Skip to main content
Download PDF

The factor structures and correlates of PTSD in post-conflict Timor-Leste: an analysis of the Harvard Trauma Questionnaire

BMC Psychiatry volume 17, Article number: 191 (2017) Cite this article

Abstract

Background

Post-traumatic stress disorder (PTSD) is the most widely assessed form of mental distress in cross-cultural studies conducted amongst populations exposed to mass conflict and displacement. Nevertheless, there have been longstanding concerns about the universality of PTSD as a diagnostic category when applied across cultures. One approach to examining this question is to assess whether the same factor structure can be identified in culturally diverse populations as has been described in populations of western societies. We examine this issue based on an analysis of the Harvard Trauma Questionnaire (HTQ) completed by a large community sample in conflict-affected Timor-Leste.

Method

Culturally adapted measures were applied to assess exposure to conflict-related traumatic events (TEs), ongoing adversities, symptoms of PTSD and psychological distress, and functional impairment amongst a large population sample (n = 2964, response rate: 82.4%) in post-conflict Timor-Leste.

Results

Confirmatory factor analyses of the ICD-10, ICD-11, DSM-IV, four-factor Emotional Numbing and five-factor Dysphoric-Arousal PTSD structures, found considerable support for all these models. Based on these classifications, concurrent validity was indicated by logistic regression analyses which showed that being a woman, trauma exposure, ongoing adversity, severe distress, and functional impairment were all associated with PTSD.

Conclusions

Although symptom prevalence estimates varied widely based on different classifications, our study found a general agreement in PTSD assignments across contemporary diagnostic systems in a large conflict-affected population in Timor-Leste. Further studies are needed, however, to establish the construct and concurrent validity of PTSD in other cultures.

Peer Review reports

Background

Post-traumatic stress disorder (PTSD) is the most widely assessed form of mental distress in cross-cultural studies conducted amongst populations exposed to mass conflict and displacement [1]. Nevertheless, there have been longstanding concerns about the universality of PTSD as a diagnostic category when applied across cultures. One approach to examining this question is to assess whether the same factor structure can be identified in culturally diverse populations as has been described in populations of western societies. We examine this issue based on an analysis of the Harvard Trauma Questionnaire (HTQ) completed by a large community sample in conflict-affected Timor-Leste.

The HTQ is the most widely used measure for assessing PTSD across post-conflict societies of diverse cultural backgrounds [2]. The PTSD symptoms of the HTQ were initially derived from the third edition of the Diagnostic and Statistical Manual (DSM-III) [3], although later studies demonstrated that the items assessed conformed to the three-factor structure (re-experiencing, avoidance/numbing, and arousal) of DSM-IV-TR [4, 5]. The HTQ has been adapted and translated into multiple languages, being applied to conflict-affected and refugee populations from diverse regions of the world, including in Asia [2, 4, 6], the Former Yugoslavia [5, 7], the Middle East [8], and Sub-Saharan Africa [9, 10]. Extensive testing of the measure [5, 7, 11] has supported the validity and reliability of the HTQ PTSD measure in a range of cultures [12].

Establishing the factorial structure of PTSD, even in western populations, has been made more difficult by the serial changes made to the criteria for diagnosing the disorder across successive revisions of the DSM. Although DSM-IV broadly followed the structure of the preceding DSM-III and DSM-III-R by defining three symptom domains of re-experiencing, avoidance/numbing, and arousal [13], in DSM-5, PTSD has undergone a fundamental reformulation with the separation of the numbing and avoidance clusters into two distinct constellations [14].

A recent systematic review of studies undertaken on a variety of measures of PTSD (not including the HTQ) in high income countries has yielded mixed findings for competing models of PTSD in that there was broad evidence for a four-factor structure Emotional Numbing (EN) as well as a five factor Dysphoric-Arousal (DA) structure, with recent studies suggesting that the DA model demonstrated a good fit in several trauma samples in Anglophone countries [15]. Whereas the EN model is consistent with DSM-5 structure, the DA model comprises five domains of intrusions, avoidance, numbing, anxious arousal (startle response, hypervigilance) and dysphoric-arousal (concentration impairment, irritability, and insomnia) [15]. Remarkably, although previous studies have investigated the factorial structure of PTSD based on the HTQ, none has examined specifically for the aforementioned four- and five-factor models [16,17,18,19,20] amongst culturally diverse populations exposed to mass conflict.

In undertaking a comprehensive analysis of the factorial structure of PTSD based on the HTQ, it is necessary also to consider the proposed ICD-11 formulation of the category, especially because it differs substantially from that of DSM-5, both in the number of symptom domains specified and the symptom composition of each. Derived from the ICD-10 structure, ICD-11 represents PTSD according to the conventional three domains (re-experiencing/intrusions, avoidance, and hyper-arousal), although in the more recent revision, the number of items in each domain has been substantially reduced [21]. Studies undertaken in the USA, most amongst survivors of childhood sexual abuse and mass shooting incidents, have supported both the ICD-10 and the ICD-11 PTSD structures [22]. Although the prevalence of ICD-11 PTSD has been examined amongst samples from Cambodia (ICD-11: 8.1%; DSM-IV: 11.2%) and Columbia (ICD-11: 44.4%; DSM-IV: 55%), no study has investigated the factorial structures of both the ICD-10 and ICD-11 PTSD in a culturally diverse population exposed to mass conflict.

