Introduction

The Latino population in the United States has increased 43 % from 2000 to 2010 [1]. It is estimated that 37 % of Latino youth in the United States are children of immigrants, and 34 % are immigrants themselves. They are the largest and youngest minority group in the United States, making up 20 % of school children in this country. Sixty percent of parents of Latino children in the United States encourage their children to speak Spanish. Though 36 % of Latino youth in the United States report to be English dominant, 41 % report that they are bilingual, and 23 % are Spanish dominant. Most Latino youth in the United States (70 %) report that they often use “Spanglish,” or a combination of Spanish and English [2].

Despite the presence of this growing population, there is still a dearth of literature regarding Latinos and their mental health. Currently, the literature suggests that exposure to potentially traumatic events is relatively common in the Latino children [3]. Moreover, most Latino children experience at least one potentially traumatic event in their childhood [3]. For example, results from the National Survey of Adolescents found that Latino adolescents are significantly more likely than Whites to have been physically assaulted (20.7 vs. 15.5 %) [4]. In addition to higher rates of trauma exposure, higher rates of PTSD or severe PTSD symptoms are reported in Latinos compared to other populations [5]. Given that 1.8 million Latino children age 5–17 in the US are not fluent in English [2], the need for Spanish validated measures has increased dramatically. In 2010, 32.4 % of Hispanic families did not have health insurance [1]. This, and the lack of English language fluency contributes to the fact that Latino children and adolescents in need of mental health services are less likely to receive them [6]. Therefore, it is critical that research be conducted to develop validated and sensitive measures of PTSD symptoms for Latino children and adolescents.

Latino children who have experienced a traumatic event are likely to have both internalizing and externalizing problems, such as depression and behavior problems [7]. Latino children who have experienced a traumatic event are at risk for PTSD and depression [8]. A national survey of 4,023 adolescents between the ages of 12 and 17 sampled in the United States found that Latino adolescents reported higher PTSD and delinquency rates than Whites [9]. Surpisingly, less research have been conducted on PTSD in the Latino pediatric population.

Development of the CPSS Scale

The Child PTSD Symptom Scale (CPSS [10]) was developed as the child version of the Post-Traumatic Diagnostic Scale. Seventeen of the items directly correspond with the criteria for PTSD listed in the DSM-IV, and seven of the items inquire about effects of these symptoms on social relationships, school-work, and other aspects of everyday functioning. Foa et al. [10] conducted the initial validation study for the CPSS, which involved 75 school children ages 8–15 who experienced the 1994 Northridge earthquake with a magnitude of 6.6. The study found high internal consistency among the CPSS items for the re-experiencing, avoidance, and arousal subscales, as well as between the subscales and the total severity score. The functional impairment items were highly correlated after removal of the item related to “general happiness with life.” Test–retest reliability coefficients of the subscale scores, total severity score, and functional impairment score were moderate to high. The CPSS total score correlated strongly with a widely used clinical interview measure of PTSD in children and moderately correlated with measures of clinical depression and anxiety.

Currently, the CPSS is has been translated into many languages other than Spanish; e.g., Korean, Russian, Indonesian, Nepalese [11, 12]. The scale has been translated by utilizing indigenous professional translators, mental health providers, and focus groups, as well as blind-back translation methods [11–13]. However, few studies have been conducted specifically to validate the CPSS in languages other than English. For example, the Nepalese version of the CPSS was found to have good sensitivity and specificity [11]. Additionally, a validation study was conducted on the CPSS in a Hebrew-speaking sample of Israeli children seeking psychological treatment following the experience of a traumatic event [13]. The results evidenced good internal consistency, test–retest reliability, and concurrent validity [13].

Kataoka et al. [14] developed the Spanish version of the CPSS to measure intervention outcomes in Latino immigrant children in the Los Angeles Unified School District (LAUSD) with PTSD symptoms. The LAUSD Translations Unit translated the CPSS from English into Spanish. After the Spanish translation of the measure was reviewed by bilingual and bi-cultural clinicians, it was administered to a sample of Latino immigrant children who were subsequently interviewed to assess their comprehension and interpretation of the items on the CPSS. Minor adjustments were made to the Spanish version of the CPSS based on the feedback of this sample. The CPSS was then given to the study participants, half of which received intervention for PTSD and half, which were placed in an intervention wait-list group. Results indicated that the intervention group had significantly lower scores on the CPSS. The authors noted that their Spanish translation of the CPSS was not yet validated for Latino immigrant youth at the time of their study, these results suggest that the Spanish version of the CPSS is sensitive to changes in PTSD symptoms during an intervention in Spanish-speaking youth. Although the CPSS in English has been used frequently, there has been a dearth of studies examining the psychometric properties of the CPSS [15], and none to date on the CPSS in Spanish.

