Factor validation of the International Trauma Questionnaire in a sample of trauma-exposed Colombian adults in the MI-VIDA study

Abstract

Background

The ICD-11 introduced a new diagnosis of complex post-traumatic stress disorder (CPTSD) defined by disturbances in self-organisation in addition to traditional post-traumatic stress disorder symptoms. The International Trauma Questionnaire (ITQ) is the established measure of this construct and has been validated for use in a variety of populations and languages; however, evidence for the measure's use in Latin America is limited.

Aims

This study sought to validate the factor structure of the Latin American Spanish version of the ITQ in a trauma-exposed sample in Colombia.

Method

Confirmatory factor analysis was used to assess a range of factor models validated previously, including first- and second-order factor models.

Results

Assessment of fit indices demonstrated that a correlated six-factor model comprised of re-experiencing, avoidance, sense of threat, affect dysregulation, negative self-concept and disturbed relationships provided the best fit for these data. Factor loadings for this model were found to be high and statistically significant.

Conclusion

Results concur with prior research validating the use of alternative language versions of the ITQ internationally, and with the theoretical underpinnings of the CPTSD diagnostic category. The ITQ is therefore a valid measure of CPTSD in this Latin American sample. Further validation research is needed in clinical populations in this region.

The latest iteration of the ICD (ICD-11)¹ includes post-traumatic stress disorder (PTSD) and complex post-traumatic stress disorder (CPTSD) as two related, but distinct, diagnostic categories. PTSD is defined by re-experiencing symptomology (feeling as if a past traumatic experience is happening here and now), avoidance (avoiding internal and external reminders of the trauma) and a persistent sense of threat (SoT; heightened arousal and hypervigilance). The related diagnosis of CPTSD is defined by the presence of these symptoms, plus disturbance in self-organisation (DSO), which encapsulates the three domains of affect dysregulation, negative self-concept (NSC) and disturbance in interpersonal relationships. According to ICD-11, caseness or (probable) diagnosis of PTSD is met when persisting in the previous month with at least one symptom from each domain plus symptom-related functional impairment being endorsed, while CPTSD caseness requires these criteria be met plus endorsement of at least one symptom in each of the DSO domains plus functional impairment related to these symptoms.² There is a concerted effect to ensure the cross-cultural validity of these diagnostic concepts in the ICD-11.³

Conflict-related stress and displacement

CPTSD has been related to prolonged and inescapable trauma exposure and thus is thought to be of particular concern for those with prolonged exposure to conflict⁴ and forced displacement,⁵ that is, being forced to move from one's home and community owing to external factors such as conflict or natural disaster, as these groups experience an undermining of personal resources and safety. These groups may be reconsidered as externally displaced/refugees; those who have crossed an international border to flee such exposures or have been internally displaced; and those who have left their homes for similar reasons but not crossed an international border.⁶ Such experiences come with compound risk factors for traumatic stress outcomes arising from potentially traumatic events (e.g. injury, violence and death) and loss of personal and psychosocial resources.⁷ Indeed, cumulative experience of traumatic stressors is associated with increased risk for CPTSD; reasoning such interrelated stressors represents complex trauma exposure, that is, stressor experiences that are prolonged and inescapable.^8,9 This is supported by evidence suggesting that a significant burden (2–86%) of CPTSD exists in populations of refugees and displaced persons post-migration.¹⁰

Colombia has nationally experienced a prolonged period of trauma and forced displacement through civil conflict between government and paramilitary groups stemming from inter-group discord that has affected population health and well-being in the country.⁷ Epidemiological evidence has suggested lifetime PTSD prevalence rates of between 1.8% and 6.7% in Colombia.¹¹ It has been suggested that those living in rural areas in the country may be at disproportionately heightened risk for traumatic stressor exposure (e.g. physical and armed assault, exposure to injury and death) and adverse mental health outcomes because of the increased levels of civil disorder, violence and displacement experienced.¹² For instance, in Buenaventura, an area with greater exposure to armed conflict, violence and criminal activity, estimated rates of PTSD are as high as 38.7%.¹³ Further to this, it is argued that longstanding exposure to contextual stressors is characteristic of complex trauma, and thus owing to this history of conflict and violence the impact of CPTSD difficulties is considered specifically.¹⁴