The majority of studies using the HTQ have applied a predefined symptom threshold (2.5) for identifying clinical cases of PTSD, a score derived from the addition of item scores, each rated on a frequency scale of 1 to 4, the sum being divided by the number of items. The conventional 2.5 threshold was identified over two decades ago in convergence studies in which DSM –III based clinical interviews were used to calibrate the HTQ amongst refugees from Southeast Asia. In contrast, our more recent concordance study undertaken in Timor-Leste, in which we calibrated the HTQ against the gold-standard clinical structured interview for DSM-IV disorders, found that a threshold score of 2.2 yielded the best balance between specificity and sensitivity on the former measure [11, 23]. As yet, however, there are no data comparing case assignments based on symptom score cut-offs on the HTQ with formulations of PTSD based on the contemporary DSM or ICD systems.

One method for assessing the concurrent validity of contemporary structures of PTSD is by comparing case assignments of competing formulations with known correlates of the disorder. For example, there is extensive evidence in the post-conflict field that PTSD is associated with female gender, high levels of exposure to trauma and ongoing conditions of adversity [1, 24,25,26]. In addition, examining which formulation of PTSD is associated with an index of functional impairment offers a further test of the relative validity of each structure. We therefore examine associations of alternative structures of PTSD with these correlates in our present study amongst the Timorese.

As a study site, Timor-Leste offered an opportunity to test aspects of the construct of PTSD in a society that is culturally distinct from high income settings in which most factorial studies have been undertaken. At the time of the study, the population had minimal exposure to western concepts of traumatic stress or other constructs of mental disorder. The population was exposed to high levels of trauma during the prolonged period of conflict during the Indonesian occupation of the territory (1975–1999), a period of low-grade war in which the indigenous population was exposed to atrocities, extrajudicial murders, incarceration and torture [27]. Many Timorese died as a consequence of violence, forced displacement, famine and untreated disease. Following national independence in 2002, a further period of internal conflict (2006–7) resulted in deaths, injuries, burning of houses and internal displacement of communities [11].

The aims of our analysis were to 1) assess the factorial structure of PTSD according to the ICD-10, ICD-11, DSM-IV, four-factor Emotional Numbing (DSM-5 consistent) and five-factor Dysphoric-Arousal models, respectively. In so doing, we note that the range of items in the HTQ precludes a direct examination of the DSM-5 criteria, the four-factor Emotional Numbing model tested herein therefore representing the closest approximation to that structure; 2) compare the prevalence of PTSD assignments based on these criteria and by the conventional HTQ cut-off score; and as a measure of concurrent validity 3) examine PTSD case assignments with established correlates of PTSD including sociodemographic characteristics, trauma count (TC), adversity count (AC), an index of severe psychological distress, and functional impairment.

Methods

Sample

The study was conducted between May, 2010 and November, 2011, involving a household survey of all men and women, 18-years and older, residing in two villages in Timor-Leste. The sites were an urban administrative area (suco) in Dili, the capital of Timor-Leste, and a rural village located an hour’s drive away. We selected these sites for our earlier survey in 2004 because the Timor-Leste National Directorate of Statistics judged the two resident communities as reflecting the broad range of socio-demographic characteristics of the national population as a whole. Each of the two administrative units is defined by contiguous hamlets (aldeias) under the administration of one chief (chefe). Both locations were directly affected by the longstanding resistance war against the Indonesian occupation and by the subsequent episode of internal conflict that occurred in 2006–7. We used GPS coordinates and aerial maps produced by the Office of Statistics to locate all dwellings in the two locations.

The study was approved by the Human Research Ethics Committee of the University of New South Wales, the Ministry of Health of Timor-Leste, and the chiefs of each village. Participants provided written or witnessed verbal consent.

Measures

We undertook extensive qualitative and quantitative research, serially field testing and refining mental health measures (with reference to a committee including expatriate and Timorese members) to ensure that the mental health constructs we sought to examine were recognized and regarded as commonly experienced in the community. In the process, we refined items to ensure their cultural, semantic and linguistic appropriateness when translated and applied in Timor-Leste, all interviews being conducted in the lingua franca, Tetum [28].

Posttraumatic stress disorder (PTSD) symptoms and psychological distress.

PTSD symptoms were assessed using the relevant section of the HTQ [2], comprising 16 items scored on a four-point scale (1 = none, 2 = some of the time, 3 = a lot of the time, 4 = most of the time). The adapted HTQ included an additional symptom of physiological reactivity in response to reminders of the trauma, dividing the original single item into the two DSM-IV criteria which differentiate between psychological and physiological reactions to reminders.

To assess psychological distress, we used the Kessler-10 scale, consisting of 10 items indexing depressive but also anxiety and somatic symptoms, each item scored on a five-point scale (1 = not at all, 2 = a little of the time, 3 = some of the time, 4 = most of the time, 5 = all of the time).