PTSD Factor Models

The PTSD single-factor, first order model loads all 17 symptoms on one general posttraumatic stress factor. Little research has been done to explore this model in children. The single-factor model was supported in one study in which children and adolescents who experienced the 1999 Athens earthquake were given the Children’s Revised Impact of Event Scale [16]. However, the single-factor model has not been explored using the CPSS.

The DSM-IV three-factor model proposes three symptom clusters: intrusive recollections/re-experiencing, avoidance/numbing symptoms, and hyper-arousal symptoms [17]. The DSM three-factor model has been found to fit well in several studies using the CPSS. Gillihanand colleagues found good internal consistency for intrusive recollection/re-experiencing, avoidance/numbing, and hyperarousal subscales (Cronbach’s α = 0.74, 0.71, and 0.58, respectively) in a sample of 91 adolescent females with PTSD following sexual trauma. Internal consistency for intrusive recollections/re-experiencing, avoidance/numbing, and hyperarousal scales were 0.77, 0.67, and 0.72, respectively, in a sample of traumatized Israeli youth on the Hebrew version of the CPSS [13]. In a study with a sample of 161 Latino children at high risk of exposure to community violence, the three-factor model provided the best fit compared to other models. Additionally, internal consistency was high in both Spanish and English, and the CPSS total scores were positively correlated with exposure to violence [7].

The Numbing four-factor model considers intrusive recollections/re-experiencing, avoidance, numbing, and hyper-arousal as separate symptom clusters [18], and the Dysphoria four-factor model includes intrusive recollections/re-experiencing, avoidance/numbing, hyperarousal, and dysphoria (general distress) as distinct factors. Kassam-Adams et al. [19] found that both the Numbing four-factor model and the Dysphoria four-factor model fit well in a sample of children and adolescents that experienced unintentional injury using the CPSS.

Purpose of Study

The current study aimed to examine the psychometric properties of the Spanish and English versions of the CPSS in children who recently experienced a traumatic event. For the measure in both English and Spanish we examined reliability (internal consistency) and construct validity (via confirmatory factor analyses, CFA).

Methods

Participants

We examined 259 English- or Spanish-speaking children and adolescents with recent exposure to a potentially traumatic event. Children were recruited from health care and community-based social service settings as part of a larger study on assessing acute stress disorder in English- and Spanish- speaking children [20]. For the larger study, 817 children and adolescents were approached and 479 (59 %) were eventually enrolled. Participants were recruited in three US [20] cities: Philadelphia (n = 138), Los Angeles (n = 109), and Miami (n = 12). The study was approved by the Institutional Review Board (IRB) at each participating institution. Children were eligible for the study if they were age 8–17 years, had experienced a potentially acute traumatic event 2 days–1 month earlier, and spoke English and/or Spanish. Because of the aims of the larger study, we excluded index events (i.e., family violence, disclosure of ongoing maltreatment or abuse), which represent repeated relational trauma. Events in which the child sustained a moderate to severe head injury or the child or his/her parent was a perpetrator of violence, and events that led to arrest or a realistic risk of legal liability for the child or his/her parent were also excluded. Table 1 displays the amount of English- and Spanish-speaking participants by type of traumatic event. The majority of English and Spanish participants reported unintentional injury (English: n = 117; Spanish: n = 65) and medical injury (English: n = 22; Spanish: n = 36).

