Background
Recently, researchers have shown an increased interest in individual differences in how people respond to trauma [
1,
2]. Some people experience minimal emotional distress and show only minor and transient disruptions in their ability to function, while others suffer from more intense pain that lasts longer [
1]. Researchers have addressed post-traumatic cognitions as one of the factors influencing the severity and persistence of pathological responses to trauma [
3‐
6]. Traumatic events significantly alter survivors’ cognitions and beliefs about themselves, the world, and their future, possibly leading to negative emotional responses and maladaptive actions, which in turn contributes to the development and maintenance of PTSD [
4,
5]. The importance of such trauma-related cognitions is reflected in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders’ [
7] (DSM-5; American Psychiatric Association 2013) revised diagnostic criteria for PTSD. One of the symptom clusters listed among the DSM-5′s diagnostic criteria for PTSD is “negative alterations in cognitions and mood,” which includes criterion D2 (“persistent and exaggerated negative beliefs or expectations about oneself, others, or the world [e.g., ‘I am bad’ and ‘no one can be trusted’])” and D3(“persistent, distorted cognitions about the cause or consequences of the traumatic event(s) that lead the individual to blame himself/herself or others.”) (p. 272). Such changes in diagnostic criteria emphasize the importance of assessment of trauma-related cognitions.
Considering the need for a valid and reliable instrument to assess trauma-related cognitions, Foa et al. [
6] developed the Post-traumatic Cognitions Inventory (PTCI). This inventory consists of 33 items that comprise three factors: “negative cognitions about the self,” “negative cognitions about the world,” and “self-blame.” The inventory was translated and tested on diverse samples in countries such as Germany [
8], the Netherlands [
9], Korea [
10], and Taiwan [
11], where its factor structure was repeatedly verified, and its reliability and validity were confirmed. In these studies, it was reported that certain characteristics like sex, type of trauma experienced, and cultural background could affect PTCI scores and its psychometric properties.
Various studies have shown that cognitive models of PTSD can be extended to children and adolescents; however, they have also indicated the need to consider developmental aspects [
12‐
14]. Therefore, Meiser-Stedman et al. [
15] developed the Child Post-traumatic Cognitions Inventory (CPTCI) to assess the post-traumatic cognitions of children and adolescents. They made age-appropriate modifications to the PTCI items, and added some items based on a cognitive model of PTSD to construct an initial 41-item questionnaire that was used with a community sample comprised 223 children and adolescents. Based on the survey results, the researchers performed item reduction to arrive at the final 25-item questionnaire and validate it in two other sets of samples. Unlike the adult version, the CPTCI comprises only two subscales. First, the “permanent and disturbing change” subscale (CPTCI-PC) comprises 13 items and focuses on the negative effects that a frightening event has on a child and the child’s perception of the future. Second, the “fragile person in a scary world” subscale (CPTCI-SW) comprises 12 items and assesses the child’s sense of vulnerability and perception of the world and other people as threatening. One of the factors of the PTCI, “self-blame,” is not included in the CPTCI.
The CPTCI turns out to be a valid and reliable measure regarding multiple criteria, benefits from being standardized within a large population of children and adolescents [
15]. Based on the theoretical model of PTSD, it has been proposed that the cognitive therapy of PTSD should target post-traumatic cognitions, and studies treating post-traumatic cognitions as a mediator of therapeutic change are being conducted using CPTCI [
16‐
20]. Furthermore, a recently published study updated the CPTCI and evaluated its utility and psychometric properties, providing additional information on the test–retest reliability of CPTCI, as well as suggesting a short form of CPTCI and cutoff points in CPTCI scores for clinical use [
21].
The CPTCI has been translated into several languages, and different versions have been validated and their psychometric properties have been reported in Germany [
22], the Netherlands [
17], Brazil [
23] and Taiwan [
24]. Previous studies generally report adequate levels of reliability and validity. Moreover, in all these samples the two-factor structure emerged as the best solution. These studies, however, showed that the original two-factor structure of CPTCI exhibits unsatisfactory model fit indices that do not meet the widely accepted criteria [
17,
21,
22,
24,
25]. Authors attribute these discrepancies to sample characteristics and cultural differences. In the original study, most of the participants were exposed to traumatic events that did not last for longer than a few minutes and affected few people (e.g., motor vehicle accidents). In contrast, Taiwanese sample predominantly comprises natural disaster survivors. Meanwhile, the majority of children participated in the Brazilian CPTCI study experienced multiple traumas, such as ongoing physical or sexual abuse. The Dutch version and The German version of the CPTCI were also validated in the samples including survivors of interpersonal violence.
To address the issue, the Brazilian version used an exploratory factor analysis to derive a new two-factor model with items that were different from those of the extant two-factor model [
23]. In the Chinese version of the CPTCI developed in Taiwan, researchers revised the original PTCI by deleting five items based on the results of confirmatory factor analysis (CFA). Both methods result in theoretically less sound models because the models were modified based on the results of the analysis. The models neither have enough empirical grounds. Therefore, the models need to replicate, in new sets of samples [
26].
