Measuring post-traumatic stress disorder and complex post-traumatic stress disorder using the International Trauma Questionnaire: results from a Hungarian clinical and non-clinical sample

ABSTRACT

Background: The 11th revision of the International Classification of Diseases (ICD-11) simplified the description of post-traumatic stress disorder (PTSD) and also introduced a new trauma-related diagnosis called complex post-traumatic stress disorder (CPTSD). CPTSD is linked to earlier, prolonged interpersonal trauma, and is characterized by a broader range of symptoms, in addition to the core PTSD symptoms. The International Trauma Questionnaire (ITQ) has been developed to assess the new diagnostic criteria.

Objectives: The primary aim of our study was to test the factor structure of the ITQ in a clinical and a non-clinical Hungarian sample. We also examined whether the degree of traumatization or the type of trauma experienced was associated with meeting the criteria for PTSD or CPTSD, or with the severity of PTSD or disturbances in self-organization (DSO) symptoms, in both samples.

Method: A trauma-exposed heterogeneous clinical sample (N = 176) and a non-clinical sample (N = 229) filled out the ITQ and a modified version of the Life Events Checklist (LEC-5). The factor structure of the ITQ was tested by examining the model fit of seven competing confirmatory factor analysis models.

Results: A two-factor second-order model with a second-order PTSD factor (measured by three first-order factors) and a DSO factor (measured directly by six symptoms) had the best fit to the data in both samples if an error correlation was allowed between negative self-concept items. Those in the clinical group who reported more interpersonal and childhood trauma experienced more PTSD and DSO symptoms. Also, there were significant, positive, and weak associations between the total number of different traumas and PTSD and DSO factor scores in both samples.

Conclusion: ITQ was found to be a reliable tool to differentiate between PTSD and CPTSD, two related but distinct constructs in a clinical and a non-clinical trauma-exposed sample in Hungary.

Highlights:

• The distinction between PTSD and DSO as related but separate constructs was validated using the ITQ in a Hungarian clinical and non-clinical sample.

• CPTSD was more frequent than PTSD among general help-seeking clinical clients, while in the non-clinical sample PTSD had a higher prevalence rate.

• Interpersonal trauma in childhood and adulthood was associated with more PTSD and DSO symptoms.

1. Introduction

The 11th revision of the International Classification of Diseases (ICD-11) (World Health Organization [WHO], 2018) simplified the description of post-traumatic stress disorder (PTSD), in comparison to the way it had been formulated in ICD-10, and introduced a new stress-related diagnosis called complex post-traumatic stress disorder (CPTSD).

A growing level of dissatisfaction has been expressed among clinicians and researchers with the previous definition of PTSD, as it covered too many symptoms – including general reactions to stress (e.g. sleep problems) – that could also be found, for example, among the symptoms of generalized anxiety disorder and depression (Brewin et al., 2017; Byllesby et al., 2016; Durham et al., 2015; Stein et al., 2014). This led to high comorbidity and a very heterogeneous PTSD patient group, reducing the clinical utility of the diagnosis.

When developing the new ICD-11 PTSD diagnosis, the goal was to reduce the number of symptoms, and focus on core symptoms that are specific to the disorder: re-experiencing the traumatic event in the present, avoidance of internal or external traumatic reminders, and persistent perceptions of heightened current threat (Maercker et al., 2013; Reed, 2010).

The term CPTSD was first introduced by Herman (1992) as a syndrome that follows prolonged interpersonal trauma, and is characterized by a wider range of symptoms, and higher levels of impairment, compared to PTSD. To be given an ICD-11 CPTSD diagnosis, besides satisfying the diagnostic criteria of ICD-11 PTSD, one must experience additional symptoms representing ‘disturbances in self-organization’ (DSO): affective dysregulation, negative self-concept, and difficulties in forming and maintaining relationships. For both diagnoses, the symptoms should be linked to traumatization, persist for at least several weeks, and cause significant impairment in everyday functioning (WHO, 2018).

Over the past few years, several studies – using clinical and non-clinical samples from different countries – have supported the claim that PTSD and CPTSD represent discrete mental disorders (Brewin, 2020; Brewin et al., 2017), and that they can be adequately assessed by the International Trauma Questionnaire (ITQ) (Cloitre et al., 2015, 2018). The ITQ is a self-report measure specifically developed to assess the criteria of ICD-11 PTSD and CPTSD, which has been translated into more than 20 languages.

In clinical samples, the latent structure of the ITQ was generally best represented by a two-factor (PTSD and DSO) second-order model (where each is measured by three first-order factors) (Karatzias et al., 2016; Vallières et al., 2018; Murphy et al., 2020), while in community and student samples, a correlated six-factor model proved to be the best fitting one in most cases (Ben-Ezra et al., 2018; Ho et al., 2019, 2020). Nonetheless, studies validating the factor structure of the 12-item version (i.e. the final form) of ITQ (Cloitre et al., 2018) are still scarce, and thus, further examination is warranted. Sele et al. (2020), for example, were unable to validate the model with two second-order factors and six first-order factors in a Norwegian clinical sample when the 12-item ITQ was used.

Although the diagnoses of ICD-11 PTSD and CPTSD are differentiated based on their respective symptoms, and not on the type of trauma that generates the symptoms, many studies have attempted to link specific trauma types to these two diagnoses. Knowing more about the possible background of the symptoms is valuable, as it can help the development of effective treatments. Several studies showed that there is a positive association between the number of trauma types experienced by a person, and the severity and complexity of his or her trauma symptoms (Anders et al., 2012; Briere et al., 2008; Cloitre et al., 2001; 2019; Green et al., 2000; Wilker et al., 2015). Childhood trauma, especially sexual abuse, was found to be more strongly associated with DSO than with PTSD (Hyland et al., 2017; Karatzias et al., 2016, 2017, 2019; Maercker et al., 2018; Murphy et al., 2020; Shevlin et al., 2017). Childhood emotional and physical neglect were both shown to increase the risk of developing emotion-regulation problems, and problematic interpersonal behaviours later in life (both are related to DSO items) (Cloitre et al., 2009), and childhood neglect was found to be associated with both PTSD and DSO symptoms (Gilbar et al., 2018; Ho et al., 2021; Murphy et al., 2020; Shevlin et al., 2017). In addition, Frewen et al. (2019) found that childhood neglect was more predictive of PTSD than Criterion A events. In the study by Karatzias et al. (2017), having experienced childhood emotional and physical neglect increased the likelihood of CPTSD classification by almost four times. Another important factor is who is committing the abuse (in cases of interpersonal abuse). Childhood abuse committed by a caregiver or someone close was shown to lead to worse mental health problems than abuse committed by a non-caregiver (Cook et al., 2005; Edwards et al., 2012; Goldberg & Freyd, 2006), and was significantly associated with a diagnosis of CPTSD (Cloitre et al., 2019). It can be seen as betrayal trauma (Freyd, 1996), that is, trauma perpetrated by someone whom a child trusts and on whom they depend for their survival.