A central goal of the ICD-11 Disorders Specifically Associated with Stress chapter is to provide a diagnostic framework that is applicable in diverse contexts, including across cultures and languages.³ There is a need to validate the instruments used to assess the diagnostic concepts within different populations and alternative language administrations to fully ensure cross-cultural validity.³ In line with the goals of the ICD-11 revision, the International Trauma Questionnaire (ITQ) has been developed as a self-report measure of PTSD and CPTSD, and made freely available in a diverse array of languages to remove barriers to its validation and use in varied international contexts.¹⁵

Factor structure of the ITQ

A range of factor models of the ITQ have been proposed,¹⁶ ranging from a univariate model comprising all CPTSD symptoms, to hierarchical models distinguishing PTSD and DSO symptomology by superordinate factors. These alternative factor models that have been theorised in response to ICD-11 proposals for PTSD and CPTSD, but have found differing levels of support related to their underlying dimensionality.¹⁷ Thus, there remains a need to understand these stressor-related disorders and their factor analytic structure from a culturally informed perspective to better inform the assessment of these diagnostic concepts.¹⁸

Studies assessing the factor validation of cross-cultural and alternative language administrations of the ITQ have lent support for the universal applicability of ICD-11 CPTSD, with most studies commonly supporting a six-factor correlated model or two-factor higher-order model.¹⁷ The six-factor correlated model commonly supported consists of re-experiencing, avoidance, SoT, affect dysregulation, NSC and disturbance in interpersonal relationships factors, while the two-factor second-order structure also includes the aforementioned factors plus superordinate PTSD and DSO factors.¹⁷ These findings suggest that both six-factor correlated and higher-order factorial models may provide a useful illustration of CPTSD symptomology and expression.¹⁶ An example of this is the Dari-language ITQ, which has been previously validated with externally displaced asylum seekers from Afghanistan who were resettled in Austria; the study found a PTSD prevalence rate of 31.2%, with the majority of respondents further screening positively for CPTSD (20.7%), highlighting the burden of trauma in this displaced group.¹⁹ Factor analytic results showed a second-order factor structure provided the best fit for the data, a finding in keeping with the ICD-11 proposal and previous evidence.¹⁹ The authors did also note the acceptable, but sub-optimal, fit of the correlated six-factor model of CPTSD in these data. Similar higher-order factor structures have been supported in refugee populations using an English-language proxy measurement of ICD-11 CPTSD.

Non-English language administration of the ITQ

The structure of CPTSD as measured by the Korean-language version of the ITQ administered to North Korean defectors was reported by Baek et al,²⁰ similarly highlighting that the two-factor higher-order and six-factor correlated model provided excellent fit to the data; however, it also concluded that the six-factor correlated model provided the best fit to these data. The authors likewise concluded that these findings support the structural validity of the ITQ, and highlighted that the ultimate conclusion to endorse the six-factor correlated model as the best fitting model is in keeping with results from other non-treatment seeking samples.^17,21,22 In contrast, in a validation of the Arabic-language ITQ in a sample of Syrian refugees now resident in Lebanon, a higher relative rate of CPTSD to PTSD (36.1% v. 25.2%) was found, and data better fit a factor model comprising higher-order PTSD and DSO factors.²³ While both samples were comprised of displaced persons, the variation in context and language administration may be argued to give rise to these differentiated results.

It should be noted that the aforenoted works^19,20,23 specifically contribute validation of the alternate language versions of the ITQ in a variety of displaced groups. Across these studies, support has been found for ITQ validity and consistency with ICD-11 PTSD diagnostic criteria from both correlated and higher-order models in displaced groups. There remains a noted lack of evidence concerning the use of the ITQ to assess these difficulties in Latin America. Fresno et al,²⁴ however, provide an initial validation of the ITQ in an adult Chilean sample replicating the findings previously discussed, where six-factor correlated and two-factor higher-order models provided acceptable fit for the data. The authors concluded that the two-factor higher-order model provided optimal fit, a finding that contrasts with previous evidence that has shown the six-factor model to provide the best fit to data from non-clinical samples.^17,24