Sexual violence is a type of trauma that leads to severe psychological aftereffects. Sexual assault and sexual violence jointly make up the second largest share of traumas causing PTSD worldwide [
27]. Especially, Child sexual abuse is associated with numerous adverse sequelae during childhood including depression, anxiety, behavioral problems, and post-traumatic stress disorder (PTSD), and is also correlated with an increased risk for mental health problems in adulthood [
28,
29]. Several studies have shown that post-traumatic cognitions in survivors of sexual violence play a significant role in how they adapt afterwards [
30,
31]. Some studies utilizing PTCI reported that a higher proportion of sexual violence survivors had maladaptive post-traumatic cognitions and beliefs as compared to survivors of other types of trauma [
6,
11]. Moreover, since it is now well established that child sexual abuse survivors benefit from TF-CBT targeting maladaptive post-traumatic cognitions, assessment of post-traumatic cognitions in these populations is crucial for intervention [
32,
33]. To date, however, the CPTCI has not been tested on a sample consisting solely of sexual violence survivors to determine its psychometric properties.
Consequently, this study has the following goals. First, we aimed to verify the factor structure of the CPTCI regarding child and adolescent survivors of sexual violence in Korea. Specifically, we sought to determine whether the original two-factor structure derived in the process of developing the scale could be used without adapting it for cultural differences or types of trauma. Second, we aimed to determine the convergent validity and discriminant validity of the CPTCI in comparison with scales that measure the severity of trauma symptoms, anxiety, and depression. Third, we examined the factor structure, reliability, and validity of the short form of the CPTCI (CPTCI-S) [
21].
Discussion
We investigated the psychometric properties of the Korean version of the CPTCI by examining child and adolescent survivors of sexual violence in Korea. This study is the first to validate the CPTCI among Koreans and the first to apply the scale in a sample of survivors exposed to one specific type of trauma (i.e., sexual violence).
Our confirmatory factor analysis revealed that the original two-factor model has the best fit to data among the 25-item models subjected to comparison. Additionally, each of the two factors is loaded on all the items at appropriate levels in the factor matrix, which seems to support the two-factor model. Moreover, model comparison via the χ
2 test showed that the original two-factor model was superior to the one-factor model [
23]. Nevertheless, it was revealed that some model fit values for the original two-factor model fell short of the criteria set based on earlier studies.
The 20-item Chinese version showed better fit indices than the original version. This finding may have significant implications for understanding cultural effects on response to trauma. In this model, the items 3 (I am a coward), 12 (I have to watch out for danger all the time), and 25 (I have to be really careful because something bad could happen) were deleted because their standardized factor loadings were insufficient [
24]. Researchers inferred that in Chinese culture, such cognitions of preparing for dangers are common in parenting and are internalized in children’s self-discipline. It is interesting to note that the current study also found that Item 3, 12 and 25 yielded relatively low factor loadings in CFA. A possible explanation for this similarity might be that Confucianism-based societal norms that East Asian societies have in common.
Nevertheless, we did not adopt the model for following reasons. First, although goodness-of-fit values of the model are better than those of the original model, they still fell short of the criteria set based on earlier studies, and are inferior to those of the CPTCI-S. Second, the version does not include one of the items which consist the CPTCI-S, making it difficult used along with the short form. Third, the item selection was based on the results of CFA in the study which it originates, which can be methodologically problematic. Last, utilizing the version which comprises different items from the original one would not allow the opportunity to compare research on the CPTCI across the regions.
The repetitive failures of different versions of the CPTCI to replicate the original factor structure seems to be related to the characteristic of different sample. In the original study, Researchers have raised the possibility that being at different stages of post-traumatic reactions may have an impact on factor structure and factor loadings. The participants in this study, in many cases, completed the questionnaire when they had visited the support center to receive crisis intervention immediately after their exposure to sexual violence. It is believed that trauma-related cognitions exhibited during the acute phase of traumatic stress may differ in kind and degree from cognitions exhibited after some passage of time when they have naturally recovered or become negatively distorted and consolidated [
15,
43,
44].
Furthermore, the type of trauma experienced by the sample group in this study differs from that experienced by the samples in the original study. The original study used its scale on children who were exposed to a single traumatic event; more specifically, a traffic accident or a violent incident, and derived its factor structure from this basis. Therefore, negative cognitions related to physical injury and internal vulnerability could have become more salient. In contrast, this study was conducted with child and adolescent survivors of sexual violence, representing a mix of single, multiple, or complex trauma survivors. Other studies that have translated and validated the CPTCI, unlike the original study, included many participants who were exposed to continuous trauma like sexual violence and abuse. These studies have likewise reported that they could not confirm a good enough model fit for the original two-factor structure [
17,
22,
23].