Previous studies that tested the factor structure of ITQ in clinical samples typically involved refugees, veterans, or clients of trauma centres, while the use of general help-seeking clinical samples has been scarce (Cloitre et al., 2018; Møller et al., 2020). Furthermore, the ITQ has not been validated in Hungary before now.

Thus, our aims were two-fold: (1) to test the factor structure of the 12-item ITQ (Cloitre et al., 2018) in a general clinical sample and in a non-clinical sample in Hungary; and (2) to examine whether the number of different personally experienced traumas, or the type of trauma, was associated with meeting the criteria for PTSD or CPTSD, or with the severity of PTSD or DSO symptoms, in both samples.

2. Method

2.1. Participants and procedures

Data were collected from 482 volunteers, all above the age of 18 years. They included psychiatric patients with mental disorders (clinical sample, N = 211) and a non-clinical sample (N = 271). Out of these individuals, we excluded those from both groups who did not report, did not remember, or did not want to share an index trauma on the ITQ (described in Section 2.2). The final clinical sample consisted of 176 participants and the non-clinical sample contained 229 people.

Our participants were reached via convenience sampling. The clinical sample was recruited at the Department of Psychiatry and Psychotherapy, Semmelweis University, Budapest, Hungary. Their mental disorders were assessed by structured diagnostic interviews [Structured Clinical Interview for DSM-5 Personality Disorders (SCID-5-PD) (First et al., 2018) and Structured Clinical Interview for DSM-5 Disorders – Clinician Version (SCID-5-CV) (First et al., 2016)] and by psychiatrists in the first week of a 4 week inpatient psychotherapy programme. Patients in an acute state of psychosis, patients with developmental disorders or neurocognitive disorders, and patients who could not complete the questionnaire owing to practical or cognitive limitations were not included in this study. The non-clinical sample was recruited by university students at Pázmány Péter Catholic University, Budapest, via their own social networks, using convenience sampling with strata regarding relatively equal proportions of gender and various age groups (18–20, 21–30, 31–40, 41–50, and 51–60 years). Being in treatment or having been treated for mental disorders were the exclusion criteria for the non-clinical sample. Characteristics of the two sample groups are described in Table 1. A significant difference was shown between the clinical and non-clinical samples in terms of gender, with a lower proportion of males in the clinical sample.

Table 1.

Comparison of the clinical and non-clinical samples.

	Clinical sample (N = 176)	Non-clinical sample (N = 229)	Test statistics (p)	Effect size
Gendera, female	109 (71.24%)	118 (51.53%)	χ2 = 14.78 (< .001)*	φ = .20
Age (years)	33.24 (11.67)	34.84 (13.16)	t = 1.25 (.211)	d = .13
Mean number of clinical diagnoses	2.40 (1.40)
Total number of types of personally experienced traumatic experiencesb	3.93 (2.70)	2.33 (1.92)	t = 6.14 (< .001)*	d = 0.71
Presence of at least one type of personally experienced traumatic experiencec	155 (88.07%)	184 (80.35%)	χ2 = 4.35 (.037)*	φ = .10
Number of types of personally experienced interpersonal traumatic experiencesd
No traumatic experience	56 (38.89%)	135 (59.73%)	χ2 = 22.63 (< .001)*	φ = .25
One traumatic experience	46 (31.94%)	65 (28.76%)
At least two traumatic experiences	42 (29.17%)	26 (11.50%)
Number of types of personally experienced non-interpersonal traumatic experiencesd
No traumatic experience	56 (38.89%)	72 (33.64%)	χ2 = 4.78 (.092)	φ = .12
One traumatic experience	30 (20.83%)	67 (31.31%)
At least two traumatic experiences	58 (40.28%)	75 (35.05%)
Number of types of personally experienced childhood traumatic experiencesd
No traumatic experience	51 (35.42%)	176 (77.19%)	χ2 = 76.39 (< .001)*	φ = .45
One traumatic experience	46 (31.94%)	41 (17.98%)
At least two traumatic experiences	47 (32.64%)	11 (4.82%)
Presence of personally experienced traumatic experiences by its typee
Natural disaster	13 (9.03%)	34 (14.98%)	χ2 = 2.82 (.093)	φ = .09
Fire or explosion	6 (4.17%)	14 (6.17%)	χ2 = 0.69 (.406)	φ = .04
Traffic accident	43 (29.86%)	75 (32.89%)	χ2 = 0.38 (.540)	φ = .03
Serious accident at work, home, or during recreational activity	19 (13.19%)	35 (15.42%)	χ2 = 0.35 (.554)	φ = .03
Exposure to toxic substance	8 (5.56%)	7 (3.14%)	χ2 = 1.30 (.254)	φ = .06
Physical assault	55 (38.19%)	67 (29.39%)	χ2 = 3.11 (.078)	φ = .09
Assault with a weapon	13 (9.03%)	7 (3.07%)	χ2 = 6.16 (.013)*	φ = .13
Sexual assault	35 (24.31%)	14 (6.17%)	χ2 = 25.29 (< .001)*	φ = .26
Other unwanted or uncomfortable sexual experience	49 (34.03%)	28 (12.28%)	χ2 = 25.43 (< .001)*	φ = .26
Combat or exposure to a war-zone	1 (0.69%)	1 (0.44%)	–f	–
Captivity	4 (2.78%)	2 (0.88%)	–f	–
Life-threatening illness or injury	17 (11.81%)	9 (3.98%)	χ2 = 8.24 (.004)*	φ = .15
Severe human suffering	18 (12.50%)	9 (3.96%)	χ2 = 9.51 (.002)*	φ = .16
Sudden, violent death	6 (4.17%)	1 (0.44%)	–f	–
Sudden, unexpected death of someone close to you	46 (31.94%)	83 (37.05%)	χ2 = 1.01 (.316)	φ = .05
Serious injury, harm, or death you caused to someone else	7 (4.86%)	5 (2.19%)	–f	–
Any other stressful event or experience	75 (52.08%)	74 (32.89%)	χ2 = 13.44 (< .001)*	φ = .19
Childhood physical abuse	64 (44.44%)	44 (19.30%)	χ2 = 27.09 (< .001)*	φ = .27
Childhood sexual abuse	22 (15.28%)	6 (2.63%)	χ2 = 20.28 (< .001)*	φ = .23
Neglect (before the age of 18)	87 (49.43%)	13 (5.70%)	χ2 = 101.98 (< .001)*	φ = .50

Note: Data are shown as N (%) or M (SD). χ² = Pearson chi-squared statistic; t = independent samples t-test (in absolute value); φ = phi effect size measure (in absolute value); d = Cohen’s d effect size measure (in absolute value). Percentages in each column represent proportions within the category presented in the given column.

^aCoded as: 0 = male, 1 = female.

^bTotal score of the personally experienced exposure of traumatic experiences based on data from participants who had valid responses on all questions of the Life Events Checklist for DSM-5 (LEC-5) (clinical sample: N = 144; non-clinical sample: N = 210).