Both the PTSD (based on 4-point Likert scale) and Kessler-10 (K10) scales demonstrated high levels of internal reliability (HTQ PTSD, Cronbach’s α = 0.95; K-10, α = 0.92). A convergence study conducted previously amongst a subsample of respondents recruited from the survey, compared the HTQ and K10 with the relevant categories of PTSD and major depressive disorder of the Structured Clinical Interview for the Diagnostic and Statistical Manual IV (SCID) applied in a blinded manner by experienced psychologists [23]. There was a sound level of convergence for both indices: Area Under the Curve (AUC) for PTSD 0.82 (95% CI: 0.71–0.94) and for the K10 0.79 (95% CI: 0.67–0.91). An HTQ score of 2.2 provided the best cut-off for PTSD: sensitivity 77.3%, specificity 87.5%, and correct classification 83%. The dichotomized HTQ item pool showed sound reliability (Kuder-Richardson coefficient/KR20 = 0.83). For the K10, the international cut-off score of 30 or more provided the highest level of convergence: sensitivity 92.3%, specificity 66%, and correct classification 71% [29]. The lower specificity is likely to reflect the inclusion of anxiety and somatic symptoms in addition to depressive symptoms in the K10.

Exposure to conflict-related traumatic events

We assessed the 23 conflict-related traumatic events (TEs) listed in the HTQ [2], modified to the context of Timor-Leste. TEs were assessed for both the Indonesian occupation (1975–1999), and the subsequent period following national independence which included the episode of internal conflict of 2006–07. Items involved traumas directed at the self and others, including losses and separations. Typical items included political imprisonment, assault, torture, witnessing murder, exposure to atrocities, losses/separations of family or close others, and severe deprivation of medical care for self or others. We generated a composite trauma count (TC) by collapsing responses assessed for both historical periods; an item endorsed for one or both historical periods was assigned a score of 1 whereas a score of 0 indicated no exposure to that event for either of the two historical periods.

Ongoing adversities

We applied an inventory of ongoing adversities based on community consultations and refinement of items via an iterative process of piloting and feedback. Items included, amongst others, insufficient food, inadequate finances (for school fees, to meet traditional obligations to family), poor shelter, unemployment, and experiences of ongoing conflict (with spouse, children, extended family, young people, and the wider community). Each item was rated on a five-point scale (1 = not a problem, 2 = a bit of a problem, 3 = moderately serious problem, 4 = a serious problem, 4 = a very serious problem). We applied an adversity count (AC) in the present analysis.

Functional impairment

Functional impairment was assessed using a community-derived index. Prior to the survey, the index was developed based on qualitative data gathered from key informant interviews and two focus groups (comprising men and women 18 to 70 years old) involving chiefs of each village and community members [30]. Participants were asked to rate on a five-point scale (1 = not at all, 2 = little, 3 = moderate, 4 = a lot, 5 = often can’t do task) the level of difficulty they experienced in undertaking or performing activities related to four specific items/domains including domestic duties, working/studying, taking care of family, and socializing. We created a composite index based on an addition of all endorsed functional domains using dichotomized items (0 = none, 1 = little/moderate/extreme difficulties).

Field personnel training

Eighteen field personnel received two-weeks training followed by 2 months of field testing and piloting of survey measures supervised by expatriate staff. Pairs of interviewers were required to achieve a consistent 100% inter-rater reliability over five interviews on the symptom measures prior to commencing the study. The interviews lasted an hour and were conducted in participants’ homes in a semi-structured format in which questions were read verbatim to participants, most of whom had low literacy, with additional clarifications and explanations provided as needed to ensure full comprehension.

Statistical analysis

Frequency of endorsement (and percentages) were calculated for individual HTQ symptoms of PTSD. Our preliminary analysis indicated that the responses of HTQ items skewed towards the lower end of the severity spectrum, providing the grounds for dichotomizing scores on statistical ground (0 = not at all/a little and 1 = quite a lot/extremely) [31].

Confirmatory Factor Analysis (CFA) was conducted based on the ICD-10 and the proposed ICD-11 symptom constellations for PTSD as well as for the DSM-IV, the four factor Emotional Numbing and the Dysphoric-Arousal models which approximated the DSM5 structure.

CFA models were estimated using the robust mean- and variance-adjusted Weighted Least Square method (WLSMV), an established statistical procedure recommended for analysing data involving dichotomous variables [31, 32].

We evaluated model fit by using recommended goodness-of-fit and comparative indices, including the chi-square(χ2) test, Comparative Fit Index (CFI), Tucker Lewis Index (TLI), and Root Mean Square Error of Approximation (RMSEA). Specifically, a CFI or TLI above 0.95 and a RMSEA below 0.06 indicate a good fit between the model and the data. A moderate fit is indicated by a CFI above 0.90 and a RMSEA below 0.08 [33,34,35,36]. Given the large sample size, as indicated by our past modelling analyses [37], we anticipated that good fitting model(s) would have a statistically significant chi-square. In the CFA, we calculated standardized factor loadings and the covariance across factors. In general, a factor coefficient of 0.70 or above is considered to be a reliable indicator of a strongly loaded item; and a cross-factorial correlation of 0.90 or above indicates a high correlation between factors.