Administration of the ITQ in Colombia

Notwithstanding this single prior study utilising data from Latin American participants,²⁴ validating the factor structure of the ITQ with regard to the Latin American Spanish version of this measure with displaced groups and in Colombia is lacking. This is of particular relevance in northern Colombia, a region historically affected by civil conflict and high rates of internal displacement.²⁵ The study of traumatic stress is of great importance in the region as many individuals may report a complex trauma signature, that is, multifaceted experiences of stressors related to conflict violence and displacement, increasing risk for adverse mental health outcomes.⁷ While the extant evidence available highlights the relevance and validity of CPTSD for forcibly displaced persons in a variety of contexts, evidence is critically lacking regarding internally displaced population samples. There is, hence, an explicit need to validate the factor structure of the Latin American ITQ in this context of individuals potentially at-risk for traumatic stress outcomes.

Aims

This study aims to contribute a novel factor validation of the ITQ: the first investigation of the Latin American Spanish version of the measure applied in a trauma-exposed Colombian sample. Drawing on previous work with trauma-exposed displaced person samples, analyses focused on five factor models representing CPTSD, hypothesising that the correlated six-factor and two-factor higher-order models would provide the best fit to these data.¹⁷

Method

Data and sample

These data were collected as part of the Estudio de la Vida Bajo Estres (The Study of Life After Stress/‘MI-VIDA’) study, a case–control investigation of risk and resiliency factors associated with PTSD conducted as a collaboration between researchers in the UK and Colombia. Data were collected in the Cesar and Atlántico departments, northern regions historically affected by the armed conflict in the Colombia.²⁶ Recruitment from this region was specifically targeted given the relative need and research underrepresentation of this population.

Procedure

Participants were eligible for inclusion in this study if they had endured any lifetime experience of trauma. Surveys were administered electronically while participants were attending clinics or recruitment events. All were provided with the opportunity to respond to the self-report measures, or to have a researcher assist them where they were unable to complete the survey independently. Study measures were administered in Latin American Spanish, and data are presented in English within this manuscript.

All data were collected using Research Electronic Data Capture (REDCap) tools hosted by La Miscordia Clinica International.²⁷ All individuals were offered nominal time and travel compensation for their participation so as to not disadvantage any prospective respondents. Written consent was provided by all respondents before participation. Study procedures were approved by the Queen's University Belfast Engineering and Physical Sciences Research Ethics Committee [EPS 21_296] and Universidad Simon Bolivar [PRO-CEI-USB-CE-0351-00].

Materials

Exposure to potentially traumatic life events was measured using an adaptation of the Life Events Checklist [LEC-5].²⁸ The current study utilised the 17 items included in the LEC-5, plus four additional items judged to be potentially relevant to the study population, namely: exposure to physical torture, psychological torture, property damage and forced displacement, adapted from previous assessment of exposure to political violence in the region.²⁹ Exposure to events were binary coded, 1 ‘Yes’, 0 ‘No’, and index trauma was recorded as the ‘worst event experienced’.

Complex PTSD symptomology was measured using the Latin American version of the ITQ.² The ITQ contains 12 items measuring symptoms in line with ICD-11 diagnostic criteria: six symptoms of PTSD and six DSOs. Respondents are asked to rate the extent to which they have been bothered by symptomology and impairment on a 5-point Likert scale from 0 ‘Not at all’ to 4 ‘Extremely’. Diagnostic screening criteria specify that at least one symptom in each domain be endorsed at 2 ‘Moderately’ or above, and each domain plus related impairment be endorsed to meet criteria for probable diagnosis.² In these data the ITQ displayed good psychometric properties (Cronbach's α = 0.92), as did the constituent subscales (PTSD; Cronbach's α = 0.90, DSO; Cronbach's α = 0.90).