^cCalculated based on data from participants who had at least one valid response on the LEC-5 (clinical sample: N = 176; non-clinical sample: N = 229).

^dCalculated based on data from participants who had valid responses on all questions of the given subscale of the LEC-5.

^eCalculated based on data from participants who had a valid response on the specific LEC-5 item (clinical sample: N = 144–176; non-clinical sample: N = 223–228).

^fχ² (and φ) were not calculated as in multiple cells of the cross-table the expected count was less than 5; thus, the assumption of the χ² was violated.

*Significant (p < .050) test statistics.

We obtained informed consent from all participants, and the Research Ethics Committee of Semmelweis University approved the study protocol.

2.2. Measures

2.2.1. Traumatic experiences

A modified version of the Life Events Checklist for DSM-5 (LEC-5) (Weathers et al., 2013) was used to assess traumatic experiences. This self-report measure screens for 16 potentially traumatic events during the responder’s lifetime and includes an additional category for any other traumatic event that was not listed before. Three further items were added to enquire about childhood traumas (before the age of 18): childhood physical abuse (being hit, punched, or hurt by someone responsible for caregiving, such as a parent, foster parent, teacher, or coach), childhood sexual abuse (being touched sexually or being sexually assaulted by an older caregiver, such as a parent, foster parent, teacher, or coach), and childhood neglect (not being properly clothed or fed, or being left without physical care by a caregiver, such as a parent, foster parent, or relative). The total score was calculated for events that respondents had experienced personally (i.e. happened to them), and ranged from 0 to 20.

Furthermore, we defined three subcategories: lifetime interpersonal trauma, lifetime non-interpersonal trauma, and childhood traumatic experiences. The first two categories were defined following Ehring and Quack’s (2010) scoring guidelines (also used by Cloitre et al., 2018 and Hyland et al., 2019). Interpersonal trauma included physical assault, assault with a weapon, sexual assault, other unwanted or uncomfortable sexual experiences, combat or exposure to a warzone, captivity, and serious injury and/or harm and/or death that you caused to someone else. Items related to non-interpersonal trauma were natural disaster, fire or explosion, transportation accident, serious accident at work or home or during recreational activity, exposure to toxic substance, life-threatening illness or injury, severe human suffering, sudden and violent death, and sudden and unexpected death of someone close to you. We did not include the ‘any other very stressful event/experience’ in either category, as the nature of this trauma varied among respondents. Childhood trauma consisted of the three added childhood items.

2.2.2. ICD-11 PTSD and CPTSD

After the subjects had filled out the modified LEC-5, they were instructed to fill out the ITQ (Cloitre et al., 2018), a self-report measure developed to assess ICD-11 PTSD and CPTSD symptoms. The ITQ was translated into Hungarian, back-translated into English, and then reviewed by the authors of the original measure.

The ITQ first asks about the most stressful event in the respondent’s life (index trauma) and the time that has passed since then. The answers to the first question on the ITQ were coded by two researchers into the following categories: childhood interpersonal traumatic experience, adult interpersonal traumatic experience, death or disease or suffering, non-interpersonal traumatic experience, and other, uncategorized types of traumatic experience. Because the coding was not straightforward, we used the distribution of traumas detected on the LEC-5 to explore the frequency of the different types of traumatic experiences.

PTSD is assessed by six symptoms: two items for the following three clusters: re-experiencing (RE), avoidance (AV), and current sense of threat (TH); and three items that enquire about functional impairment linked to these symptoms, all experienced in the past month. DSO are assessed by an additional six items, covering three clusters: affective dysregulation (AD), negative self-concept (NSC), and disturbed relationships (DR); and respondents have to indicate how much these symptoms typically bother them, and whether they have experienced any functional impairment linked to them in the past month.

To rate symptom severity and level of impairment, a five-point Likert scale (from 0 = not at all to 4 = extremely) is used.

The PTSD criteria are fulfilled by those who endorse at least one symptom at a moderate or higher level (score ≥ 2) in each PTSD symptom cluster, and report significant (score ≥ 2) functional impairment in at least one important area of life.

CPTSD criteria are presented if someone, besides satisfying the diagnostic criteria for PTSD, endorses at least one item from each DSO symptom cluster at a moderate or higher severity level, and experiences functional impairment in at least one important life area at a moderate or higher level, linked to the DSO symptoms.

Only one of the two diagnoses can be given to a person, not both.

We also identified those who fulfilled the DSO criteria, irrespective of how they scored on the PTSD criteria.

2.3. Data analysis

First, independent samples t-test and chi-squared statistics were used to compare the clinical and non-clinical samples on different measures of traumatic exposure, and the ITQ criteria (PTSD, CPTSD, DSO), symptom endorsement rates, and severities.

To find the best fitting measurement model in both samples related to the ITQ, using confirmatory factor analysis (CFA), seven competing models were estimated (Figure 1), based on the study by Karatzias et al. (2016). Model 1 is a unidimensional structure, where all symptoms load on a CPTSD factor. Model 2 represents a correlated two-factor structure, where all PTSD and DSO symptoms load directly on their respective factors. Model 3 is a correlated six-factor structure, with factors of RE, AV, TH, AD, NSC, and DR. Model 4 is a second-order factor structure, in which correlations between the six first-order factors are explained by a higher-order CPTSD factor. Model 5 describes a second-order factor structure, with six first-order factors and two correlated higher-order factors (PTSD and DSO). RE, AV, and TH load on the PTSD factor, while AD, NSC, and DR load on DSO. Model 6 is a two-factor second-order model, where PTSD symptoms load on their respective first-order factors (RE, AV, and TH), which load on the second-order PTSD factor, while DSO is measured directly by six symptoms. Model 7 is similar to Model 6, but here PTSD is measured by six items, and second-order DSO is measured by three first-order factors (AD, NSC, and DR).

Figure 1.

Competing measurement models. CPTSD = complex post-traumatic stress disorder; PTSD = post-traumatic stress disorder; DSO = disturbances in self-organization; RE =  re-experiencing; AV = avoidance; TH = current sense of threat; AD = affective dysregulation; NSC = negative self-concept; DR = disturbed relationships.

All models were specified using the maximum likelihood robust to non-normality (MLR) estimation method to avoid bias due to non-normal distributions on the observed continuous indicator variables. Each model was evaluated in terms of model fit, based on values of the root mean squared error of approximation (RMSEA), the standardized root mean square residual (SRMR), the Comparative Fit Index (CFI), and the Tucker–Lewis Index (TLI). In the cases of the RMSEA and SRMR, values < 0.080 indicate adequate model fit and values < 0.050 show good model fit, while for the CFI and the TLI, values > 0.900 indicate adequate model fit and values > 0.950 present good model fit (Brown, 2015). To decide which measurement model presents the optimal solution, models with the highest levels of model fit in each sample were also compared, based on the Satorra–Bentler scaled chi-square difference test (Δχ²). A significant result on the Δχ² can indicate that a given model has a significantly closer fit to the data than a given alternative model (Satorra & Bentler, 2010). The best fitting model was retained for invariance testing and validation analyses.

