study of reliability, criterion validity, and construct validity. Scandinavian Journal of Child and Adolescent Psychiatry and Psychology, 3(1), 80–96. https://doi.org/10.21307/sjcapp-2015-008. 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 Research, 220(3), 883–889. https://doi.org/10.1016/j.psychres.2014.10.005. Levey, E. J., Apter, G., & Harrison, A. M. (2016). The global problem of child maltreatment: Perspectives on mechanisms of influence and illness presentation. International Journal of Culture and Mental Health, 10(1), 90–96. https://doi.org/10.1080/17542863.2016.1264440. Lorenzo-Seva, U., & ten Berge, J. M. F. (2006). Tucker’s congruence coefficient as a meaningful index of factor similarity. Methodology, 2(2), 57–64. https://doi.org/10. 1027/1614-2241.2.2.57. 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. https://doi.org/10.1111/jcpp.12713. Moody, G., Cannings-John, R., Hood, K., Kemp, A., & Robling, M. (2018). Establishing the international prevalence of self-reported child maltreatment: A systematic review by maltreatment type and gender. BMC Public Health, 18(1), https://doi.org/10.1186/s12889-018-6044-y. 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 abuse. Advances in Psychiatric Treatment, 13(2), 101–110. https://doi. org/10.1192/apt.bp.106.002840. Ryu, E. (2011). Effects of skewness and kurtosis on normal-theory based maximum likelihood test statistic in multilevel structural equation modeling. Behavior Research Methods, 43(4), 1066–1074. https://doi.org/10.3758/s13428-011-0115-7. Sacchi, C., Vieno, A., & Simonelli, A. (2018). Italian validation of the Childhood Trauma Questionnaire-Short Form on a college group. Psychological Trauma, 10(5), 563–571. https://doi.org/10.1037/tra0000333. Schäfer, I., & Fisher, H. L. (2011). Childhood trauma and psychosis - what is the evidence? Dialogues in Clinical Neuroscience, 13(3), 360–365. 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. Child Abuse Review, 24(1), 37–50. https://doi.org/10.1002/car.2353. Thombs, B. D., Bernstein, D. P., Lobbestael, J., & Arntz, A. (2009). A validation study of the Dutch Childhood Trauma Questionnaire-Short Form: Factor structure, reliability, and known-groups validity. Child Abuse & Neglect, 33(8), 518–523. https://doi.org/10.1016/j.chiabu.2009.03.001https://doi.org/10.1016/j.chiabu. 2009.03.001.

9

Child Abuse & Neglect 94 (2019) 104026

M.T. Kongerslev, et al.

Thombs, B. D., Lewis, C., Bernstein, D. P., Medrano, M. A., & Hatch, J. P. (2007). An evaluation of the measurement equivalence of the Childhood Trauma Questionnaire–Short Form across gender and race in a sample of drug-abusing adults. Journal of Psychosomatic Research, 63(4), 391–398. https://doi.org/10.1016/ j.jpsychores.2007.04.010. Trauelsen, A. M., Bendall, S., Jansen, J. E., Nielsen, H. G. L., Pedersen, M. B., Trier, C. H., ... Simonsen, E. (2015). Childhood adversity specificity and dose-response effect in non-affective first-episode psychosis. Schizophrenia Research, 165(1), 52–59. https://doi.org/10.1016/j.schres.2015.03.014. Tucker, L. R. (1951). A method for the synthesis of factor analytic studies. Washington, DC: Department of the Army. Vachon, D. D., Krueger, R. F., Rogosch, F. A., & Cicchetti, D. (2015). Assessment of the harmful psychiatric and behavioral effects of different forms of child maltreatment. JAMA Psychiatry, 72(11), 1135–1142. https://doi.org/10.1001/jamapsychiatry.2015.1792. 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 sex workers. Journal of Trauma & Dissociation, 5(3), 33–41. https://doi.org/10.1300/J229v05n04_03. Widaman, K. F., Ferrer, E., & Conger, R. D. (2010). Factorial invariance within longitudinal structural equation models: Measuring the same construct across time. Child Development Perspectives, 4(1), 10–18. https://doi.org/10.1111/j.1750-8606.2009.00110.x. 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. Translational Psychiatry, 6(5), https://doi.org/10.1038/ tp.2016.61 e799–e799. World Health Organization (2017). Child maltreatment: Infographics. Geneva: World Health Organization. Zanarini, M. C. (1992). Revised childhood experiences questionnaire (CEQ-R). Belmont, MA: McLean Hospital, Harvard Medical School. 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. The American Journal of Psychiatry, 163, 827–832. https://doi.org/10.1176/ajp.2006.163.5.827. 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 the last decade. Psychiatry Research, 256, 353–358. https://doi.org/10.1016/j.psychres.2017.06.082. Zeanah, C. H., & Humphreys, K. L. (2018). Child abuse and neglect. Journal of the American Academy of Child and Adolescent Psychiatry, 57(9), 637–644. https://doi. org/10.1016/j.jaac.2018.06.007.

10