Analysis

Confirmatory factor analysis was applied to assess the models specified based on previous factor analytic studies using the ITQ.^4,5,16,17 The models tested are represented graphically in Fig. 1. All analyses were performed using R version 4.1³⁰ (R Core Team; see https://www.R-project.org/) and model fit was estimated using the maximum likelihood algorithm in the ‘lavaan’ R package.^30,31 Model fit was assessed using a range of fit indices: the chi-square test (χ²), comparative fit index (CFI), Tucker–Lewis index (TLI), standardised root mean squared residual (SRMR) and root mean square error of approximation (RMSEA). A non-significant chi-square test is considered indicative of acceptable model fit; however, this index is influenced by larger samples sizes and thus a significant value should not be basis for model rejection.³² Increasing CFI and TLI values (≥0.95) and decreasing SRMR and RMSEA values (≤0.06) are considered indicative of excellent or ‘close’ model fit.³³ Items should also load on to their respective factors to a moderate degree or higher to be said to sufficiently capture the latent construct, ≥0.30.³⁴

Fig. 1 Factor models of the ICD-11 complex post-traumatic stress disorder symptomology tested.

CPTSD, complex post-traumatic stress disorder; PTSD, post-traumatic stress disorder; DSO, disturbance in self-organisation; Re, re-experiencing; Av, avoidance; SoT, sense of threat; AD, affect dysregulation; NSC, negative self-concept; DR, disturbance in interpersonal relationships.

Results

Demographic information and trauma endorsements

These data consisted of N = 557 individuals surveyed at baseline between February 2022 and June 2023 in the MI-VIDA study. The majority of the sample were female (70.32%), heterosexual (81.50%) and a plurality were in a relationship (44.78%). Half the sample identified as Hispanic or indigenous (34.29 and 20.48%, respectively). The majority (98.21%) of the sample reported being in the lower stratums of socioeconomic status. Almost the entire sample (95.91%) reported being a victim of the armed conflict, in line with expectations given the recruitment procedure for the current study. Comprehensive sociodemographic details of the sample may be found in Table 1.

Table 1 Sample sociodemographic information

Inclusion criteria for the primary data collection in the wider MI-VIDA study required participants to endorse at least one lifetime trauma exposure. The mean number of lifetime trauma endorsements in the current study sample was 8.24 (s.d. = 4.17). The most commonly reported lifetime trauma events in the sample were exposure to physical assault (52.06%), exposure to combat/living in a warzone (52.96%) and severe human suffering (54.04%). Notably, the majority of the sample (88.51%) reported experience of forced displacement. With regard to index (self-rated worst) trauma experience, the most commonly reported were exposure to forced displacement (48.64%), sudden/violent death (11.95%) and fire or explosion (10.51%). Summary statistics for lifetime and index trauma exposure may be found in Table 2.

Table 2 Lifetime and index trauma endorsements

Mean (s.d.); n (%).

ICD-11 diagnostics

Symptom endorsements and probable diagnosis according to ICD-11 criteria using the ITQ were inspected (see Table 3). PTSD symptoms were endorsed (item response ≥2) by the majority of participants, as was affect dysregulation and disturbance in interpersonal relationships symptoms. The NSC domain was endorsed less frequently by this sample. In total, over half those surveyed (57.09%) met the criteria for probable PTSD according to scoring criteria. Of these, most met the additional criteria for CPTSD, meaning the effective prevalence rates for PTSD and CPTSD in this sample were 23.69 and 33.40%, respectively.

Table 3 Complex post-traumatic stress disorder (CPTSD) symptom endorsements and diagnostic classification according to ICD-11 criteria

PTSD, post-traumatic stress disorder; DSO, disturbance in self-organisation.

n (%).

Confirmatory factor analysis

The factor models previously specified (see Fig. 1) were estimated. Model fit indices were compared using established cut-offs and relative comparisons (see Table 4). Models 2–5 displayed adequate fit indices according to the CFI and TLI, but only Models 3 and 5 displayed a close fit (≥0.95). In a comparison of these, it was found that both displayed excellent fit on the SRMR, but only Model 3 yielded a RMSEA result below the suggested threshold (≤0.06). Model 3 was therefore selected as the best fitting for these data.

Table 4 Fit indices for factor models compared

Best fitting model denoted by bold text. CFI, comparative fit index; TLI, Tucker–Lewis index; RMSEA, root mean square error of approximation; SRMR, standardised root mean squared residual.