Three levels of measurement invariance were tested between the clinical and non-clinical samples: (1) configural (i.e. factor loadings and intercepts are freely estimated in both samples); (2) metric- (i.e. factor loadings are fixed at equals in both samples); and (3) scalar invariance models (i.e. factor loadings and intercepts are fixed at equals in both samples) were estimated. Consecutive invariance models were compared in terms of model fit, and the more restrictive invariance model was accepted if the decrease in model fit was below 0.005, 0.010, and 0.025 according to the CFI, RMSEA, and SRMR, respectively (Chen, 2007).

Bivariate correlations were calculated between the total number of personally (directly) experienced types of traumatic experiences and ITQ-based symptom severities. The latter constructs were represented by standardized factor scores, based on the best fitting measurement model. We used independent samples t-tests to compare the average number of traumatic experience types in the two samples and to compare ITQ-based criteria categories in terms of the total number of different types of traumatic experiences. To examine the relationships between symptom severity and the number of personally experienced interpersonal, non-interpersonal, and childhood traumatic experience types (with three categories in each: absence, one, and two or more types of traumatic experiences), one-way analyses of variance (ANOVAs) were performed. Standardized factor scores were calculated based on the best fitting measurement model and were used as dependent variables in the ANOVA models. Chi-squared statistics (χ²) were applied to investigate the links between ITQ-based criteria categories and the number of different types of personally experienced interpersonal, non-interpersonal, and childhood traumatic experiences.

Mplus 8.0 (Muthén & Muthén, 2017) and IBM SPSS Statistics 25.0 statistical software were used to perform the analyses. For CFA, the full information maximum likelihood (FIML) missing data handling method was used. For validation analyses, listwise missing data handling was applied.

3. Results

3.1. Preliminary analyses

Table 1 summarizes the average number of different personally experienced trauma types, and the frequency of different types of traumatic experiences, in the clinical and non-clinical samples.

The mean number of different types of personally experienced traumatic experience, measured by the modified LEC-5, was 3.93 (SD = 2.70) in the clinical sample and 2.33 (SD = 1.92) in the non-clinical sample, and this difference was significant. Significant differences were also shown between the samples in terms of the number of personally experienced interpersonal and childhood traumatic experience types (i.e. higher proportion of those experiencing two or more types in the clinical sample).

In the clinical group, the most frequently indicated trauma types were ‘any other stressful event or experience’, childhood neglect, childhood physical abuse, physical assault, and other unwanted, or uncomfortable sexual experience (52.08% to 34.03%). In the non-clinical group, the trauma types cited most frequently were sudden, unexpected death of someone close, traffic accident, ‘any other stressful event or experience’, physical assault, and childhood physical abuse (37.05% to 19.30%).

The average number of different types of personally experienced trauma was 3.72 (SD = 3.32) in the excluded clinical and 1.31 (SD = 1.38) in the excluded non-clinical group, and this difference was significant [t(35.24) = 3.69, p = .001, d = 1.01].

Table 2 presents the prevalence rates for the ITQ-based PTSD and CPTSD criteria, and PTSD and DSO symptom and dysfunction scores, in the clinical and non-clinical samples.

Table 2.

Post-traumatic stress disorder (PTSD) and disturbances in self-organization (DSO) symptom and dysfunction scores; and PTSD, DSO, and complex post-traumatic stress disorder (CPTSD) criteria prevalence rates.

	Clinical sample (N = 176)	Non-clinical sample (N = 229)	Test statistics (p)	Effect size
Presence of ITQ-based PTSD criteria	15 (8.52%)	17 (7.42%)	χ2 = 0.17 (.684)	φ = .02
Presence of ITQ-based DSO criteria	102 (57.95%)	28 (12.23%)	χ2 = 95.47 (< .001)*	φ = .49
Presence of ITQ-based CPTSD criteria	46 (26.14%)	8 (3.49%)	χ2 = 44.15 (< .001)*	φ = .33
Total PTSD symptom score	10.07 (6.13)	5.83 (5.26)	t = 7.32 (< .001)*	d = 0.75
Total DSO symptom score	12.92 (5.67)	4.95 (4.47)	t = 15.29 (< .001)*	d = 1.58
Total PTSD dysfunction score	5.95 (3.80)	1.99 (2.87)	t = 11.51 (< .001)*	d = 1.20
Total self-organization dysfunction score	7.19 (3.41)	2.15 (2.70)	t = 16.13 (< .001)*	d = 1.67
Endorsement of CPTSD symptom and dysfunction categoriesa
Re-experiencing	104 (59.09%)	87 (38.16%)	χ2 = 17.46 (< .001)*	φ = .21
Avoidance	119 (67.61%)	101 (44.10%)	χ2 = 22.17 (< .001)*	φ = .23
Threat	114 (64.77%)	84 (36.68%)	χ2 = 31.43 (< .001)*	φ = .28
Affective dysregulation	144 (81.82%)	122 (53.28%)	χ2 = 35.97 (< .001)*	φ = .30
Negative self-concept	137 (77.84%)	62 (27.19%)	χ2 = 101.94 (< .001)*	φ = .50
Disturbed relationships	134 (76.14%)	68 (30.09%)	χ2 = 83.92 (< .001)*	φ = .46
PTSD dysfunction	132 (75.00%)	63 (27.63%)	χ2 = 89.25 (< .001)*	φ = .47
Self-organization dysfunction	152 (84.36%)	75 (32.75%)	χ2 = 116.12 (< .001)*	φ = .54

Note: Data are shown as N (%) or M (SD). χ² = Pearson chi-squared statistic; t = independent samples t-test (in absolute value); φ = phi effect size measure (in absolute value); d = Cohen’s d effect size measure (in absolute value). Percentages in each column represent proportions within the category presented in the given column.

ITQ = International Trauma Questionnaire.

^aCoded as: 0 = absence, 1 = presence of at least one endorsed symptom or dysfunction.

*Significant (p < .050) test statistics.

In the clinical sample, PTSD and CPTSD prevalence rates were 8.5% and 26.1%, respectively, while in the non-clinical sample 7.4% fulfilled the PTSD criteria and 3.5% fulfilled the criteria for CPTSD. The DSO criteria (endorsed at least one symptom of all three DSO symptom clusters and had at least one dysfunction at a moderate level) were satisfied in 57.9% of the clinical sample and 12.2% of the non-clinical sample, but only a fragment of them fulfilled the CPTSD criteria, as the rest did not show all the necessary core PTSD symptoms.

The clinical sample had significantly higher levels of PTSD and DSO symptom and dysfunction scores, while participants in the clinical sample also had higher endorsement rates in each trauma symptom cluster and dysfunction category, as well as a higher prevalence rate for CPTSD criteria, compared to individuals in the non-clinical sample.