The CPTCI was shown to be highly correlated with scales measuring PTSD symptoms, depression, and anxiety. This may be due to the fact that PTSD symptoms are frequently accompanied by depression and anxiety, and it is consistent with findings from previous studies that showed high correlations between post-traumatic cognitions and depression and anxiety symptoms in children and adolescents [
15,
17,
22,
45]. PTSD symptoms and CPTCI scores were significantly correlated even when depression and anxiety scores were controlled for, indicating that the correlation between these two sets of variables is not merely an artifact due to depression or anxiety, but rather due to cognitions and responses specific to traumatic experiences that are shared between the two sets of variables. Traumatic experiences are associated not only with PTSD, but also with various types of psychopathology, and it seems possible to examine post-traumatic cognitions as a transdiagnostic target of therapy and intervention [
17,
19,
46,
47].
Earlier studies found that there were no significant differences in CPTCI results per age [
15,
17]. However, we revealed the opposite. However, the sample characteristics may have affected our results. Previous studies reported that adolescent survivors of sexual violence are closely associated with more violent and severe assault characteristics like penetrative sexual assault, paid sex, brokering, and exhibit more serious and extensive psychological aftereffects than do child survivors [
48,
49]. The adolescents included in the sample of this study also had a higher rate of exposure to rape rather than non-penetrative sexual harassment when compared to children, and their experiences were frequently accompanied by physical violence, multiple assailants, and so on. Other reasons for the CPTCI score differences per age may be related to cognitive and emotional development. In adolescence, more elaborate and complex emotions develop, and there is also the maturing of one’s self-concept and self-consciousness. Accordingly, one’s post-traumatic cognitions concerning threats to oneself, which are also one’s higher cognitions mediating secondary emotions, tend to become negatively distorted and exaggerated [
48,
50‐
52]. Therefore, it is necessary to consider such age characteristics when interpreting the CPTCI results. In addition, future studies need to investigate whether CPTCI reveals any differences per age in the severity and persistence of maladaptation and psychological distress resulting from exposure to sexual violence. As for sex differences, which were not evident, the sample included few male survivors; therefore, it is difficult to interpret and generalize the research findings in this respect.
We also sought to verify the reliability and validity of the CPTCI-S. It was confirmed that the internal consistency, convergent validity, and discriminant validity of the CPTCI-S were similar to those of the CPTCI’s total score. Moreover, our confirmatory factor analysis showed that CPTCI-S had better overall model fit than the original 25-item scale, which was consistent with previous findings [
21]. Among the fit indices for the CPTCI-S, the RMSEA did not have a good fit; however, this may be because the index in question has the property of yielding poor fit when there are only a few items or measurement variables and consequently few degrees of freedom [
53]. When other indices such as the CFI, TLI, and SRMR were considered, they support the two-factor structure. Consequently, the CPTCI-S is expected to be useful in clinical practice and its subscales seem amenable to interpretation.
This study had some limitations. First, instead of using structured interviews with clinicians to perform PTSD diagnoses, the cutoff point for the self-report CRIES was used to distinguish the high PTSD-risk group and the low PTSD-risk group. The Korean version of the CRIES was found to have high sensitivity (.88) and specificity (.85; [
35]); therefore, we felt it could be used to diagnose PTSD with relative accuracy. However, it is necessary to confirm the validity of the CPTCI through more precise criteria in the future. Second, formal backward translation has not been done. Third, this study was conducted only on survivors of sexual violence; therefore, it is difficult to generalize our results to groups exposed to other types of trauma. However, it must be made clear that this limitation is at the same time a strength of this study. Previous studies have shown that CPTCI scores and its factor structure may vary per type of trauma [
15]. For this reason, the original CPTCI paper mentioned the need to apply the scale to various types of samples. Until now, however, no studies had confirmed the psychometric properties of CPTCI as applied solely to survivors of sexual violence. Another limitation is gross underrepresentation of males in the sample. Due to nature of the sexual violence, the sample consists mostly of females. Further study is needed to identify the characteristics of male survivors of the sexual assault.
Despite these limitations, this study is the first to use the CPTCI on child and adolescent survivors of sexual violence, thereby adding new evidence on the scale’s applicability. The present study may extend our understanding of the CPTCI by validating the scale in a different cultural context to previous studies, and in a homogenous sample regarding types of trauma. Further research should be undertaken to investigate the utility of the CPTCI and distinct response patterns considering types of trauma, the phases of response to trauma, and cultural differences.
Authors’ contributions
HYC and KMS designed the study. HYC and HBL wrote the draft of the paper. YKC, YJS oversaw the whole process and provided critical comments. KMS, HBL and MH collected and organized the data from the Sexual Assault Center. USC, SMB and NK helped the statistical analysis and validation process. All authors read and approved the final manuscript.