Table 1 Demographic and event characteristics for Spanish and English samples
Full size table

Of the 259 children studied, 106 completed study measures in Spanish (58 boys and 48 girls) and 153 in English (96 boys and 57 girls). The sample consisted of 114 Latino and non-Latino children who identified as either Spanish-speakers 104 (98.1 %) or English-speakers 38 (24.8 %) (see Table 1) for demographic and event characteristics, hereafter referred to as English- and Spanish- speaking samples. There was no significant difference in age between the English-(M = 12.70, SD = 2.7) and Spanish-(M = 13.4, SD = 2.5) samples, t (182) = −0.44, p = .66. The type of acute event reported from children varied from unintentional to intentional injury. Specifically, 61 % (N = 65) of Spanish-speaking children and 76 % (N = 117) of English-speaking children reported experiencing an unintentional injury. Medical events (non-injury) were the second most frequently reported event by both the Spanish-speaking group (N = 36, 34 %) and the English-speaking group (N = 22, 34 % or 14 %?). Two percent (N = 2) of the Spanish-speaking children and 6 % (N = 9) of the English-speaking children reported experiencing an intentional injury. Three percent (N = 3) of Spanish-speaking children and 1 % (N = 2) of English-speaking children reported witnessing or being a victim of interpersonal violence. Finally, 2 % (N = 3) of English-speaking children reported a disaster, a fire, or other as their acute traumatic event. There was a significant difference among the type of acute events between the Spanish (see Table 1).

Procedures

After parental consent, children assented to participation in the study. Parents completed demographic questionnaires in their preferred language. If the child was bi-lingual, they completed measures in both English and Spanish. As part of a larger study, the children completed measures and interviews of stress, anxiety, and depression symptoms. Three months after the initial assessment, research assistants administered the CPSS (in person if possible, by telephone if preferred by the participant). For any child who needed reading assistance when conducted in person, research assistants read checklist items aloud and the child marked his or her responses.

Measures

Child PTSD Symptoms

The CPSS [21] is a 24-item self-report measure of PTSD symptom severity (17 items) and impairment in daily functioning (?) (7 items). Children rate the frequency that they experience each symptom using a 4-point Likert scale, with 0 meaning “not at all or only one time,” and 3 meaning “5 or more times a week, almost always” [21]. The CPSS yields a total severity score, and three subscale scores (re-experiencing, avoidance, and arousal [21]). It can also be scored to determine whether the child meets symptom criteria for PTSD (per DSM-IV). The CPSS (in English) has shown excellent internal consistency, test–retest reliability, and convergent/discriminant validity with other measures of traumatic stress and depression, respectively [21]. The CPSS was translated into Spanish for a study of school-based treatment of traumatic stress symptoms [14]. While a formal psychometric study has not been undertaken, this translation evidenced excellent internal consistency and showed decreases in symptom severity scores with treatment for posttraumatic stress [14].

Data Analyses

Descriptive analyses were conducted for all variables in the study (see Table 1). Preliminary analyses examined differences between English- and Spanish- speaking samples in the PTSD criteria as well as possible age and gender differences. We examined internal consistency (Cronbach’s alpha, α) for the total symptom scale and subscales in each language version. Analyses were conducted with IBM SPSS Statistics Version 20. We conducted confirmatory factor analyses to examine construct validity with MPlus Version 7.

We tested the fit of 4 alternative factor models, which have been supported in prior research: (1) PTS single-factor model, (2) DSM-IV three-factor model, (3) Numbing four-factor model, and (4) Dysphoria four-factor model (see Table 2) for symptom clusters of each model tested.

Table 2 Posttraumatic stress disorder symptom clusters for each model tested
Full size table

Results

Descriptive and Preliminary Analyses

Means and standard deviations for the CPSS total score and subscale scores are presented in Table 3. We summed PTSD symptom severity ratings to create total symptom severity scores (17 symptom items; possible score range from 0 to 51), and derived categorical variables indicating whether a child met symptom criteria within each DSM-IV symptom category. Figure 1 presents the percentage of participants meeting DSM-IV criteria for PTSD (?). Chi square analyses determined that the percentage of participants that met each criteria (B, C, and D) did not differ by child language. To examine possible age differences, both English- and Spanish-speaking samples were split into younger (8–12) and older (13–17) age groups. Across both English- and Spanish-speaking samples, there were no significant differences between age groups in total and subscale scores (see Table 3). We also examined gender differences in both language samples and found no significant gender differences in amount of symptoms reported in the English sample, where as in the Spanish sample, girls reported significantly more PTSD symptoms than boys (see Table 3).

Table 3 CPSS descriptive and internal consistency results for English and Spanish samples
Full size table
Fig. 1
figure 1

Percentage of participants meeting DSM-IV PTSD criteria and overall PTSD

Full size image

Construct Validity

Both language versions demonstrated modest fit to the data for all four models of PTS symptom structure that have been supported in other research. More notably, the English version demonstrated superior fit compared to the Spanish versions. Additionally, the CFI approached acceptable fit for all models in both language versions. The best fitting models were the Dysphoria four-factor model and Numbing four-factor model for both Spanish and English versions (see Table 4). Comparatively, the Dysphoria four-factor model had a slightly better fit in the Spanish sample and the Numbing four-factor model had a slightly better fit in the English sample. The least best-fitting model for both language samples was the PTS single-factor model.