3.2. Confirmatory factor analysis

Model fit indices for the tested measurement models separately in the clinical and non-clinical samples are presented in Table 3.

Table 3.

Model fit of the measurement models.

	χ2	df	p	RMSEA [90% CI]	CFI	TLI	SRMR
Clinical sample (N = 176)
Model 1	329.41	54	< .001	.170 [.153–.188]	.565	.469	.112
Model 2	192.32	53	< .001	.122 [.104–.141]	.780	.726	.084
Model 3a	62.21	39	.011	.058 [.028–.084]	.963	.938	.051
Model 4b	–	–	–	–	–	–	–
Model 5c	–	–	–	–	–	–	–
Model 6	121.08	50	< .001	.090 [.070–.110]	.888	.852	.071
Model 7c	–	–	–	–	–	–	–
Modified Model 6 with error correlations between NSC1 and NSC2	68.98	49	.031	.048 [.015–.073]	.968	.958	.057
Non-clinical sample (N = 229)
Model 1	390.60	54	< .001	.165 [.150–.181]	.518	.411	.117
Model 2	137.03	53	< .001	.083 [.066–.100]	.880	.850	.074
Model 3d	54.24	39	.053	.041 [.000–.066]	.978	.963	.057
Model 4b	–	–	–	–	–	–	–
Model 5c	–	–	–	–	–	–	–
Model 6	87.15	50	< .001	.057 [.036–.077]	.947	.930	.065
Model 7c	–	–	–	–	–	–	–
Modified Model 6 with error correlations between NSC1 and NSC2	78.25	49	.005	.051 [.028–.072]	.958	.944	.068

Note: χ² = chi-squared statistic; RMSEA [90% CI] = root mean squared error of approximation [90% confidence interval]; CFI = Comparative Fit Index; TLI = Tucker–Lewis Index; SRMR = standardized root mean square residual; NSC = negative self-concept.

^aMeasurement model is not considered owing to a non-positive latent variable covariance matrix. Problems involving the following parameter: correlation between AD (affective dysregulation) and DR (disturbed relationships) factors (i.e. estimate is > 1).

^bMeasurement model is not considered as it was not possible to compute the robust χ² and the standard errors of the model parameter estimates. Problems involving the following parameter: factor loading of the first-order AD factor on the second-order complex post-traumatic stress disorder (CPTSD) factor.

^cMeasurement model is not considered as it was not possible to compute the robust χ² and the standard errors of the model parameter estimates. Problems involving the following parameter: factor loading of the first-order AD factor on the second-order disturbances in self-organization (DSO) factor.

^dMeasurement model is not considered owing to a non-positive latent variable covariance matrix. Problems involving the following parameters: correlations of the AD factor with the other factors (i.e. estimates are > 1).

Models 3, 4, 5, and 7 were not considered in either sample, because these models had non-positive latent variable covariance or first-order derivative product matrices, or it was not possible to compute the robust χ² and the standard errors of the model parameter estimates because of issues related to the AD factor (e.g. correlations that are greater than 1 between AD and the other factors, problematic factor loadings of the first-order AD factor on the second-order CPTSD or DSO factors). Therefore, only Models 1, 2, and 6 were considered to decide which measurement model presents the optimal solution and has the highest levels of model fit in each sample.

Models 1 and 2 had poor model fit in both samples, based on all indices. The highest values on the CFI and TLI and the lowest values on the RMSEA and SRMR were shown for Model 6 in both samples. In the clinical sample, Model 6 showed adequate level of model fit based on the SRMR, in addition to low rates of model fit on the CFI, TLI, and RMSEA. In the non-clinical sample, Model 6 was characterized by adequate model fit on all indices. Owing to the inadequate levels of model fit of Model 6 in the clinical sample, modification indices were scanned if it was possible to obtain a modified version of Model 6, which had more optimal model fit characteristics. Modification indices suggested that closer fit to the data can be attained if error covariance is allowed between the NSC items (i.e. ‘I feel like a failure’ and ‘I feel worthless’). It was considered that it can be appropriate and reasonable to assume correlation between NSC1 and NSC2 items (over the common effect of the DSO factor) because of the similar wording of these two items. That is, this may represent a possible source of unintended measurement error and may contribute to a tendency to give similar answers to NSC1 and NSC2 items. The modified version of Model 6, with an error correlation allowed between NSC1 and NSC2 items, presented adequate to optimal levels of model fit on all indices in both samples. The modified Model 6 demonstrated a significantly closer fit to the data compared to Model 6 (without an error covariance between NSC1 and NSC2) in the clinical sample [Δχ²(1) = 52.35, p < .001] as well as in the non-clinical sample [Δχ²(1) = 49.86, p < .001].

Taken together, in both samples the modified version of Model 6 presented the most optimal rates of model fit, and therefore this model was retained for further analyses. It represents a two-factor second-order model, where PTSD symptoms load on their respective first-order factors (RE, AV, and TH), which load on the second-order PTSD factor, while DSO is measured directly by six symptoms, and error covariance is allowed between NSC1 and NSC2 items. Figure 2 summarizes factor loadings, correlation estimates between latent factors and observed items, and reliability indices for the modified Model 6.

Figure 2.

Factor loadings, correlation estimates, and reliability indices for the modified Model 6. λ = standardized factor loading; r = correlation estimate; ω = model-based internal reliability omega estimate. Estimates with subscript C are presented in the clinical sample, while estimates subscript NC are presented in the non-clinical sample. All estimates are significant at at least the p ≤ .003 level. PTSD = post-traumatic stress disorder; DSO = disturbances in self-organization; RE =  re-experiencing; AV = avoidance; TH = current sense of threat; AD = affective dysregulation; NSC = negative self-concept; DR = disturbed relationships.

All items presented positive, significant, and at least moderately strong factor loadings in both samples. Except for the items of AD1 and AD2, all factor loadings showed strong loadings on the latent factors in both samples. However, the factor loading of AD1 in the non-clinical sample, as well as the factor loadings of AD1 and AD2 in the clinical sample, indicated only moderately strong associations with the DSO factor. The first-order factors of RE, AV, and TH had positive, significant, and strong factor loadings on the PTSD second-order factor in both samples. Positive, significant, and strong correlation was demonstrated between PTSD and DSO latent factors in the clinical sample, while in the non-clinical sample there was a positive, significant, and moderately strong correlation between PTSD and DSO. Error covariance between NSC1 and NSC2 items was positive and significant in both samples, with strong correlation in the clinical sample and moderately strong correlation in the non-clinical sample. Also, RE, AV, TH, PTSD, and DSO factors presented acceptable rates of internal consistency in both samples.