We assigned cases in each model based on the appropriate HTQ cut-off score and/or HTQ algorithms, in the latter case based on a mapping of items according to ICD-10, ICD-11 and DSM-IV criteria. The Z-test was used to examine for significant  differences in prevalence of PTSD according to these diagnostic criteria. Cohen’s kappa was calculated to assess the level of diagnostic concordance between PTSD assignments. Finally, a series of logistic regression analyses were conducted to examine associations between PTSD assignments according to ICD-10, ICD-11 and DSM-IV criteria and the clinical threshold of 2.2, with relevant socio-demographic characteristics (model 1); trauma count (TC) and adversity count (AC) (model 2); severe psychological distress and incremental levels of functional impairment (model 3). The analysis was not possible for the four factor models as there were no criteria for assigning caseness in these models. The logistic regression results were expressed as odds ratios with 95% confidence intervals (CI). Analyses were performed using STATA version 13 [38] and Mplus version 7 [32].

Results

Socio-demographic characteristics

From the eligible pool of 3597 adults identified in the catchment areas, 2964 completed interviews, a response rate of 82.4% (non-response was due to refusal, and inability of our field staff to make contact in spite of five visits to the dwelling). The analytic sample included 1451 men (49%) and 1513 women (51%) with an overall mean age of 36.4 years.

Table 1 indicates that 62% of the participants resided in rural area and about two thirds (67.9%) were married, a quarter being single/never married (25.5%) and the remainder were widowed or divorced/separated. About 23.9% had completed junior school, 26.3% senior high school, and 10.7% had received post-school education (college/university). A third (34%) were engaged in paid employment (in a range of work including government and private sectors), 35% were unemployed and 6.1% retired; the remainder were involved in subsistence farming/ domestic duties or were unable to work because of physical disability.

Table 1 Socio-demographic characteristics and mental health characteristics of the sample (n = 2964)
Full size table

Threshold scores for symptoms of PTSD severe psychological distress and functional impairment

One in seven (n = 453; 15.3%) met criteria for PTSD based on the clinical HTQ threshold (≥2.2); 15.1% (n = 447) reported severe psychological distress (K10 ≥ 30), and 82.4% (n = 2442) reported difficulties in at least 1 domain of functioning including performing domestic duties, attending school, going to work, attending to the needs of family members, and socializing with others (Table 1).

Confirmatory factor analysis (CFA)

Table 2 maps the constituent items of the HTQ based on the ICD-10, the proposed ICD-11, the DSM-IV, the four factor Emotional Numbing and Dysphoric-Arousal models. Standardized factor loadings for all models tested are presented in Table 3. A good fit was achieved for the three-factor models based on ICD-10 (χ2 (62 df) = 688.59, P ≤ 0.001, CFI = 0.93, TLI = 0.91, RMSEA = 0.058) and ICD-11 (χ2 (6 df) =34.16, P ≤ 0.001, CFI = 0.99, TLI = 0.98, RMSEA = 0.04). The DSM-IV three-factor model (χ2 (116 df) =981.14, P ≤ 0.001, CFI = 0.93, TLI = 0.92, RMSEA = 0.05), the four-factor Emotional Numbing (χ2 (113 df) = 995.95, P ≤ 0.001, CFI = 0.93, TLI = 0.92, RMSEA = 0.051), and the five-factor Dysphoric-Arousal (χ2 (109 df) = 964.93, P ≤ 0.001, CFI = 0.93, TLI = 0.91, RMSEA = 0.05) models each produced moderately good fitting solutions. Table 4 reports the goodness of fit statistics for the sequence of CFA models based on the HTQ PTSD symptom list. Standardized factor loadings for all models tested are presented in Table 3.

Table 2 Mapping items of the Harvard Trauma Questionnaire (HTQ) based on the ICD-10, ICD-11, DSM-IV, four-factor Emotional Numbing, and five-factor Dysphoric-Arousal models
Full size table
Table 3 Standardised factor loadings for ICD-10, ICD-11, DSM-IV, four-factor Emotional Numbing, and five-factor Dysphoric-Arousal models
Full size table
Table 4 Model Fit indices for tested confirmatory factor analysis (CFA) models
Full size table

Prevalence estimates of ICD-10, ICD-11, and DSM-IV PTSD based on symptom criteria

PTSD symptom criteria were met by 46.2% (n = 1369) of the sample for ICD-10, 33.7% (n = 998) for ICD-11 and 38% (n = 1126) for DSM-IV criteria (data not shown). Although comparisons of each classification with another showed differences that in some instances were statistically significant (Table 5), there was a substantial level of agreement across systems in general, specifically between DSM-IV and respectively, the ICD-10 (kappa = 0.83, Z-score = 45.8, P < 0.001) and ICD-11 assignments (kappa = 0.79, Z-score = 43.1), and between ICD-10 and ICD-11 assignments (kappa = 0.74, Z-score = 41.9, P < 0.001). Moderate agreement was found between the ICD-11 and HTQ clinical threshold assignments (kappa = 0.51, Z-score = 31.7, P < 0.0001). In contrast, low agreement was found between the ICD-10 and HTQ clinical threshold case assignment (kappa = 0.35, Z-score = 69.1).

Table 5 Percentage of agreement (kappa) across symptom case assignments derived from the DSM-IV, ICD-10, the proposed ICD-11 PTSD criteria, and the community threshold
Full size table

Assessment of concurrent validity

Multivariate logistic regression analyses were applied to examine associations of PTSD models with sociodemographic variables (model 1); trauma count and adversity count (model 2); and severe psychological distress (K10 > 30) and functional impairment (model 3). Adjusted odds ratios with 95% CIs are presented in Table 6.