All items displayed strong and significant factor loading on their respective factors (see Table 5). All factors were significantly (P < 0.001) and positively correlated (see Supplementary Material available at https://doi.org/10.1192/bjo.2024.752). PTSD factors were more strongly correlated with one another (r ≥ 0.80), as were DSO factors (r ≥ 0.77). Inter-factor correlations between PTSD and DSO factors remained acceptable, with correlations ranging from 0.50 to 0.72. Inspection of internal reliability for each factor likewise showed these to have good consistency: re-experiencing (α = 0.84), avoidance (α = 0.85), SoT (α = 0.78), affect dysregulation (α = 0.70), NSC (α = 0.86) and disturbance in interpersonal relationships (α = 0.86).

Table 5 Factor loadings for Model 3 (six-factor correlated)

SoT,  sense of threat; NSC,  Negative self-concept; PTSD, post-traumatic stress disorder; DSO, disturbance in self-organisation.

Item labels available in Supplementary File 1.

Discussion

The current study presents the first factor analytic investigation of the Latin American ITQ in a sample of individuals exposed to trauma in Colombia. These findings support the factor structure of this language version, and suggest that the latent structure is in line with results from administrations of other language versions and populations, including of the Latin American ITQ in Chile.^17,24 The correlated six-factor model of CPTSD was found to provide the best fit to these data, a result consistent with those obtained in English-speaking community samples.^4,17

CPTSD risk is attributed to the loss of resources and personal connection; however, the identity memory theory of CPTSD proposed that re-experiencing may be triggered by low-level sensory cues, thus maintaining pathology and distress.³⁵ It is therefore proposed that where individuals remain in contexts when widespread trauma is experienced, and where reminders and cues that suggest negative evaluation of one's identity are present, such as threats posed by other political identities or civil conflict factions, CPTSD may be a prevalent issue. The case is therefore considered of internally displaced persons as these individuals may experience a number of psychosocial stressors and be presented with cues that exacerbate the experience of CPTSD, such as places or people that prompt memories of prior traumatic experiences.

These findings notably align with those of Fresno et al²⁴ who similarly assessed the factor validity of the Latin American Spanish version of the ITQ, concluding that both the six-factor correlated and two-factor higher-order model provided acceptable fit to these data. The conclusions in the current study are, however, more consistent with expectations from previous research suggesting that the correlated six-factor structure is optimal in community samples.¹⁷ The factor structure findings in the current study are likewise reminiscent of those obtained in externally displaced person samples, suggesting simultaneous applicability of the six-factor correlated and higher-order models in such groups.^19,20,23 The divergence in factor solution ultimately providing optimal fit in Latin American Spanish administrations may be attributed to different characteristics of trauma exposure in these samples, as has been considered in previous factor analytic investigations with alternative language administrations of the ITQ.²⁴ For instance, in the current sample the most frequently endorsed potentially traumatic events were severe human suffering (54.04%), exposure to combat/living in a warzone (52.96%) and physical assault (52.06%), whereas those people surveyed by Fresno et al²⁴ most often reported unexpected and violent death of a family member or close person, and sexual assault. Review evidence has suggested that trauma experiences, symptom severity and sample demography may influence results regarding optimal factor model fit across samples.¹⁷ The findings of the current study, of acceptable and optimal model fit supporting the theorised model of ICD-11 (C)PTSD, are in concert with prior evidence from displaced groups;^20,23 however, there is limited evidence relating to internally displaced groups, highlighting the need to extend this evidence base.

The broad consistency in latent structure supported in different language administrations is argued to further support the international and multilingual validity and application of the ITQ as a measure of CPTSD. This finding is consistent with that of previous research in community and forcibly displaced populations.^17,19 The sum of this evidence suggests that both these factor models may be considered valid representations of the latent CPTSD construct, and are replicated in multiple population and alternate language administrations.¹⁷

Factor loading of items was excellent in the current sample, with all loading on to their respective items at 0.60 or greater. It should be noted that most items loaded at 0.80 or greater; however, two items; PTSD5 ‘Being super-alert, watchful, or on guard’ and DSO1 ‘When I am upset, it takes me a long time to calm down’, displayed lower factor loading on SoT and affect dysregulation, respectively. These items may be argued to not represent these latent concepts as strongly as other indicators represent respective latent factors. The finding that these items load with less strength is, however, consistent with other evidence with validation studies using the English-language ITQ,^4,36 theoretically consistent and highly favourable; thus, this finding should not be considered to undermine the structure and utility of the ITQ.