The results of the invariance testing for the modified Model 6 between the clinical and non-clinical samples are described in Supplementary Table S1. The configural and metric invariance models demonstrated predominantly adequate and optimal levels of model fit on all indices (except for the SRMR in the case of the metric invariance model). The scalar invariance model showed low levels of model fit. The comparison of the configural and metric invariance models showed mixed results. On the one hand, the Δχ²-based comparison indicated a significant decrease in model fit for the metric invariance model, in addition to a decrease > 0.005 in the level of the CFI. On the other hand, decreases in the levels of the RMSEA and SRMR were < 0.010 and < 0.025, respectively, which indicated only minor decreases in model fit for the more restrictive metric invariance model compared to the configural model. The Δχ²-based comparison, as well as changes on the CFI, RMSEA, and SRMR, indicated substantial levels of decrease in model fit between the metric and scalar invariance models. Overall, these findings suggested that the factor structure of the modified Model 6 was invariant between the clinical and non-clinical samples. However, it was not possible to demonstrate unequivocally the invariance of the factor loadings between the two samples, while the invariance of the item intercepts was rejected.

3.3. Associations between categorical and continuous measures of the ITQ and the level of personally experienced traumatic exposure types

Significant, positive, and weak associations were shown between the total number of personally (directly) experienced types of traumatic experience and RE symptom scores (clinical sample: r = .18, p = .027; non-clinical sample: r = .16, p = .020), TH symptom scores (clinical sample: r = .21, p = .011; non-clinical sample: r = .17, p = .014), PTSD symptom scores (clinical sample: r = .25, p = .003; non-clinical sample: r = .17, p = .015), and DSO symptom scores (clinical sample: r = .24, p = .003; non-clinical sample: r = .18, p = .009). In the clinical sample, AV symptom scores had a significant, positive, and weak correlation with the total number of personally (directly) experienced types of traumatic experience (r = .27, p = .001), while in the non-clinical sample this correlation was non-significant (r = .09, p = .203).

In the clinical sample, those who fulfilled the DSO criteria experienced significantly more types of traumatic experiences (with moderate effect size) than those without a DSO criterion (Table 4). We did not find significant differences between the average number of types of traumatic experiences of individuals diagnosed with PTSD or CPTSD criteria compared to those without, in either sample.

Table 4.

Comparison of the International Trauma Questionnaire (ITQ)-based post-traumatic stress disorder (PTSD), disturbances in self-organization (DSO), and complex post-traumatic stress disorder (CPTSD) diagnostic groups in terms of the total number of personally experienced trauma exposure types.

	Absence of the given criteria, M (SD)	Presence of the given criteria, M (SD)	t (p)	d
Clinical sample (N = 144)
PTSD criteriaa	3.95 (2.66)	3.77 (3.19)	0.23 (0.822)	0.07
DSO criteriab	3.24 (2.42)	4.41 (2.79)	2.62 (0.010)*	0.44
CPTSD criteriac	3.70 (2.74)	4.53 (2.51)	1.65 (0.101)	0.31
Non-clinical sample (N = 210)
PTSD criteriad	2.29 (1.90)	2.76 (2.14)	0.98 (0.330)	0.25
DSO criteriae	2.24 (1.84)	2.96 (2.37)	1.76 (0.080)	0.38
CPTSD criteriaf	2.29 (1.90)	3.38 (2.20)	1.58 (0.117)	0.57

Note: t = independent samples t-test (in absolute value); d = Cohen’s d effect size measure (in absolute value). The calculations are based on data from participants’ who had valid responses on all questions of the modified version of the Life Events Checklist for DSM-5 (LEC-5).

*Significant (p < .050) test statistics.

^aAbsence: N = 131; presence: N = 13.

^bAbsence: N = 59; presence: N = 85.

^cAbsence: N = 104; presence: N = 40.

^dAbsence: N = 193; presence: N = 17.

^eAbsence: N = 185; presence: N = 25.

^fAbsence: N = 202; presence: N = 8.

We then explored whether there is an association between a growing level of different types of traumatic experience (interpersonal, non-interpersonal, childhood trauma) and PTSD and DSO symptom scores (Table 5) and PTSD and CPTSD criteria. Specific, first-order PTSD symptom scores were not considered during the analyses in order to control for family-wise type I error rates. Moreover, strong correlations between RE, AV, and TH symptom scores in both samples (clinical sample: r = .60–.74, all p < .001; non-clinical sample: r = .70–.86, all p < .001) and their strong correlations with the general (second-order) PTSD symptom score (clinical sample: r = .81–.98, all p < .001; non-clinical sample: r = .82–.98, all p < .001) also explained why the general (second-order) PTSD symptom score was used in the ANOVA models and the specific (first-order) PTSD symptom scores were not.

Table 5.

Comparison of groups with increasing levels of types of interpersonal, non-interpersonal, and childhood traumatic experiences in terms of post-traumatic stress disorder (PTSD) and disturbances in self-organization (DSO) standardized factor scores.

	PTSD						DSO
	M (SD)	F (p)	η2	Post-hoc test: 0 vs 1	Post-hoc test: 0 vs 2+	Post-hoc test: 1 vs 2+	M (SD)	F (p)	η2	Post-hoc test: 0 vs 1	Post-hoc test: 0 vs 2+	Post-hoc test: 1 vs 2+
Clinical sample
Personally experienced interpersonal traumatic experiences
No traumatic experience (N = 56)	−0.17 (1.01)	3.83 (.024)*	.05	p = .653 d = 0.17	p = .018 d = 0.56*	p = .166 d = 0.39	−0.18 (0.99)	3.13 (.047)*	0.04	p = .786 d = 0.14	p = .032 d = 0.52*	p = .199 d = 0.37
One type of traumatic experience (N = 46)	0.00 (0.98)						−0.04 (1.04)
At least two types of traumatic experiences (N = 42)	0.38 (0.95)						0.32 (.92)
Personally experienced non-interpersonal traumatic experiences
No traumatic experience (N = 56)	−0.06 (1.10)	0.64 (.530)	0.01	p = .581 d = 0.23	p = .791 d = 0.12	p = .834 d = 0.14	−0.10 (1.06)	0.54 (.584)	0.01	p = .758 d = 0.17	p = .591 d = 0.19	p > .999 d = 0.00
One type of traumatic experience (N = 30)	0.19 (1.13)						0.08 (1.12)
At least two types of traumatic experiences (N = 58)	0.06 (0.82)						0.08 (0.88)
Personally experienced childhood traumatic experiences
No traumatic experience (N = 51)	−0.17 (0.99)	7.39 (< .001)*	0.10	p > .999 d = 0.00	p = .002 d = 0.69*	p = .003 d = 0.69*	−0.23 (0.99)	7.73 (< .001)*	0.10	p = .979 d = 0.04	p < .001 d = 0.76*	p = .004 d = 0.70*
One type of traumatic experience (N = 46)	−0.17 (1.00)						−0.19 (1.04)
At least two types of traumatic experiences (N = 47)	0.48 (0.88)						0.46 (0.82)
Non-clinical sample
Personally experienced interpersonal traumatic experiences
No traumatic experience (N = 135)	0.02 (0.98)	2.24 (.109)	0.02	p = .481 d = 0.17	p = .401 d = 0.32	p = .154 d = 0.48	−0.10 (0.94)	1.55 (.214)	0.01	p = .431 d = 0.19	p = .433 d = 0.32	p = .864 d = 0.13
One type of traumatic experience (N = 65)	−0.15 (0.97)						0.08 (0.99)
At least two types of traumatic experiences (N = 26)	0.34 (1.16)						0.22 (1.25)
Personally experienced non-interpersonal traumatic experiences
No traumatic experience (N = 72)	−0.09 (0.91)	0.96 (.385)	0.01	p = .708 d = 0.13	p = .361 d = 0.22	p = .825 d = 0.10	−0.02 (0.93)	0.20 (.823)	0.00	p = .992 d = 0.02	p = .879 d = 0.08	p = .839 d = 0.10
One type of traumatic experience (N = 67)	0.03 (0.98)						−0.04 (1.06)
At least two types of traumatic experiences (N = 75)	0.14 (1.15)						0.06 (1.04)
Personally experienced childhood traumatic experiences
No traumatic experience (N = 176)	−0.03 (0.96)	0.88 (.418)	0.01	p = .502 d = 0.22	p = .973 d = 0.06	p = .649 d = 0.25	−0.08 (0.94)	2.61 (.076)	0.02	p = .144 d = 0.39	p = .784 d = 0.22	p = .863 d = 0.15
One type of traumatic experience (N = 41)	0.19 (1.18)						0.31 (1.21)
At least two types of traumatic experiences (N = 11)	−0.09 (0.85)						0.13 (0.98)