Table 6 Logistic regression analyses of socio-demographic (Model 1), psychosocial (Model 2), and mental health predictors (Model 3) of positive PTSD assignments based on the ICD-10, ICD-11, DSM-IV CFA models and the HTQ 2.2 community threshold (PTSD ≥ 2.2)
Full size table

Findings revealed that as compared to men, women were more likely to meet symptom criteria for PTSD according to ICD-10 (OR = 1.55, CI: 1.31–1.84), ICD-11 (OR = 1.86, CI: 1.55–2.22), DSM-IV (OR = 1.80, CI: 1.51–2.15), and the HTQ clinical threshold (OR = 2.36, CI: 1.85–3.01). We note that age, occupational and residency (urban/rural) are likely to be context specific so that the associations shown with different categorizations of PTSD may not have general significance and are not emphasized here.

Trauma count and adversity count all showed associations with ICD-10, ICD-11, DSM-4, and the clinical threshold PTSD assignments. In relation to severe psychological distress and functional impairment, we found a dose-response association with all four classification methods, that is positive case assignment for PTSD categorizations based on ICD-10, ICD-11, DSM-IV and the HTQ clinical threshold assignments were associated statistically with psychological distress and functional impairment, respectively (Table 6).

The adjusted odds ratios (ORs) were largest where the highest level of impairment was reported in all four domains ranging from 1.93 (95%CI: 1.35–2.77) for the DSM-IV assignment to 10.97 (95%CI: 3.48–34.57) for the HTQ clinical threshold assignment (Table 6).

Discussion

Our study is unique in exploring the ICD-10, proposed ICD-11, DSM-IV, four-factor Emotional Numbing and five-factor Dysphoric-Arousal PTSD structures for PTSD according to the HTQ in a culturally distinct population, in this instance, based on a large sample in post-conflict Timor-Leste. Our CFA results found support for all contemporary PTSD factorial models in this population with only marginal differences between them, a result that is consistent with other inquires undertaken amongst Anglophone populations in developed countries. Consistent with the literature and providing support for the concurrent validity of our findings, we found statistical associations between all PTSD models and gender (women reporting a higher prevalence) [39], the quantum of trauma exposure, an index of ongoing adversity, a measure of severe psychological distress, and levels of functional impairment. Notably, however, the prevalence rates of PTSD showed marked variation across the models, with a greater number of persons meeting ICD-10, ICD-11 and DSM-IV criteria compared to those who reached the clinical HTQ threshold.

The strengths of our study include the large sample, the careful approach to recruitment, and high response rate (82.4%). The restriction of our sample to two villages means that further studies will be needed to test the generalizability of our findings to populations in Timor and wider afield. In spite of our systematic approach in adapting and translating our measures [40], we cannot discount the risk of transcultural errors in assessment. Although anamnestic bias can lead to inaccuracy in recording trauma and losses, we note that, in general, the events recorded are consistent with the known history of conflict in Timor-Leste. The culturally adapted HTQ included symptoms based on the DSM-IV-TR, thereby precluding a direct examination of the DSM-5 criteria, the four-factor Emotional Numbing model we tested representing the closest approximation to that structure. The clinical threshold of 2.2 we applied to generate PTSD caseness was based on our clinical calibration of the measure compared with the structured clinical interview for DSM-IV disorders [23]. This finding illustrates the need to redefine the threshold in each setting in that our cutoff differed from the HTO cutoff established in other contexts [4, 8, 10]. Finally, our analysis was restricted to the symptom domains of PTSD given that the HTQ is not designed to assess all aspects of caseness, in particular, associated functional impairment.

These caveats notwithstanding, our key findings, based on the sequence of CFAs conducted, provide support for the capacity of the HTQ to assess PTSD symptoms in this transcultural setting. The measure was found to yield findings consistent with a range of established PTSD factorial models, including the ICD-10, the proposed ICD-11, the DSM-IV, four-factor Emotional Numbing (DSM-5 consistent) and the five-factor Dysphoric-Arousal models of PTSD, an important finding in the transcultural field. The only broadly relevant study in the field was one that found support for the Dysphoric-Arousal model [15] amongst a clinic sample of Arabic speaking refugees undergoing psychiatric treatment in Denmark [41]. With the exception of our study amongst refugees from West Papua [42], no studies have tested the ICD-10 or the proposed ICD-11 PTSD structure in a large post-conflict population. Our findings therefore add further evidence in support of the factorial structures of a range of PTSD classification systems including the ICD-10, ICD-11, DSM-IV, four-factor Emotional Numbing and the five-factor Dysphoric-Arousal models. The finding that all PTSD structures tested in this population provided similarly adequate solutions accords with a recent systematic review of the contemporary PTSD structures in which the current body of studies provided support for both the EN and DA models (noting that the former corresponds directly to the DSM-5 structure) across diverse trauma samples from western countries. Together, the findings provide strong evidence for the separation of numbing and arousal symptoms into two distinct constellations as formulated in DSM-5. In deriving these conclusions, it is important to recognize that CFA allows assessment of the correspondence between the constituent items and their respective domains, only one source of evidence to determine which cluster symptoms are most appropriate to making a clinical diagnosis. Hence a range of studies using various methodologies is needed in order to determine more clearly what group of symptoms best represents a universal constellation of PTSD at a universal level.