Table 4 Fit indices for confirmatory factor analyses: models of posttraumatic stress for English and Spanish samples
Full size table

Reliability Analyses

Theses total symptom scale demonstrated excellent internal consistency in Spanish and in English (Spanish: α = 0.88 and English: α = 0.89) (see Table 3). Internal consistency was moderate to good within the sub-scale symptom categories (i.e., re-experiencing, avoidance, and hyper-arousal) for both Spanish and English measures (range: α = 0.71–0.84).

Discussion

The purpose of the present study was to assess the psychometric properties of the Spanish and English versions of the CPSS. Examining the psychometric properties of translated measures is critical in ensuring that clinical measures are valid and reliable for use in other populations [22]. Research on PTSD in Latino children and adolescents has increased, which is reflective of the rise of the Latino population in the United States. Therefore, the current findings are important to both researchers and clinicians alike inaccurately/reliably identifying PTSD symptoms/disorders for the purposes of PTSD prevalence, screening, identification, and ultimately treatment in Latino children and adolescents.

The current findings lend support and confirm previous research on English speaking children and adolescents who experienced injury [19] that the CPSS is a reliable measure with construct validity, and extend these findings to Latino children and adolescents. The two models that demonstrated the best fit were the Numbing and Dysphoria models, both of which are consistent with proposed DSM-5 symptom structure in separating active from passive avoidance. This research confirms previous research that avoidance and numbing are distinct symptom clusters [19, 23] and that dysphoria may be as well [19].

Most studies examining the latent structure of PTSD have been conducted with English speaking children or adults. The current results are important in providing an examination of PTSD symptom structure in Spanish-speaking children and adolescents. As mentioned earlier, the Latino population has grown significantly and continues to grow [1]. Additionally, 14.4 % of children ages 5–17 in the United States reside in Spanish-speaking homes, and only 78 % of these youth reported the ability to speak English “very well” [24]. This suggests that a significant portion of Latino and/or Hispanic youth in the United States are more comfortable speaking Spanish than English. Therefore, mental health providers need to be equipped with appropriate tools for assessment and intervention for Spanish-speaking youth.

Results of the current study support the reliability and validity of the Spanish version of the CPSS in Spanish-speaking youth by finding strong support for its internal consistency. However, regarding construct validity, the Spanish version fit only modestly well with the proposed models. Especially given that the CPSS is undergoing revisions due to the DSM-V diagnoses, additional research will need to confirm whether it is valid in Spanish-speaking samples. Since PTSD may be especially prevalent in Latino youth, it is crucial that practitioners utilize measures of acute stress and PTSD that are appropriate for this population, as well as understand the issues and specific characteristics of PTSD in Latinos. For example, unlike the English language sample in this study, girls reported significantly more PTSD symptoms than boys within the Spanish-speaking sample. Previous research has found that girls report significantly more PTSD symptoms compared to boys [25, 26]. This was confirmed in the Spanish sample and not in the English sample.

There are a few limitations in the present study. The current study enrolled children and adolescents exposed to a range of potentially traumatic events, but the sample was largely made up of children experiencing injury or other acute medical events. Additionally, the current study did not assess convergent or divergent validity. Future studies should examine the models tested in children who experienced other types of potential trauma and examine both convergent and divergent validity of the CPSS in Spanish. Moreover, the Spanish-speaking sample (n = 106) was much smaller compared to the English-speaking sample (n = 153), which may have had an influence on the superior fit of the models of the English-speaking sample. With the introduction of additional symptom criteria for PTSD in DSM-5, future studies will need to address the psychometric properties of an expanded CPSS in both English and Spanish.

Summary

Overall, the current study suggests that the CPSS can be a useful tool for clinical practice and research among English-speaking children, but provides less support for its validity in Spanish-speaking children. Given that Latino Spanish-speaking youth comprise the largest growing child/adolescent population in the US [1], our findings lend support for more robust and valid instruments for detecting key posttraumatic mental health concerns in potentially traumatized youth seeking hospital-based services.