Note: Both dependent variables were measured as standardized factor scores. η² = eta-squared effect size measure. Post-hoc tests were conducted using the Games–Howell test. Abbreviations for groups in post-hoc tests: 0 = absence of traumatic experiences; 1 = one type of traumatic experience; 2+ = two or more types of traumatic experiences. d = Cohen’s d effect size measure.

*F-statistics and post-hoc tests with significant (p < .050) results.

In the clinical sample, interpersonal trauma exposure and childhood traumatic experience had significant overall effects on PTSD and DSO symptom scores. Findings of the post-hoc tests indicated that those who reported two or more types of interpersonal traumatic experience had significantly higher levels of PTSD and DSO symptoms (with moderate effect sizes) compared to those participants without interpersonal trauma. Also, those who reported two or more types of childhood traumatic experience had significantly higher levels of PTSD and DSO symptoms (with moderate to strong effect sizes) compared to participants with zero or one type of childhood trauma. In the non-clinical sample, no significant association was found between the growing level of type of interpersonal or childhood traumatization and PTSD and DSO symptom scores. The number of types of non-interpersonal traumatic experience was not linked to higher symptom scores in either sample.

Lastly, in the clinical sample, the presence of two or more types of childhood traumatic experience was significantly and positively associated with the presence of CPTSD criteria, compared to those who reported no childhood trauma (χ² = 9.94, p = .007; odds ratio [95% confidence interval] = 3.31 [1.35–8.14]). Other bivariate associations were non-significant in both samples (see details in Supplementary Table S2.)

4. Discussion

The primary goal of our study was to test the factorial validity of the short form of the ITQ (Cloitre et al., 2018) in a Hungarian clinical and non-clinical sample. The factor structure of the 12-item ITQ has scarcely been examined so far. Thus, the present study may increase the current knowledge on the latent structure of the 12-item ITQ by considering and comparing multiple competing measurement models. By contrast, for example, Sele et al. (2020) tested the model fit of three measurement models in the case of the 12-item ITQ, but only one model was in line with the ICD-11-based CPTSD conceptualization.

In both samples, a two-factor second-order model was proven to be the best fitting one, where PTSD symptoms load on their respective first-order factors (RE, AV, and TH), which load on the second-order PTSD factor, while DSO is measured directly by six symptoms, and error covariance is defined between the NSC items. However, the original version of this two-factor second-order model without the latter residual covariance did not present sufficient levels of model fit in the clinical sample. It was assumed that the correlation between NSC1 and NSC2 items (over the common effect of DSO factor) may represent a possible source of unintended measurement error because of the similar wording of these two items.

On the one hand, this result fits the theoretical model of CPTSD as formulated in the ICD-11, as it differentiates between two separate trauma groups (PTSD and DSO) in addition to discriminating among and containing the three specific dimensions of PTSD (RE, AV, and TH). On the other hand, specific dimensions of DSO (AD, NSC, and DR) were not specified in the model, as the symptom severity of DSO was measured directly by six symptoms. Furthermore, the two-factor second-order model with six first-order factors (which is in line with the ICD-11-based CPTSD model) was not supported by our data owing to non-convergence. Sele et al. (2020), who used a clinical sample, ran into similar problems when they tested the latter measurement model with the 12-item version of the ITQ. It was suggested that the problem may have been explained by the fact that each symptom cluster was described by only two items (Sele et al., 2020). That is, it may be possible that empirical underidentification was presented for some of the estimated measurement models, despite the presence of mathematical identification. The relatively small sample size, as well as lower levels of correlation between some of the indicator variables (e.g. the correlation between AD1 and AD2 items was r = 0.17 and r = 0.00 in the clinical and non-clinical samples, respectively), may also have contributed to convergence issues in the present study. In those models, which were disregarded because of statistical problems, issues were mostly related to the AD factor (e.g. correlations that are greater than 1 between AD and the other factors, problematic factor loadings of the first-order AD factor on the second-order CPTSD or DSO factors). Thus, future studies are warranted to examine further the psychometric properties of the AD factor in the 12-item version of the ITQ. Moreover, model non-convergence of the second-order factor structure, with six first-order factors and two correlated higher-order factors (PTSD and DSO), may highlight the need for revising the 12-item version of the ITQ. For example, by adding new indicators for each specific PTSD and DSO factor, it may be possible to avoid model non-convergence (e.g. by having three indicators per factor). Because of the statistical issues related to AD, it may also be worth considering specifying separate factors of hyperactivity and hypoactivity (instead of a common factor of AD) in the shortened version of the ITQ, and including new indicators separately for hyperactivity and hypoactivity (Vallières et al., 2018).

In our clinical sample, the prevalence rate of CPTSD (26.1%) was higher than that of PTSD (8.5%), which is in line with previous findings (Cloitre et al., 2018; Hyland et al., 2017; 2018; Karatzias et al., 2016; Møller et al., 2020; Murphy et al., 2020; Vallières et al., 2018). Our results are closest to those of Møller et al. (2020), who found that 36% fulfilled the criteria for CPTSD and 8% for PTSD in a Danish psychiatric outpatient sample. The other studies referred to above, which used more traumatized populations, had higher prevalence rates for both diagnoses (up to 37% for PTSD and up to 62.5% for CPTSD).