Our demonstration of a dose-response relationship between trauma exposure, ongoing adversity and PTSD as assessed by all the models tested is consistent with a well-established association in the post-conflict and refugee field [43]. In addition, logistic regression analysis found that all models of PTSD symptoms were associated with severe psychological distress and incremental levels of functional impairment (in 3 and 4 functional domains), a finding that accords with the post-conflict mental health literature in general [44, 45]. This anticipated pattern of correlates of PTSD across all methods of categorization provides further support for the construct validity of PTSD and the use of the HTQ as a screening measure in this transcultural population. Together, our findings suggest that the adapted HTQ may have ongoing utility in capturing the contemporary construct of PTSD and hence can be used validly as a screening and monitoring instrument in the Timorese population as a whole.

Conclusions

Our study found considerable support for the ICD-10, ICD-11, DSM-IV, four-factor Emotional Numbing (consistent with the DSM-5 formulation of PTSD) and Dysphoric-Arousal PTSD structures in a large conflict-affected population in Timor-Leste. Case assignments using various models showed consistent associations with female gender, trauma exposure, ongoing adversity, severe distress, and functional impairment, providing evidence of concurrent validity of the HTQ symptom measure. Although symptom prevalence estimates varied across classifications, there was adequate agreement in PTSD assignments across the systems. Together, the data suggest that the HTQ represents a robust measure for assessing PTSD symptoms across several models of the disorder, adding to the growing body of evidence supporting the utility of the measure in the transcultural setting.

Abbreviations

A:

Avoidance

AA:

Anxious Arousal

AN:

Avoidance/Numbing

CFA:

Confirmatory factor analysis

CFI:

Comparative Fit Index

DA:

Dysphoric Arousal

DSM:

Diagnostic and Statistical Manual

H:

Hyperarousal

HTQ:

Harvard Trauma Questionnaire

I:

Intrusion

ICD:

International Classification of Diseases

N:

Numbing

PTSD:

Posttraumatic stress disorder

RMSEA:

Root Mean Square Error of Approximation

TLI:

Tucker Lewis Indhex

References

  1. Steel Z, Chey T, Silove D, Marnane C, Bryant RA, Ommeren M. Association of torture and other potentially traumatic events with mental health outcomes among populations exposed to mass conflict and displacement: A systematic review and meta-analysis. J Am Med Assoc. 2009;302(5):537–49.

  2. Mollica RF, Caspi-Yavin Y, Bollini P, Truong T, Tor S, Lavelle J. The Harvard Trauma Questionnaire: Validating a cross-cultural instrument for measuring torture, trauma, and posttraumatic stress disorder in Indochinese refugees. J Nerv Ment Dis. 1992;180(2):111–6.

    Article  CAS  PubMed  Google Scholar 

  3. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. Third ed. Washington, DC: American Psychiatric Association Press; 1980.

    Google Scholar 

  4. Fawzi MC, Pham T, Lin L, Nguyen TV, Ngo D, Murphy E, Mollica RF. The validity of posttraumatic stress disorder among Vietnamese refugees. J Trauma Stress. 1997;10(1):101–8.

    CAS  PubMed  Google Scholar 

  5. Oruc L, Kapetanovic A, Pojskic N, Miley K, Forstbauer S, Mollica FR, Henderson DC. Screening for PTSD and depression in Bosnia and Herzegovina: validating the Harvard Trauma Questionnaire and the Hopkins Symptom Checklist. Int J Cult Ment Health. 2008;1(2):105–16.

    Article  Google Scholar 

  6. Lhewa D, Banu S, Rosenfeld B, Keller A. Validation of a tibetan translation of the hopkins symptom checklist-25 and the harvard trauma questionnaire. Assessment. 2007;14(3):223–30.

    Article  PubMed  Google Scholar 

  7. Jakobsen M, Thoresen S, Johansen LEE. The validity of screening for post-traumatic stress disorder and other mental health problems among asylum seekers from different countries. J Refug Stud. 2011;24(1):171–86.

    Article  Google Scholar 

  8. Shoeb M, Weinstein H, Mollica R. The Harvard trauma questionnaire: adapting a cross-cultural instrument for measuring torture, trauma and posttraumatic stress disorder in Iraqi refugees. Int J Soc Psychiatry. 2007;53(5):447–63.

    Article  PubMed  Google Scholar 

  9. de Fouchier C, Blanchet A, Hopkins W, Bui E, Ait-Aoudia M, Jehel L. Validation of a French adaptation of the Harvard Trauma Questionnaire among torture survivors from sub-Saharan African countries. Eur J Psychotraumatol. 2012;3. doi:10.3402/ejpt.v3i0.19225.

  10. Kleijn WC, Hovens JE, Rodenburg JJ. Posttraumatic stress symptoms in refugees: Assessments with the Harvard Trauma Questionnaire and the Hopkins symptom checklist-25 in different languages. Psychol Rep. 2001;88(2):527–32.