Also considered in this investigation were the probable diagnoses according to ICD-11 criteria in this trauma-exposed sample. The prevalence rates obtained for probable PTSD diagnostics in the current sample were reminiscent of evidence from previous studies of forcibly displaced persons,^19,23 whereby CPTSD was more prevalent relative to PTSD and caseness represented one third to one half of responses. The pooled morbidity rate for PTSD and CPTSD was also comparable to that obtained by Vallieres et al²³ among Syrian refugees. It is therefore suggested that CPTSD may be of greater concern for internally displaced groups relative to classic PTSD difficulties. Notably, the rate of PTSD and CPTSD difficulties in the current study were elevated relative to nationwide estimates of PTSD pathology in Colombia,¹¹ but lower than estimates from areas with greater conflict exposure.¹³ This finding may be attributable to the assessment of a community sample living in an area historically exposed to a great degree of conflict-related stressors.²⁶

It is noted that experience of forced displacement was highly endorsed in the current study sample, and cited by almost half of participants as their most stressful experience. This finding highlights the salience of displacement as a potentially traumatic experience. Forced displacement is associated with diverse compounded trauma experiences and stressors that may undermine personal resources and well-being,³⁷ highlighting the relevance of this experience as a risk factor for CPTSD outcomes and a potential contributing factor to the elevated estimates reported in the current study. The rates of disorder in the current study are reminiscent of those from refugee population samples varied by level of conflict exposure and setting.³⁷ While those with refugee/external migration experiences are faced with difficulty with migration and resettlement in unfamiliar surroundings, those who are internally displaced are met with other difficulties. For instance, trauma cues/reminders in their environment may serve to increase perceived threat and contribute to complex traumatic stress and increase the risk of CPTSD outcomes.^9,38 Further research is required to assess specific cognitions and appraisal of trauma experiences and reminders to test this hypothesis.

Strength and limitations

This study contributes a novel investigation validating the factor structure of the ITQ using a Latin American Spanish administration; however, the findings should be reviewed considering some study limitations.

The representativeness of the sample should be considered. This consists of those recruited to a case–control study advertised as investigating ‘life and health after stress’. There is therefore consideration of some self-selected biases. Likewise, the demographic background of the current sample (majority female, married and low-income status) cannot be determined to be representative of the wider population. It should also be noted that the current study sample was recruited from a community population, and there remains a need to validate these findings in clinical populations in Latin America. These findings should be interpreted with this in mind, and generalisation to wider populations and contexts should be done with caution.

These data and results and based entirely on self-reported symptomology. Conclusions may be strengthened by triangulations of diagnostic factor structure using clinical interviews, for instance using the International Trauma Interview, the diagnostic interview schedule developed in line with ICD-11 PTSD criteria that has seen favourable validation in initial studies.^39,40 Likewise, experience of displacement and victim status were assessed by brief screening questions. Future studies integrating clinician assessment of symptomology may also consider utilising structured interviews of trauma experiences and related appraisals.

Despite these limitations, the contribution of this research as outlined provides valuable evidence supporting the use of the Latin American ITQ and the understanding of ICD-11 CPTSD in a trauma-exposed sample in Colombia. There remains a call for further validation and investigation in Latin America to assess the utility of the ITQ in congruence with clinical assessment.

Despite the limitations noted, this study demonstrated a novel replication of the factor structure of the ITQ administered in Latin American Spanish to a sample exposed to trauma in Colombia. These results support the factor validity of the ICD-11 CPTSD diagnostic construct and support the use of the ITQ as a measure to identify this. The finding that CPTSD is more prevalent than PTSD in this sample suggests that this may be a relevant concern among those internally displaced. The comparable findings of the current study and those previously conducted with trauma-exposed displaced persons suggests support of the ITQ to serve as an internationally validated and standardised measure of traumatic stress. Further research is required to extend and replicate these findings with diverse international samples to realise this goal.