In our non-clinical sample, more individuals fulfilled the PTSD criteria than the CPTSD criteria (7.4% vs 3.5%). While this result is in line with theoretical expectations (as PTSD is a less severe disorder), findings are mixed in this regard when using community/general population samples (Ben-Ezra et al., 2018; Cloitre et al., 2018, 2019; Ho et al., 2019; Hyland et al., 2019). In non-clinical sample, rates were found between 3.4% and 9% for PTSD, and between 2.6% and 13% for CPTSD, and our data fit in these ranges.

We found that 31.8% of our clinical sample and 8.7% of our non-clinical sample fulfilled the DSO criteria but not the PTSD criteria, while their traumatization was comparable to that of the PTSD group. Maercker et al. (2018) suggested investigating whether a CPTSD variant with subclinical PTSD symptoms should be considered. Prevalence rates in the present study may also underline the possible relevance of a CPTSD variant with subclinical PTSD; thus, future studies are warranted to examine further the clinical characteristics of this category.

While childhood neglect was one of the most frequently cited experiences in the clinical group (reported by 49.43%), only 5.7% of the non-clinical group experienced this. In the latter group, sudden death of a loved one was the most frequent trauma (indicated by 37%). Furthermore, significantly more people in the clinical sample reported childhood physical and sexual abuse, sexual assault, and other unwanted sexual experiences than in the non-clinical group. In both groups, the ‘other stressful event’ category was indicated among the most frequent traumas, which suggests that the other LEC-5 categories may not capture some of the experiences that cause major stress for people. Ben-Ezra et al. (2018) also kept the ‘other’ category of the LEC-5 and obtained a similar result to ours, using a nationally representative sample from Israel. In their study, it was the second most frequently cited trauma type in their CPTSD group and the third most frequently cited trauma in their PTSD group. Even if some participants’ experiences listed under the ‘other stressful event’ category would not be considered a Criterion A event if probed further, they could still be linked to the trauma symptoms. The need to assess trauma beyond the scope of Criterion A events was raised by several studies (Anders et al., 2012; Green et al., 2000; Hyland et al., 2021), as non-Criterion A events were also associated with elevated trauma symptoms. For example, in the study by Hyland et al. (2021), bullying, stalking, emotional abuse, rejection, and neglect were shown to be as strongly associated with PTSD and CPTSD as the Criterion A events.

As expected, those who experienced a greater number of different traumas showed more PTSD and DSO symptoms, in both samples. Furthermore, in the clinical sample, those with two or more lifetime interpersonal or childhood traumas reported more PTSD and DSO symptoms, while a growing level of non-interpersonal trauma did not lead to more symptoms. We did not find any significant associations between specific types of traumas and PTSD or DSO symptoms in the non-clinical group.

Contrary to our expectations, individuals who fulfilled either PTSD or CPTSD criteria did not report a significantly higher number of different traumas than those who did not fulfil such criteria, in either sample. The only significant result that we found between level of traumatization and the odds of fulfilling PTSD or CPTSD criteria was in the clinical sample. Here, having two or more reported childhood traumas was positively associated with satisfying the CPTSD criteria, compared to those who did not have any personally experienced childhood trauma. This finding could partly be linked to the small number of individuals fulfilling the ITQ-based criteria in the samples. In the clinical sample, we had more cases when we considered those who fulfilled the DSO criteria as a separate group (more than 50% of the clinical sample belonged to this category), and we found that those who fulfilled the DSO criteria reported a significantly higher level of traumatization than those who did not fulfil it. The lack of significant results here could also be linked to the use of categorical (absence or presence of diagnoses) variables instead of continuous variables. Other studies that examined the complex relationships between traumatization and trauma symptoms/diagnoses typically used much larger samples than us (> 500, or even > 1000 participants) (Cloitre et al., 2018; Ho et al., 2020; Hyland et al., 2019; Karatzias et al., 2019). Lastly, we measured the severity of traumatization by summing up the number of different types of traumatic events that a person has experienced, and we did not take into account repeated incidents of the same trauma type. The Stressful Life Events Screening Questionnaire (SLESQ) (Goodman et al., 1998) and the Early Trauma Inventory – Self Report (ETI-SR) (Bremner et al., 2007) would be suitable tools for this purpose.

5. Limitations and conclusions

This study has several limitations. First, because of the relatively small sample sizes and their lack of representativeness, our results are only cautiously generalizable to the wider clinical or non-clinical population. Moreover, it should be considered that the relatively small sample size may have contributed to convergence issues in some of the tested models. Secondly, the use of self-report measures (instead of clinician-administered interviews) may have limited the precision of the responses. Thirdly, although we added three adverse childhood experiences to the LEC-5 trauma categories, a more thorough assessment of childhood traumatization is needed, even more so because CPTSD is theoretically linked to prolonged early life trauma. Moreover, it is important to note that the analyses did not consider the potential overlap between different types of traumatic experience, for example, between childhood and adult interpersonal trauma exposure. As of now, various trauma questionnaires are used together with the ITQ, and sometimes more than one questionnaire is used; for example, either the Childhood Trauma Questionnaire (CTQ) (Bernstein et al., 2003) or the Adverse Childhood Experiences (ACE) (Felitti et al., 1998) is administered together with the LEC-5. This can lead to overlap between trauma categories, which happened in our case, as well in relation to the childhood and lifetime physical and sexual abuse items. This problem can be addressed by choosing one comprehensive tool that assesses a wide range of traumatic experiences, all linked to increasing the risk of ICD-11 PTSD and CPTSD. Efforts in this direction have already been made, with the development and testing of the International Trauma Exposure Measure (ITEM) (Hyland et al., 2021). The use of a commonly accepted measure to assess trauma would also make the results of the studies in this field more comparable. Fourthly, we used the ITQ’s first open-ended question to decide which participants we saw as traumatized. Although this is in line with the instruction of the ITQ, and this solution guarantees that the trauma symptoms are linked to an index trauma, the coding of answers is not straightforward, and therefore, using a list of preselected traumatic events may be preferable in a future study.

In conclusion, our study was the first to test the factorial validity of the ITQ in a Hungarian general clinical and non-clinical sample, and supported the distinction between PTSD and DSO as related, but separate, constructs. We also found that CPTSD was a more common condition than PTSD in the clinical sample, while we found the reverse situation in the non-clinical sample. Future studies could explore how typical it is to be exposed to trauma and then fulfil only the DSO criteria, but not the PTSD criteria; and also, how often it happens that seriously traumatized people cannot identify the worst (index) trauma, and because of this, are excluded from these kinds of studies altogether. Irrespective of how trauma is measured, the ITQ could be used as a reliable and quick measure to screen for PTSD and CPTSD in different populations.

Funding Statement

This work was supported by the Hungarian National Research, Development and Innovation Fund [grant number NKFI-132546]; ÚNKP-21-4 and ÚNKP 20-3 New National Excellence Programmes of the Ministry for Innovation and Technology from the National Research, Development and Innovation Fund, and by the ELTE Thematic Excellence Programme 2020, supported by the National Research, Development and Innovation Office [grant number TKP2020-IKA-05].

Disclosure statement

No potential conflict of interest was reported by the authors.