    Article  CAS  PubMed  Google Scholar 

  11. Silove D, Manicavasagar V, Mollica R, Thai M, Khiek D, Lavelle J, Tor S. Screening for depression and PTSD in a Cambodian population unaffected by war: comparing the Hopkins Symptom Checklist and Harvard Trauma Questionnaire with the structured clinical interview. J Nerv Ment Dis. 2007;195(2):152–7.

    Article  PubMed  Google Scholar 

  12. Hollifield M, Warner TD, Lian N, Krakow B, Jenkins JH, Kesler J, Stevenson J, Westermeyer J. Measuring trauma and health status in refugees: a critical review. JAMA. 2002;288(5):611–21.

    Article  PubMed  Google Scholar 

  13. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. Forth ed. Washington, DC: American Psychiatric Association Press; 1994.

    Google Scholar 

  14. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. Fifth ed. Washington, DC: American Psychiatric Association Press; 2013.

    Book  Google Scholar 

  15. Armour C, Mullerova J, Elhai JD. A systematic literature review of PTSD's latent structure in the Diagnostic and Statistical Manual of Mental Disorders: DSM-IV to DSM-5. Clin Psychol Rev. 2016;44:60–74.

    Article  PubMed  Google Scholar 

  16. Palmieri PA, Marshall GN, Schell TL. Confirmatory factor analysis of posttraumatic stress symptoms in Cambodian refugees. J Trauma Stress. 2007;20(2):207–16.

    Article  PubMed  Google Scholar 

  17. Rasmussen A, Smith H, Keller AS. Factor structure of PTSD symptoms among West and Central African refugees. J Trauma Stress. 2007;20(3):271–80.

    Article  PubMed  Google Scholar 

  18. Shevlin M, Elklit A. The latent structure of posttraumatic stress disorder: Different models or different populations? J Abnorm Psychol. 2012;121(3):610–5.

    Article  PubMed  Google Scholar 

  19. Michalopoulos LM, Unick GK, Haroz EE, Bass J, Murray L, Bolton P. Exploring the fit of Western PTSD models across three non-Western low- and middle-income countries. Traumatology. 2015;21(55–63).

  20. Vinson GA, Chang Z. PTSD symptom structure among West African War trauma survivors living in African refugee camps: A factor-analytic investigation. J Trauma Stress. 2012;25(2):226–31.

    Article  PubMed  Google Scholar 

  21. International Classification of Diseases, Tenth Revision (ICD-10). 2009.

  22. Hyland P, Shevlin M, Elklit A, Murphy J, Vallieres F, Garvert DW, Cloitre M. An Assessment of the Construct Validity of the ICD-11 Proposal for Complex Posttraumatic Stress Disorder.  Psychological Trauma. 2017;9(1):1–9.

  23. Liddell BJ, Silove D, Tay K, Tam N, Nickerson A, Brooks R, Rees S, Zwi AB, Steel Z. Achieving convergence between a community-based measure of explosive anger and a clinical interview for intermittent explosive disorder in Timor-Leste. J Affect Disord. 2013;150(3):1242–6.

    Article  PubMed  Google Scholar 

  24. Momartin S, Silove D, Manicavasagar V, Steel Z. Dimensions of trauma associated with posttraumatic stress disorder (PTSD) caseness, severity and functional impairment: a study of Bosnian refugees resettled in Australia. Soc Sci Med. 2003;57(5):775–81.

    Article  CAS  PubMed  Google Scholar 

  25. Miller KE, Omidian P, Rasmussen A, Yaqubi A, Daudzai H. Daily stressors, war experiences, and mental health in Afghanistan. Transcult Psychiatry. 2008;45(4):611–38.

    Article  PubMed  Google Scholar 

  26. Miller KE, Weine SM, Ramic A, Brkic N, Bjedic ZD, Smajkic A, Boskailo E, Worthington G. The relative contribution of war experiences and exile-related stressors to levels of psychological distress among Bosnian refugees. J Trauma Stress. 2002;15(5):377–87.

    Article  PubMed  Google Scholar 

  27. Modvig J, Pagaduan-Lopez J, Rodenburg J, Salud CMD, Cabigon RV, Panelo CIA. Torture and trauma in post-conflict East Timor. Lancet. 2000;356(9243):1763.

    Article  CAS  PubMed  Google Scholar 

  28. Silove D, Liddell B, Rees S, Chey T, Nickerson A, Tam N, Zwi AB, Brooks R, Sila LL, Steel Z. Effects of recurrent violence on post-traumatic stress disorder and severe distress in conflict-affected Timor-Leste: a 6-year longitudinal study. Lancet Glob Health. 2014;2(5):e293–300.

    Article  PubMed  Google Scholar 

  29. Silove D, Liddell B, Rees S, Chey T, Nickerson A, Tam N, Zwi AB, Brooks R, Sila LL, Steel Z. Effects of recurrent violence on post-traumatic stress disorder and severe distress in conflict-affected Timor-Leste: a 6-year longitudinal study. Lancet. 2014;2(5):e293–300.

    PubMed  Google Scholar 

  30. Bolton P, Tang AM. An alternative approach to cross-cultural function assessment. Soc Psychiatry Psychiatr Epidemiol. 2002;37(11):537–43.

    Article  PubMed  Google Scholar 

  31. DeCoster J, Iselin AM, Gallucci M. A conceptual and empirical examination of justifications for dichotomization. Psychol Methods. 2009;14(4):349–66.

    Article  PubMed  Google Scholar 

  32. Muthen L, Muthen B. Mplus user's guide. 7th ed. Los Angeles: Muthen & Muthen; 2014.

    Google Scholar 

  33. Hair J, William B, Babin BJ, Anderson RE. Multivariate data analysis. 7th ed. Englewood Cliffs: Prentice Hall; 2010.

    Google Scholar 

  34. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equ Model. 1999;6(1):1–55.

    Article  Google Scholar 

  35. Kline RB: Principles and practice of structural equation modeling. 1998.

    Google Scholar 

  36. Muthén LK, Muthén BO. Mplus User’s Guide. 7th ed. Los Angeles, CA: Muthén & Muthén; 1998-2012.

    Google Scholar 

  37. Tay AK, Rees S, Liddell B, Tam N, Nickerson A, Steel C, Silove D. The role of grief symptoms and a sense of injustice in the pathways to post-traumatic stress symptoms in post-conflict Timor-Leste. Epidemiology and Psychiatric Sciences. 2016. In press.

  38. StataCorp. Stata Statistical Software: Release 13. College station: StataCorp LP; 2013.

    Google Scholar 

  39. Tolin DF, Foa EB. Sex differences in trauma and posttraumatic stress disorder: A quantitative review of 25 years of research. Psychol Bull. 2006;132(6):959–92.

    Article  PubMed  Google Scholar 

  40. Van Ommeren M, Sharma B, Thapa S, Makaju R, Prasain D, Bhattarai R, De Jong J. Preparing instruments for transcultural research: Use of the translation monitoring form with Nepali-speaking Bhutanese refugees. Transcult Psychiatry. 1999;36(3):285–301.

    Article  Google Scholar 

  41. Vindbjerg E, Carlsson J, Mortensen EL, Elklit A, Makransky G. The latent structure of post-traumatic stress disorder among Arabic-speaking refugees receiving psychiatric treatment in Denmark. BMC Psychiatry. 2016 (In press).

  42. Tay AK, Rees S, Chen J, Kareth M, Silove D. The structure of post-traumatic stress disorder and complex post-traumatic stress disorder amongst West Papuan refugees. BMC Psychiatry. 2015;15:111.

    Article  PubMed  PubMed Central  Google Scholar 

  43. Steel Z, Chey T, Silove D, Marnane C, Bryant RA, Van Ommeren M. Association of torture and other potentially traumatic events with mental health outcomes among populations exposed to mass conflict and displacement: A systematic review and meta-analysis. JAMA. 2009;302(5):537–49.

    Article  CAS  PubMed  Google Scholar 

  44. Mollica RF, McInnes K, Sarajlic N, Lavelle J, Sarajlic I, Massagli MP. Disability associated with psychiatric comorbidity and health status in Bosnian refugees living in Croatia. JAMA. 1999;282(5):433–9.

    Article  CAS  PubMed  Google Scholar 

  45. Mollica RF, Sarajlić N, Chernoff M, Lavelle J, Vuković IS, Massagli MP. Longitudinal study of psychiatric symptoms, disability, mortality, and emigration among Bosnian refugees. J Am Med Assoc. 2001;286(5):546–54.

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We thank our field personal and the Alola foundation for their support for this project.

Availability of data and materials

All data generated or analysed during this study are included in this published article [and its supplementary information files].

Authors’ contributions

SR, DS, ZS conceived and designed the study. SR, DS, ZS gained funding for this study. NT and Z Soares conducted the study. AT, MM, and JB undertook the statistical analysis. AT, MM, SR, ZS, DS, JB drafted the manuscript. All authors read and approved the final manuscript.

Competing interests

The authors declare that they have no competing interests.

Consent for publication

Participants provided consent to publish all data reported in this and other publications arising from the project.

Ethics approval and consent to participate

The study was approved by the Human Research Ethics Committee of the University of New South Wales, the Ministry of Health of Timor-Leste, and the chiefs of each village.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Author information

Authors and Affiliations

  1. Psychiatry Research and Teaching Unit, Academic Mental Health Unit, School of Psychiatry, University of New South Wales, Cnr Forbes and Campbell Streets, Liverpool, NSW, 2170, Australia

    Alvin Kuowei Tay, Mohammed Mohsin, Susan Rees, Natalino Tam, Zelia Soares, Jessica Baker & Derrick Silove

  2. The Black Dog Institute, Sydney, Australia

    Zachary Steel

  3. St John of God, Richmond Hospital, Richmond, NSW, Australia

    Zachary Steel

Authors
  1. Alvin Kuowei Tay
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Mohammed Mohsin
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Susan Rees
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Zachary Steel
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Natalino Tam
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Zelia Soares
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Jessica Baker
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Derrick Silove
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Alvin Kuowei Tay.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Tay, A.K., Mohsin, M., Rees, S. et al. The factor structures and correlates of PTSD in post-conflict Timor-Leste: an analysis of the Harvard Trauma Questionnaire. BMC Psychiatry 17, 191 (2017). https://doi.org/10.1186/s12888-017-1340-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12888-017-1340-0

Keywords

BMC Psychiatry

ISSN: 1471-244X

Contact us